I think about AI adoption using a simple model: the Signal to Noise Curve.

Early in a technology cycle, everyone experiments. Most of those experiments don’t last, which creates a lot of noise. Over time, the experiments that don’t create real value fade away. What’s left is the signal.

Which Type of Persona Do You Want to Be?

To understand where you want to sit on the curve, the first question is simple but important: Do you want to be an innovator or an early adopter? There are important differences between Innovators and Early Adopters on the Product Adoption Curve.

When AI came out, some of my more innovative friends were trying to learn all about the transformer and next token prediction. A lot of that felt like learning TCP/IP during the early internet. I’ve built a lot of successful internet products. I still can’t explain TCP/IP succinctly. It’s too deep in the stack to matter for the leverage I’m trying to create. It made me realize I’m a different persona, and that’s okay. If you’re an early adopter, don’t feel guilty about not being an innovator. Just change your consumption patterns to match.

Should You Be Either of These Personas?

The main way to answer this question is: what is required for your job? AI is still just a tool. Focus on what the tool is useful for in your job or company’s strategy. If you’re responsible for selling AI tooling into the enterprise, you probably need to be an innovator. You can’t have one of your prospects ask about something you haven’t heard about. It would be damaging to your brand.

But let’s think about my job. As a consumer startup founder with the product role, not the technology role, understanding how to build with AI meant I should be an early adopter there. So I forced myself to get Cursor set up and started making changes to our product.

At the same time, there were a lot of innovations happening in AI image and video generation. I didn’t think that was required for my job. I didn’t spend time mastering Midjourney prompts. I just waited until image generation got 100× easier like everyone else.

Where To Sit on the Curve?

If you can intelligently answer whether you should be an innovator vs. early adopter vs. neither, you can figure out how to ride the AI Signal to Noise Curve.

If it’s your responsibility to know a lot about AI in general, like in the previous example, you’re going to deal with a lot of noise, will need to try a lot of things and will spend time on a lot of things that don’t end up being valuable.

If you should be an early adopter, you want to balance signal and noise, and sit right in the sweet spot. If it isn’t relevant for your job, you can wait until it becomes mainstream, because that will correspond usually with a massive decrease in effort to get value from the tool.

The key rule: you shouldn’t sit in the same place on the curve for everything.

The same person can be an innovator in one area, an early adopter in another, and the early majority everywhere else. Let’s take my role again as a startup founder. For my role in setting the vision for an AI product, I need to be an innovator to build the right thing. For how I help build the product, I should be an early adopter. For everything else, I am the early majority.

Let’s see this curve mapped with some major innovations of the last couple of years.

Claude Code crossed into the early-adopter zone for product leaders in the last six months. AI meeting notes and image generation passed that threshold a long time ago. New tools like OpenClaw and Claude Cowork aren’t there yet, but innovators are tinkering with them.

–

AI doesn’t change every week. Capabilities change a few times a year. Everything else is noise. Decide where you want to sit on the curve for each topic, and ignore the rest.

Ask my SuperMe follow up questions, and I’ll probably respond.

Currently listening to my Springs Flings playlist.

Sooners and Missourians

Sooners and Missourians can sit in the same demo call and look identical on paper. Here’s how they actually behave.

Sooners

• Want to be early

• See themselves as “in the know”

• Curious by default

• Tolerant of rough edges in your product

• Motivated by status and play

Missourians

• Skeptical

• Risk-averse

• Require evidence

• Financially grounded

• Need proof before committing

Sooners adopt because they want to be early.

Missourians adopt because they want to be right.

You’re Probably Talking to the Wrong Customer

Early founders need Sooners. They try things. They explore edges. They lean into potential. Their engagement is your early signal of progress.

Missourians look valuable to founders. They usually aren’t. They give rational feedback, but don’t spend real time with the product. They point out gaps. You close those gaps. Nothing changes. What needs to be resolved first is scale and proof that others are finding value. Early on, you should largely discard Missourian feedback. That means you need to identify them quickly.

Sooners reward vision. Missourians demand infrastructure.

How To Tell Who You’re Talking To

Both personas, fortunately, reveal themselves quickly.

Chasm Crossing vs. Crossing State Lines

In enterprises, this is frequently talked about as the product adoption curve. The difference isn’t timing. It’s motivation. In Reforge, we teach that all user behavior is motivated by either social, financial, or personal utility. In simple terms, a product must help me signal identity, make money, or create personal utility.

Sooners want to accrue social status, be able to explore, and have fun with some potential upside later on. Missourians want money, risk reduction, clear ROI, and proof from their peers beforehand. Crossing the chasm is not just distribution. It’s about rearchitecting the motivations behind using your product.

Migrating from Oklahoma to Missouri

When you move from Sooners to Missourians, the product must change. Social capital must become financial capital. Attention must turn into income. Play must become business. “Show me” often means “show me the money.” This is why most direct network effect businesses evolve into cross-side network effect businesses at scale. Creator vs. consumer. Buyer vs. seller. User vs. advertiser.

“Where are the ads?”
Missourians get confused when they don’t see transaction economics. They want money made visible. It reassures them they aren’t being taken advantage of. For example, when I was early at Grubhub, someone asked us why we don’t show ads. They didn’t realize we took a percentage of every order, which, when converted to an ad-based metric like CPM, was far more lucrative than any ad unit we could run. Ads would have reduced orders and cost us money.

Similarly, people have asked me why I don’t monetize my blog. My response has been that I do. It drives leads to and improves conversion rate for my advising business and for full-time job opportunities. I make far more from those activities than I would from the occasional sponsorship.

–
Every product starts in Oklahoma. Every successful one eventually crosses into Missouri. The mistake is trying to build for both at the same time. Know which state you’re in, and listen and build accordingly.

Ask Casey AI about this topic on SuperMe.

Currently listening to Container by Lackluster.

Operating Styles Are Context Specific

Ludo and I first got exposed to each of our styles when scaling Pinterest. Initially on different teams, I got to see Ludo’s creative problem solving and intuition for when to hack and when to scale engineering systems, and I think Ludo got to see my ability to strategically prioritize and frame problems to the rest of the organization that made what we were doing more legible to them.

We started working together again when Ludo brought me in to advise Whatnot. That exposed both of us to different aspects of our management styles, both because it was a different set of circumstances compared to Pinterest and because we had both gained a lot of new reps and skills from recent experiences at Lyft and Eventbrite. Ludo was operating at more scale, and given my part-time status as an advisor, I was more direct and less patient, trying to be as efficient with the smaller amount of time I had to help the company.

The Founder Surface Area Matrix

Working with Ludo zero to one as a co-founder has also been different. While both Pinterest and Whatnot were scaling experiences for companies that had product/market fit, we started SuperMe from scratch. Over time, we noticed two key axes on which we operate materially differently. Founders make two early bets going zero to one:

• How much surface area to build?

• How much aesthetic/brand polish to invest in?

Here’s my take on what quadrants famous companies started in:

Max Features, Min Design: Facebook, Reddit, Jira, Craigslist

Max Features, Max Design: Figma, Coda, Canva, Shopify

Min Features, Min Design: ChatGPT, Google, Twitter, WhatsApp

Min Features, Max Design: Pinterest, Snapchat, Robinhood, Granola

Many great companies move quadrants over time, and neither of these approaches are default right or wrong, and founders do not need to be aligned to the same bucket to be successful together, but each has pros and cons and optimal working model.

Maximal features → faster learning, harder to prune

Minimum features → easier to prune, but slower learning

Maximal design → stronger positioning, but slower learning

Minimal design → faster learning, weaker positioning

The surface area decision determines your pruning burden. The more you build, the more you must cut. The more polished your product, the harder those cuts feel.

Distractions vs. Compounding Value

For a consumer network, as mentioned in my last essay, there are three variables: value prop, target audience, and network density, and this impacts how you act in the different modes above.

In a maximal feature model, any experiment that works needs to enable a new step in the process that clears the table of other experiments to triple down on what is working. The Instagram founders famously did this with photo filters of their original social app Brbn, as did Pinterest with saving from their original shopping app Tote.

It’s hard to give up on other experiments when you’re a feature maximalist. Some of those other experiments could also be hits with more investment or more time. Sunsetting them is brutal for builders and initial users.

To triple down on hits, you need a clear process for handling misses. The key question is always is this feature a distraction or something that compounds with higher density? In networks, the line between distraction and compounder is blurry because value increases with density.

You can’t triple down on hits if you can’t diagnose what’s wrong with your misses. If you keep pouring energy into misses, you starve your hits.

One useful process is a formal experiment review after a feature has been live for a while that answers:

• Is it a hit? If so, what are we removing from our feature set and future roadmap to focus on it?

• If it’s not a hit, is it a distraction that should be killed or is there evidence to suggest its value will compound?

• If it’s a compounder, do we iterate now, let it sit, or intentionally defer it until density is higher?

Feature maximalists are more prone to distraction risk. Feature minimalists are more prone to missing latent compounders.

Understanding the Full Vision

Consumer products don’t win by chasing engagement alone. Because LTV is lower than in SaaS, consumer interest must convert into defensibility.

One exercise that we found helpful in answering the questions above is “is this feature required for the full vision?”. In order to answer this question, it helps to go back to some of the key questions Gilad and I proposed in our platform essay:

• What does the company need to own?

• What does the company want to compete to win?

• What does the company want to attract?

• What does the company want to reject?

If a feature is firmly in the first bullet for your long term success, it’s a compounder. Any other answer makes the feature a hit or a distraction. So, it’s either working and helping you grow and get closer to vision, or it needs to be killed.

–
Zero to one isn’t about building fast. It’s about pruning fast. You will try things that don’t work. The only question is whether you can recognize them, and whether you know what you’re protecting when you cut them. If you don’t know what you need to own, you won’t know what to remove.

Feel free to ask Casey AI more about this topic.

Currently listening to BOC Maxima by Boards of Canada.

• AI should be transformative enough to create a new wave of consumer network products

• Those product ideas should be attractive to entrepreneurs

When you look at what’s happening in the market and compare it to previous platform shifts, the answer becomes clearer. New network businesses require patience, new land, and a very specific alignment of variables. None of which are enabled by the current environment.

Magic Tricks vs. Moats

During a platform shift, it’s common for founders to seize new capabilities to create magic tricks: experiences that feel novel because they lean on new tech, not new structure.

These products often:

• Go viral (or seem to)

• Use undifferentiated infrastructure (e.g., commodity LLMs)

• Have low retention

• Are easy for competitors to replicate

Viral growth alone isn’t a moat. Once the trick is explained, or once a big incumbent and startups clone the behavior, the product becomes a commodity. And commodities don’t compound.

This dynamic played out during the mobile platform shift too. App Store favorites featured in early iPhone commercials and viral hits drove spikes in downloads, but the companies that endured actually came a few years later. Instagram, TikTok, Uber, DoorDash, and Snapchat were not magic tricks. They were architected for defensibility from the start.

AI products today are repeating the pattern: impressive initial growth, shallow retention curves, and little unique ownership of a user’s attention or behavior. Unlike mobile startups of the past, many of these AI tricks monetize early, and must do so to pay for compute. This allows them to brag about ARR instead of downloads, but it only masks the underlying business’s weakness for slightly longer than viral apps during the App Store boom.

When entrepreneurs can do the cheap thing and reap all the rewards short term, why worry about the long term? It’s hard to understand, internalize or admit when you’re going so fast that you’re about to drive off a cliff. Growth momentum makes cliffs hard to see.

Even products intended as feeder experiences into bigger, more defensible plays often end up trapped in act one. Teams struggle to sustain growth in act one instead of investing in long-term defensibility of act two because act one is scaling so quickly.

Why Venture Capital Still Funds This Stuff

If the ARR isn’t eye popping enough for you with the first generation of AI companies, the valuations should be. Are VC’s just getting distracted by high flying growth? Shouldn’t they know better? You can’t blame investors for backing fast growth. Venture capital is a portfolio business, not a concentration business. For a VC, If they back ten fast-growth companies and one or two turn that growth into a sustainable business, they are one of their generation’s best investors.

Founders and employees only get one shot at a time. That misalignment of portfolio construction vs. concentration is part of why so many startups chase growth signals that look good in a pitch deck but don’t translate into durable user value as measured on a long term time horizon.

The last decade of venture has been dominated by SaaS. The muscle memory of most investors today is subscription metrics, not network density. That shapes what gets funded and how it’s evaluated.

Yes, HALOs and acquihires soften some downside risk. Yes, early secondary markets let founders and early employees capture liquidity. But those outcomes require specific talent and timing that most early-stage teams won’t have.

No New Land

Alex Zhu, the founder of Musically (which became Tiktok), gave the best explanation for the raw ingredients required for new networks to arise. Alex mentioned that building a new consumer network product is like building a new country. And it’s a lot easier to build a new country when there is new land that opens up. Mobile opened up a lot of new land in terms of internet access on the go. Instagram, Tiktok, and Snapchat consumed a bunch of free time people had waiting for coffee, at the DMV, etc. AI hasn’t yet opened up new land, which means as consumer founders we need to steal time and attention from some other product, one that is frequently extremely well optimized at keeping its users’ attention.

New land often creates new distribution channels. Consumer networks, which monetize slowly, depend on those cheap acquisition loops to reach density. Distribution tends to lag capability.

To be clear, not every consumer wave requires a new platform. In venture they call this the “why now.” Sometimes the unlock is regulatory, not technological. The deregulation of online gambling has created a surge of consumer network-oriented startups such as Kalshi, Polymarket, Underdog, Whatnot, Whop and others without any new land. The “why now” was a combination of legal permission, innovations in variable reward mechanics, and video infrastructure.

AI, so far, has increased capability. But it hasn’t yet created a comparable structural unlock for new consumer networks.

The Network Three-Body Problem

The final dimension preventing consumer network growth is structural: consumer network products don’t have a simple product/market fit geometry. They have a three-body problem.

Network product/market fit isn’t one variable. It’s an alignment problem.

1. The right target audience

2. The right network size

3. The right value proposition

If any one of these is off, you can’t tell which one is broken just by looking at the metrics. And if you misdiagnose early churn as “not enough users,” you’ll misallocate your budget to acquisition instead of fixing the real problem.

Marketplace and SaaS founders get clearer feedback loops:

• In SaaS, you tune value prop.

• In marketplaces, you tune liquidity.

• In networks, you’re tuning three variables at once.

That’s why some products burst into existence, spike in growth, and then fizzle, and why building a defensible network takes multiple cycles of learning, not just initial virality.

–
Until AI creates new land, or someone finds a way to manufacture density without it, most attempts will default to magic tricks to go viral and not create lasting value. The opportunity isn’t in the trick. It’s in compounding trust and density over time. The first wave of AI products optimized for speed. The second wave will optimize for structure.

Feel free to ask Casey AI more about this topic.

Currently listening to my Best of 2025 playlist on Spotify.

You can also listen on Spotify here:

We covered:

↳ Why meaningful expertise has disappeared from public platforms

↳ How knowledge moved into private networks, and why that’s a loss

↳ The difference between answers and perspective

↳ How SuperMe uses AI to scale access to judgment, not replace humans

↳ Why peer learning and trust matter more than ever in career growth

Most real-world problems don’t have a single right answer. What people actually want is to learn how great operators reason and then decide for themselves.

If you care about AI, expertise, and the future of professional networks, I think you’ll enjoy listening.

You can also listen on Spotify here:

The best practices of building companies are unevenly distributed. Every great company has pockets of excellence and its own blind spots. The knowledge behind the best teams lives in documents, meetings, and the minds of great operators. Until now.

We’re very excited to announce SuperMe, the AI-native professional network.

The AI-Native Profile for the Future of Work

With SuperMe, your work becomes your profile. We create an AI-native representation of you from your writing, talks, notes, and more. Your AI profile stays synced and grounded in your real words, giving others a way to learn from you while keeping you fully in control.

You shouldn’t need to become a creator to get credit for your expertise.

We know self-promotion can feel awkward and time-consuming. Yet when you stay silent, opportunities fade. Recruiting gets harder, your relevance declines, and the best work goes unseen. With SuperMe, your work speaks for you.

Anyone can talk to your profile with voice or text to learn about you or ask for your advice. Every response is grounded in your real words — no hallucinations, no guessing. If we don’t know what you think about something, we’ll say so. You can always review and refine your profile’s answers, training it to represent you even better over time. You can even send real follow ups to those who asked you questions.

Your profile isn’t a replacement for you. It’s an extension that handles common questions and makes your expertise instantly discoverable.

SuperMe has already supported over 30,000 conversations.

Product thinkers like Nikhyl Singhal and Brian Balfour use SuperMe to share career and product advice in the age of AI.

Growth leaders like Elena Verna and Adam Fishman use it to teach companies how to shape their growth.

Marketing leaders like Rachel Hepworth and Ethan Smith use it to showcase how marketing strategies and roles are evolving.

Introducing a New Approach to Search

Great insights only matter when they can be discovered. And the truth is, the most insightful operators are often the worst marketers of their own expertise.

Introducing Perspective Search: find the right expert for any business question.

Ask a question, and SuperMe surfaces the most relevant professionals. Their AI profiles answer instantly, each grounded in that person’s real work. We summarize the range of perspectives, and let you explore each one. It’s how advice really works in business: you ask multiple people, compare views, and decide. We’ve built that into the network itself.

SuperMe is built as a network from the ground up, so when you follow people, their thoughts are prioritized in search, and when they share more insights, we’ll notify you about them.

Why is now the time for a new professional network?

Professional networking will be forever changed by AI, and there are a few key trends that force us to rethink how things should work given this new technology.

Knowledge moved private.

Expertise left public spaces as algorithms rewarded virality over nuance. Private chats hide the best insights — unsearchable, unattributed, and invisible.

Search is changing.

AI assistants answer without revealing the humans behind the knowledge. Perspectives are lost as blue links turn into one-size-fits-all answers.

Network graphs broke.

“Connections” became meaningless. Vetting and filtering are nearly impossible.

Careers are faster and fuzzier.

Reputation is the new résumé, but few know how to market themselves to unlock the deal flow their expertise deserves.

Why are we the right people to build this?

We’ve been building network products for the last 20 years. While we met at Pinterest, and helped scale the network from 20 to 200 million users, we’ve also been behind the scaling of other networks such as Whatnot, Lyft, and Grubhub.

But there’s a more personal reason for this. We didn’t start our careers as Silicon Valley insiders. In fact, neither of us started our careers in Silicon Valley. We were lucky to get exposed to how powerful networks can be when they work at first outside of Silicon Valley (Ludo at Hulu, Casey at Grubhub) to make our way here. Ludo came from Bulgaria, Casey from New Orleans. Advice on how to succeed in our careers was not really accessible in either place.

We built SuperMe to give everyone the kind of network that once only existed in Silicon Valley. One that turns your experience into opportunity and allows you to learn from the best operators in the world.

Try it today or download it on the App Store.

The Traditional Purchase Funnel

A lot of unfulfilled excitement around LLM products is how LLM products like ChatGPT could compress the discovery journey. The historical marketing funnel for high consideration decisions starts with ideas and inspiration, advice and insight, pricing and comparing, and purchase and experience.

The early consumer internet excelled at pricing and comparison, and many won here by making purchasing seamless too. This is where marketplaces thrived in many industries like travel (Expedia/Booking), physical goods (Ebay/Amazon), real estate (Zillow/Redfin) and food (Grubhub/Doordash). Google as a search engine perceived its job to be to get you to the best site for this type of comparison and purchase as quickly and accurately as possible.

Much of the first two steps happened offline in the early internet, and started to be brought online in the age of social and content networks like Facebook, Instagram, Pinterest, and Tiktok. But none of these platforms have been able to extend very well into the traditional marketplace steps of pricing, comparison, and purchase despite dozens of attempts. Marketers called these destinations demand creation and demand capture, and managed them somewhat distinctly. Unifying them with metrics has always proved difficult.

LLMs can now collapse this funnel, and about a dozen years ago Google was the first to change its tune. Instead of being content with being the traffic router to the best marketplace, it started to expand its mission to solve your problem directly. Mobile made comparison slow, so Google began solving it in‑house. Google approached solving this problem by looking at the top categories of searches and either incubating (Google Local, Google Shopping) or acquiring (Google Flights) competition to the marketplaces they used to happily send this traffic to. That vision has continued to expand to this day with some success.

Discovery Agents

In an agentic world, an LLM that has a deep understanding of your preferences will be theoretically be the first to understand new intent to give you ideas and inspiration for, will have the best understanding of what choices are reasonable for you, taking away a lot of pricing and comparison work search engines and social networks have had to delegate to marketplaces in the past. This collapse of the marketing funnel changes the customer acquisition economics of every marketplace.

Bill Gurley recently argued that when ChatGPT rolls out ads, it will want a share of the billions Booking.com and others now spend on Google. If agents do the discovery, marketplaces will still bid—but far less—because they lose downstream lifetime value potential.

Why payback math breaks in an agentic world

Marketers today assume that visitors build brand affinity, convert (where the marketplace takes a cut), and can be re-marketed to. This increases the chances of direct purchases in the future. Therefore, a marketer might measure that transaction cost on a payback period basis based on the lifetime value of similar customers from that channel.

In a world where agents potentially do all that discovery and booking work for the user, the marketplace receives potentially none of those benefits. Sure, the marketplace will take the additional transaction revenue, but it will no longer use payback period math, but unit economic math for a single transaction as it can no longer feel confident it’s acquiring the ability to do future outreach, create brand preference, and increase the chances of direct purchases in the future.

The higher frequency the activity, the more likely one can create an experience that retains better than a generic discovery agent experience. Retention is primarily influenced by:

• Natural frequency: how often do I need the value prop of this service?

• Loyalty: when I have this value prop, how often will I use the same service?

Memory, or personalization, should also help existing apps that have had frequent usage in the past maintain direct relationships moving forward as vertical discovery agents. Personalization can be thought of as a first order effect of a data network effect to increase engagement. It frequently has a second order effect of making experiences for new users better as well because most new users benefit from the data of existing users on what is good and not good. This is part of how Pinterest doubled its activation rate while I was there.

How founders should react to discovery agents

• Switch to payback on first purchase for transactions that are sourced directly from discovery agents

• Fortify brand hooks with customers via the website experience post the handoff from the LLM

• Focus on high frequency markets that will benefit from optimized experiences and personalization

Transactional Agents

When agents stop at discovery, marketplaces still have a chance to build loyalty. When they also execute the purchase, the marketplace disappears entirely. This is a losing scenario.

Many marketplaces would object to transactional agents initially and try to block them. But can they? LLMs don’t have an equivalent of robots.txt that their agents will adhere to, and new protocols like MCP clear the way for these agentic workflows with personal information and credit card data regardless of a marketplace’s decision to integrate with an LLM or not.

If you can block agents, you might miss out on increased impressions from discovery agents or more transaction volume if people use a dedicated transactional agent from OpenAI or Google for most tasks. In marketplace verticals historically, the top supplier with a very high market share can opt out of participating in the marketplace, confident in its marketing ability to bring people directly to their product. Southwest Airlines not integrating with any flight marketplaces is a famous example. But everyone else needs the volume despite the lack of direct relationship with the customer. Even Domino’s eventually got on Doordash.

If the transactions still occur offline, that presents a marketing opportunity for the marketplace. If a marketplace is working with physical inventory in the real world, as it scales, it has the opportunity to use that real world inventory to promote the marketplace brand. Grubhub tended to leverage signage, dedicated business cards, stickers etc. inside the restaurants it worked with to drive more awareness. This became the source of over 15% of new users. This tactic may need to be dialed up more aggressively for marketplaces in an age of AI.

High frequency products still win as they can always optimize the experience to feel superior to more generic front door versions as vertical transactional agents. I still use Granola (notes) and Cursor (coding) over other LLM solutions. This is just like how I prefer Whatsapp over texting, Spotify over Apple Music, Chrome over Safari, Amazon over Google Shopping et al. despite distribution disadvantages for all of those companies.

And if frequency is really high, you might have multiple solutions you use directly as apps on your phone or laptop. How many of you have Uber and Lyft on your phone (even Waymo)? Some combo of Doordash, Uber Eats, and Grubhub? Whatsapp, Messenger, Signal, and Telegram? There is still considerable value in being a second app in use next to ChatGPT or Google.

How founders should react to transactional agents

• Opt out if you have incredibly large market share or until you get an incredible deal

• Fortify brand hooks with customers via the real world where the transactions frequently take place, including potentially tighter couplings with the supply

• Be the best experts in your vertical on pricing and authenticity of content or some other key lever of experience. Despite the emergence of dupe culture, there will still be some resistance to AI slop or cheap knockoffs

Supply Integration and Supply Agents

Once agents intermediate demand, suppliers will plug in directly just as every local business eventually claimed a Google Maps pin. So a marketplace’s supply, once integrated into a LLM, would not need to use the marketplace to receive more business if they can get it directly from the LLM.

The solution almost has to be a much tighter coupling of the marketplace with supply. Gilad Horev and I wrote in the past about the types of marketplace models, and many were surprised that we listed vertically integrated companies on the far right. Those aren’t marketplaces! Sameer Singh even calls these MINOs (marketplaces in name only). Why did we list them? Because, as Sameer explains, this sometimes is the only way to create a winning customer experience, despite the fact that you lose margin and network effects (both of which make growth harder).

Well, now, in addition to demand side preference on experience creating the need for vertical integration, distribution needs may force more vertical integration in some marketplace markets. Otherwise, a marketplace’s ability to acquire customers profitably and retain them may disappear, and physical real estate, standards, and the branding opportunity that affords may be the only way to survive, even at dramatically depressed margins.

Perhaps marketplaces can receive a lot of the value of vertical integration without the dramatic change in business model and maintain some of the value of their network effects in the process by integrating more directly with their supply as described in the transactional agent section.

For less physical domains, the solution may be using AI to beat AI by making agentic versions of your supply. In this model supply may have a lot of process knowledge that can be programmed into an agentic version of the supply to do tasks on their behalf, earning them more passive income than most marketplace models. Imagine Upwork or Fiverr having their agents doing 95% of the work their supply used to, at similar economics to supply.

By managing that agentic technology, that vertical agent will likely work a lot better than generic ChatGPT or any individual supplier trying to create and maintain their own agent. The agent itself can drive more discovery of other agents in the marketplace who are better at certain solutions than who a user might be engaging with at the moment. These agents improve with every “transaction” both for repeating and new customers on the demand side, creating data network effects, and their ability to recommend other agents preserves cross-side network effects as well. Each operation improves operations, gathers outcome data, and improves future matching. This is what we are working on at SuperMe.

Perhaps this leads to changes in monetization models that feel more subscription or usage-based as the marketplace now powers so many workflows for the supply. While marketplaces are no stranger to subscription models, they have historically been limited to add-ons or lead gen models where marketplaces can’t verify every transaction. Licensing data or access is not out of the question either. Cloudflare is working on this, but it feels hard for this to be the cornerstone of a strategy.

How founders should react to supply integration

• Deepen the relationship with supply via workflows, full integration, or owning their individual agents and making them self-improving

• Use supply AI to beat discovery and transactional AI

• Be willing to change your model if you have to to preserve your value prop

The Marketplace Response Playbook to Agents (2025-2030)

1. Go agentic on both sides of the marketplace. Develop your supply and demand agents and capture your own structured data. Use AI to beat AI.

2. Reorient CAC to first order economics. Stop overpaying for traffic without identity.

3. Deepen Supply Integration. Offer workflow tools and consider forms of vertical integration.

4. Exploit Offline / Packaging Surfaces. Build recall where agents can’t completely cut you out of the experience.

5. Experiment with Monetization Mix. Experiment with subscriptions and data or licensing before take rate erodes.

The marketplace landscape is changing, and I urge founders to start planning major changes in how they acquire and retain users now, and be ready for more drastic changes about how they operate their businesses generally in the future. While on average these changes may be net negative for marketplaces, I predict we’ll see some incumbents and startups alike emerge to build much stronger business than we have seen in the last decade. But those who wait are more likely to perish.

Feel free to ask Casey AI more about this topic.

Currently listening to Hinterland by Aim.

Thanks to Mike Duboe for feedback on this essay.

This post is the fourth post in an unintentional series on the levers founders reach for when trying to scale their company’s growth. Last time we talked about experimenting with new channels and what prerequisites need to be there to make that experimentation able to drive significant improvements in growth. A more nebulous topic founders look toward as a solution to slowing growth is brand marketing. Let’s dive in on why that is the case, the different components of brand marketing, and what it can and cannot do to fix slowing growth.

A common scenario is that startups find product/market fit by leveraging demand capture channels. People search for a solution, the company provides that solution, and has a high conversion rate from the people saying they have that need. But no topic has infinite search volume, and of course it’s hard to rank higher than #1 on Google Search or Google Ads. Demand capture is basically a variation of the common YC saying: “Make something people want.”

Demand creation is the opposite. You attempt to go into the world, interrupt people, and convince them they need the thing you are selling. Demand creation is “Make people want things.” Now, of course there are many amazing companies that do that like Nike and Apple. But notice how large those companies are and how few software companies specifically win with demand creation vs. demand capture. This is certainly different compared to CPG, which can basically leverage either brand marketing or promotions. But for software, the safer and more traditional option when you run out of demand capture channels is to invent more ways to capture more of that demand that exists and grow organically with hopefully increased demand over time. Rank #1 organically on Google? Well, how about you buy a competitor and rank #2 as well? It’s what worked for Expedia and Booking.com, the two largest travel companies.

When founders talk about brand marketing much earlier in their lifecycle, it’s usually because they are unhappy about either their awareness or positioning. Founders feel like they’ve invented the future, and they do have customers who see it and live in that future, but most of the world is still unaware. If they could just get more awareness of how amazing this service is, or if they could just better describe why what they do is so amazing, their usage would 10x! And founders can point to previous examples of startups providing a new way to do things being quickly adopted. Look at Uber! Airbnb! Groupon! (rip) ChatGPT!

Let’s break the problem down between awareness and positioning first. I’ve said it before, and I’ll say it again. In startups, awareness is never the problem, and never the solution. Awareness building rarely translates into faster growth, can be quite expensive, and is near impossible to measure the effectiveness of. Startups should treat all marketing like performance marketing and measure effectiveness based on payback period even if the creative or medium is heavily brand-oriented e.g. TV or OOH. Why? Because awareness oriented marketing, when it works, which is rare, tends to have payback periods longer than the runway of most startups. So if your brand is compounding, it does so too slowly to save the company before it runs out of money.

Now, let’s talk about positioning. Founders think if only they could figure out a better way to describe their product like world class marketers do at Apple or Nike they would grow much faster. This is also mostly never as big of an opportunity as you think. For most examples where founders think positioning was the driver of growth, the underlying product strength explains the growth trajectory of the company. Uber and Airbnb were better and cheaper than existing alternatives. It’s pretty easy to do positioning when you’re better and cheaper. When founders raise this positioning problem, usually the issue is they are better, but not 10x better than existing alternatives, and they’re certainly not cheaper. Your product/market fit just isn’t as strong. As the kids would say, skill issue. That does not mean more consistent positioning or more appealing brand design can’t increase conversion on ads or the website, but it’s an optimization, not a game changer.

As you get to massive scale, normally well after becoming a public company, one can lean into brand marketing, and might need to especially if competition catches up on price point or value prop. Airbnb is a great recent example of reconfiguring the entire company to run less like a marketplace and more like a hardware company by bringing in a bunch of Apple execs. This is because they are no longer cheaper than hotels, and no longer tend to have unique inventory compared to other marketplaces like Booking.com that now have the alternative inventory you used to only be able to find on Airbnb. So their strategy is to leverage all of the good will from having lower prices and cooler inventory into an enduring brand by running the company like the best brand in the world. It’s a bold strategy, Cotton. And it’s not going to be right for your startup. Booking.com of course does brand marketing too (and I’m sure Idris Elba isn’t cheap), but the core of their strategy is still performance marketing. Don’t try to emulate these companies too early in your company’s lifecycle, or you’ll tend to just waste a lot of money.

–
Brand marketing is a force multiplier, not a life preserver. If your product isn’t yet meaningfully better or cheaper, and if you haven’t exhausted high‑intent channels, pouring money into awareness or clever taglines will merely extend the runway on your burn, not your growth curve. Optimize demand capture first, measure every dollar like performance spend, then graduate to brand when scale and competitive convergence demand it, which is normally much later than you think. Until then, keep your cash and your focus on being the obvious choice, not the loudest voice.

Feel free to ask Casey AI more about this topic.

Currently listening to Orchestra of Bubbles by Ellen Allien & Apparat.

This post is the third in an unintentional series on the levers founders reach for when trying to scale their company’s growth. This is beyond the day-to-day optimization a growth team might work on. This is about the next mountain that can 3x to 10x the company’s value. And a lot of times you run out of optimization opportunities along to get there. Growth starts to plateau, and no amount of optimization seems to be able to fix the problem. You need a bigger bet. Those bigger bets are usually one of:

• New products

• M&A

• New channels

• Brand marketing

You can read the links above for how to think about new product development and M&A. Today, we’ll talk about new channels, and in my fourth installment, we’ll talk about brand marketing.

Most startups find product/market fit with one acquisition loop like virality, Adwords, SEO, or sales, but they optimize that channel to asymptotic growth, and think they need to layer on new channels to grow. Let’s take a look at these channels.

Virality

Why It Scales: Users tell other users about the product

How to Sequence to It: Build sharing into the product, and figure whether the incentive that motivates users to talk about you to others is:

• Social: I look cool or smart when I talk about this e.g. Granola/Lovable

• Financial: I make money when people use this e.g. Uber/Dropbox

• Personal: I get more value from the platform when others are on it with me e.g. Whatsapp/Notion

Problems With It: Virality can die out much quicker than other channels as people by default only talk about the exceptional, and the exceptional becomes normal very quickly. When users reject the value prop, additional invitations tend to have lower conversion rates, requiring selling a new thing to get them to pay attention to a value prop they have already rejected.

Sales

Why It Scales: Use salespeople’s compensation to manage payback period of customers they bring in.

How to Sequence to It: Increase monetization to size that can support salespeople salaries and/or commission, and there is an easy way for salespeople to find leads, such as:

• Leads generated by marketing or product teams

• The ability to find leads on their own

• Channel partnerships

Problems With It: It is very hard to have very low payback periods, and only works for incredibly high lifetime value products like in B2B

Paid Acquisition

Why It Scales: Use value of purchases or lifetime value to fund more ads

How to Sequence to It: Improve lifetime value so that paid ads can drive healthy payback period

Problems With It: It tends to perform worse over time as ads always target the best possible customers first and then need to expand to customers who are less of a fit that then convert and retain worse. Network effect businesses are usually the only companies that can offset this trend, but new products can also offset this trend

Content

Why It Scales: Create a volume of content distributed that once it gets distributed can get enough eyeballs to bring in more users who also create content that can get distribution through channels like SEO, user sharing, online communities

How to Sequence to It: Amass enough content over time, typically through users creating content with the product, that can be distributed through:

• Search engines e.g. Google

• Users sharing to each other e.g. email, text

• User sharing to social or online communities e.g. X, reddit

Problems With It: Company generated content rarely scales, so product has to get users to create content, which can take a long time to become the main driver of growth as conversion rates on content are worse than invites

As one of these channels start to asymptote on the level of new growth it can create, startups look to expand into other channels on this list or new distribution forms for the current channel. You might experiment with distributing UGC through social media in addition to search like we did at Pinterest. While at Grubhub, we built SEO, Adwords, restaurant marketing, sales, TV and out of home into material channels over time. The key is over time. Certain things can happen over time that make new channels more viable. In the case of Grubhub, more supply meant better conversion and higher frequency, which started to make some more expensive channels profitable at real scale. As we drove more demand, restaurants saw more value, so they were willing to do more work to fuel more orders.

It is unlikely you just dramatically missed an important channel early on. Your first channel for growth is likely that channel for a reason: it was the most optimal one. So when you are looking for new channels, you have to ask the question: has my scale or product offering unlocked new opportunities for growth that didn’t exist when we unlocked our first channel. Perhaps you can take more risk on payback period because you’re better funded. Or you have a lot more content that you can distribute versus when you started. Or you now have a bundle of products that change your unit economics and lifetime value.

The mistake in expanding into new channels is when those things aren’t true. A lot of founders or growth people try to layer on new channels before scale effects make them more viable, and when that happens, these new channels tend to lack scale, and can at best add 5-10 points to growth, generally not what founders are looking for. A lot of the time it’s new products that unlock new markets or improve lifetime value that eventually unlock these additional channels, and they should be barking up that, admittedly, much harder tree instead.

Engagement loops are somewhat different in this regard and are generally a good idea to layer on over time, but their impact varies. At Reforge, we tend to categorize potential engagement loops into tactics that leverage:

• People: employees engage customer directly where increased engagement can produce enough of a lifetime value increase to pay for the employee’s salary

• Product: changes to the product that make it easier to use more often or features that increase utility

• Incentives: financial, utility, or social incentives that promote deeper usage like discounts, loyalty programs, and gamification (boy do I hate that word)

• Messaging: emails, texts, or notifications that bring people back more often

While activation is normally the biggest lever to improve retention, these tactics can improve the level of engagement for retained users. For a product in a low frequency category like travel, these may not move the needle that much. The higher the frequency potential for a product, the more engagement loops can double things like lifetime value, which of course then opens up new acquisition loop opportunities as well as helps the company grow revenue a lot faster and more sustainably. The key is to understand what headroom is possible from these efforts vs. considering them a panacea for growth when they will only be able to drive incremental impact.

–
Growth plateaus are inevitable, but they’re not your destiny; they’re a signal something needs to change. When your core acquisition loop begins to flatten, the answer isn’t to frantically stack every new channel you can think of. Instead, zoom out and ask what has actually changed since you first cracked product‑market fit:

• Has scale improved your unit economics enough to tolerate longer payback periods?

• Has your product breadth or user‑generated content base grown to unlock new surfaces for distribution?

• Have engagement loops boosted retention or LTV to the point where fresh acquisition bets now will be worth the effort?

If the honest answer to any of those questions is “not yet,” your time is better spent deepening product value rather than sprinkling on channels that can only deliver marginal gains. But when those answers have positive responses, you’re ready to test the next loop with conviction and the patience to let it compound.

Breakout channels aren’t discovered; they’re earned through stronger economics, richer content, and tighter engagement. Nail those prerequisites and the next mountain of growth becomes climbable. In the final post of this series, we’ll tackle the fourth big lever brand marketing and how to think about that lever appropriately.

Feel free to ask Casey AI more about this topic.

Currently listening to Oneiric by Boxcutter.

At most companies, if you’re not an engineer, you’re frequently stuck waiting: for access, for tools, for dashboards, for decisions. In an AI-native startup, that gap shouldn’t exist. Building in the age of AI is different. The rules—and the tools—are changing fast and with that the day-to-day operations of the team. We take an AI-native approach to running our company to make us more efficient. One rule we’ve recently embraced is to “onboard everyone as if they are an engineer”.

Historically, you can judge the effectiveness of tech companies based on how well they onboard engineers. Meta famously tries to get people to ship changes in their first week during their “bootcamp” that teaches engineers how things work at Meta. At some companies, it can be months before an engineer effectively commits code. Many companies are now attempting to re-onboard their engineers into AI-native engineering, driving adoption of new tools like Cursor, Devin, or Windsurf.

At SuperMe, our Growth Ops person, Anton, joined us a few weeks ago. Instead of the usual maze of tools and docs, we decided to switch things around and onboard them as an engineer. We set up their Github account, downloaded the backend repository, and set them up with Cursor. Our thesis was that instead of building tools or having to go through engineering to get things done, we can instead use our codebase as our internal tools suite. We had a thesis that code-assisted onboarding would unlock a faster, more empowered workflow. So we tried it.

On Anton’s first day, he was able to leverage AI tools to write complicated scripts to retrieve data and get a sense of the current state of the environment. Within his first week, he was able to create a tool set that allowed him to validate and sync account information and discover gaps in both our tooling and setup and validate where things are working fine. At most companies, Ops people depend on engineering and are always begging for engineering capacity. At SuperMe, Ops is doing the engineering.

Technical barriers usually slow teams down or lock them out. Our bet was that AI could flatten those barriers. Instead of manually stitching together workflows with GSuite and having manual tasks take hours or days, many of Anton’s early tasks took minutes. When AI couldn’t get all the way there, it at least covered 80% of the way. That created a much faster ramp to productivity than we’re used to from prior companies. At previous companies, new non-technical hires would surface tooling gaps that kicked off months of work, like building dashboards, configuring Retool for new internal tools, or buying and integrating new analytics or martech stacks. With this approach, we’re no longer stuck stitching together tooling; we’re just shipping.

At SuperMe, we’ve made the decision to onboard everyone like they are an engineer moving forward. The expectation is that you are going to use code-assisted tools regardless of role to directly access the data and automate tasks. You also do not need to understand the full technical depth to get and create a lot of value. Code is leverage, and that leverage is now available to everyone at the company regardless of their engineering experience. In an AI-native world, code is no longer a moat. It’s a multiplier. And that means everyone needs to wield it.

This approach can trigger a lot of red flags in people’s heads: What about database access? This can be very unsafe! What if a mistake happens? Similar to internal tools, it’s important to have a stance on what is safe or unsafe. We do force the usage to happen through our API layer, and that, while not bullet proof, allows us to maintain a level of sanity within the rest of the system. So far we’ve realized that our APIs are our best internal tools, and we’re leaning into our APIs and using AI tooling to generate personalized, on-demand internal tools.

Code is no longer just for engineers. And onboarding like an engineer is no longer just for engineers either. It’s the fastest way to unlock your team’s potential in an AI-native world.

It’s working for us. We think it’ll work for you too.

Anton’s already talked about his experience on his profile if you want to learn more. We’re learning just as much from these new teammates as they are from the tools. And you can always ask Casey AI or Ludo AI our thoughts on building a new startup in general.

Currently listening to my Fauxld School playlist on Spotify.

Some of the response is reflexive AI backlash, which is, of course, the protesting of the future that happens with all new technologies. Some of it is a reaction to where AI is today, which is… inconsistent at best in its ability to do jobs better than humans. Some of it is seen as an excuse to layoff employees.

But setting aside the backlash, there’s something important in what these CEOs are saying. I think the spirit of these manifestos is correct. I’ve seen teams and early users of our product SuperMe unlock real productivity with these tools, while others misuse them and don’t get much material value. And I also see some entrepreneurs leveraging the tools to build real companies of the future. Some are building enduring companies. Others are just building pit stops on the road to the future. They’re useful now, but will quickly become forgotten like early mobile apps or search engines prior to Google.

One of the best video games of all time is the original Portal. It is essentially a puzzle game where you have a portal gun that allows you to create a portal between two places and walk through those portals to get to places that otherwise would be inaccessible. The game is narrated by an AI named GLaDOS. After solving some rudimentary puzzles, GLaDOS says the next task is impossible, and when you solve the puzzle, GLaDOS says, “now you’re thinking with portals.” I think what CEOs are struggling with is they see a lot of employees kind of working the same way as they were before ChatGPT and all of these other tools came out or using these in products in only slightly incremental ways. Thinking with portals means using AI not to slightly improve existing workflows, but to do things that were previously impossible. Why is this so important to CEOs? Well, one other thing GLaDOS famously says, “If at first you don't succeed, you fail.” This really does feel existential to CEOs. Succeed at becoming an AI company before other companies, or fail to be a company long term. And I can guarantee you Tobi at Shopify has played Portal.

One struggle in thinking with portals i.e. being AI-first is that many AI tools are not always “portal-worthy” themselves. That is, they cannot fulfill their promise yet without a lot of supporting work from the human. We call this work prompt engineering. There are some brilliant prompt engineers out there that get way more value than people who sort of give up when more complex prompt engineering is required.

What’s interesting about this AI revolution is that for certain tasks the AI revolution has progressed far enough to make convoluted prompt engineering irrelevant, whereas for most tasks it is still the norm to receive a good outcome from an AI tool. One example is image generation. Midjourney, while revolutionary when it launched in February 2022, required some complex prompt engineering to get great images that matched what was in people’s heads. Now, image generation in ChatGPT can reliably one shot many image requests from a normal sentence. It’s amazing. That sort of leap will happen for most other use cases over time. This creates three sort of tiers of work in an AI-native world:

• The old way

• The half measure

• The one shot

Let’s look at another progression relevant for startups: data retrieval. There is:

• The old way: write a SQL query to get data

• The half measure: use ChatGPT to write SQL queries for you to get data

• The one shot: have Cursor write Python scripts for you to get data

I actually kind of liked writing SQL, but I will most likely never do it again. Rip.

If it’s mission critical for your job, learning the half measure can still be very important to get some productivity improvement, but keep an eye out for when the one shot solution is possible, and don’t try to become a specialist in a half measure that will no longer be needed in 12-24 months. For non-mission critical elements, perhaps you can just wait for the one shot solution to emerge, because it might not be far off. Using half measures a lot of the time can make it more obvious what the eventual future will be. This is what CEOs are trying to get at with memos to their team. It’s not all about cost savings.

If an entrepreneur, you should as much as possible be building one shot product experiences. When the technology can’t quite get there, perhaps find product/market fit with a half measure product that allows you to sequence to the one shot easily. Do not build half measures that do not sequence to one shot solutions you can own. Those are escalators to nowhere. Your fast ARR growth eventually will neither be recurring nor revenue, and your half measure is just market research for the other startup or incumbent to build the one shot solution instead of you.

What are some examples of half measures vs. one shots in startup land right now? AI notetaking and AI prototyping are probably the most obvious examples of half measures. We certainly won’t be using tools like this standalone as models continue to advance. Customer support tools like Decagon, Sierra, and Fin and coding tools like Cursor and Windsurf feel more like potential one shots that are pushing toward an end state of how things will work, with rising success rates over time, and becoming even more agentic. This is how you want the products you work on to feel: occasionally magical, and on the verge of becoming reliably so.

I could be wrong, of course, about which tools ultimately become the one shot solutions. Predicting the trajectory of AI has proven humbling, and I’m still learning how to think with portals myself. But the shift is clear: the future belongs to products and teams that aim for step changes, not slight improvements.

To get there, we need to stop optimizing old workflows and start designing for what was previously impossible. That’s what CEOs are really asking when they talk about becoming AI-first. They don’t just want to save money; they’re asking their teams to get good at thinking with portals. And if GLaDOS is right, “If at first you don’t succeed, you fail.” So you better start thinking with portals.

Currently listening to my Electro playlist on Spotify.

• Product Strategy Build, communicate, and execute a cohesive product strategy across feature, growth, scaling and innovation product work. Created with Fareed Mosavat, and hosted by Michael Sippey this Spring.

• Growth Strategy (formerly known as Advanced Growth Strategy). Build and communicate a holistic growth strategy, and identify opportunities to evolve how you grow. Created with Kevin Kwok and hosted by Adam Fishman this Spring.

Explore All Live Courses Here

The cohort starts the week of April 14 and spots are filling up quickly, so get on with enrollment.

If you’d like a sneak peek of the course content, you can attend this free event I’m hosting with Fareed on April 4th (at 10am PT) about The Science of Evaluating Feature Impact.

As AI accelerates product development, the ability to rigorously evaluate feature impact will separate category leaders from the endless stream of mediocre products. You can't rely on gut feelings and vanity metrics. Instead we'll discuss a systematic evaluation framework. We'll cut through the feature chaos to deliver a battle-tested methodology for assessing true business and user value.

Learn more and RSVP here.

Hope to see you there as well as during the live cohorts.

Currently listening to Baby It’s Cold Inside (2025 Remaster) by The Fun Years.

Give me feedback here.

And if you didn’t get a chance to kick the tires of my AI profile yet, you can still do so here. If you’re wondering what people asked, we did a little cluster analysis. Story checks out.

More cool stuff on the way.

Currently listening to Songs from the Smoking Section playlist on Spotify.

With AI profiles, you can chat with voice or text to my profile and ask any question, and my AI profile will respond based on all of the content I’ve ever created and show you where that content comes from if you want to go deeper. You can also use voice to brainstorm with my AI profile on any problem you’re facing. We’re excited to see what sort of questions people will ask, and how good of an experience it creates for you.

Chat with Casey AI here.

How it works
Just start typing into the question box any business problem you are looking for help in understanding. My SuperMe looks through all my blog content, presentations I’ve done, podcasts I’ve been on, etc. to aggregate the best possible information in response.

If you want to jam, it’s best to use the voice feature where you can share more about your specific problem, and Casey AI will go into a bit more of a brainstorm mode where it asks clarifying questions and give suggestions for you to give feedback on. Really excited to see what you want to jam on.

Internally, we've been having insightful conversations with several AI profiles on the platform, and my profile is the first one to go in live preview but there are others to come. What makes the AI Profiles different in our view is that they are grounded to how we as individuals approach problems. We strive to make the AI profiles opinionated and sourcing from individual experiences and helping promote critical thinking vs. be a yes man. They are a great way to learn about people’s experience, how they think about problems, and get their unique take on any business challenge you’re facing.

This is a special preview for my Substack subscribers. This is not being shared with the general public, and it’s still in the experimental phase and might break, have bugs, or not say the right thing. All feedback is appreciated. Feel free to send any to casey@superme.ai. Thanks as always for your support.

Currently listening to my For Carnivals and Music Boxes playlist on Spotify.

One key point to consider is that there are numerous companies that are experiencing rapid growth, reaching $10 million to $100 million in annual recurring revenue faster than we previously thought possible. However, this does not necessarily mean these will be beneficial companies to join from a career standpoint. While initial revenue growth may be enticing, it's crucial to evaluate if these opportunities have longevity, especially given the rapid changes brought about by AI.

When evaluating career prospects, it’s important to take a long-term view and consider whether these roles will remain relevant and secure in the future. The classic Bezos quote of what won’t change in the next ten years has never been more relevant. Despite initial success and excitement, a lot of the first wave of AI companies could be half measures to deeper technological innovation that may win out long term. Or incumbents with better distribution may cut off the growth of some of these companies with copycats. A key question here is do these companies control their own growth loops, or are they mostly word of mouth. Word of mouth growth always dies off after periods of technological change slow down.

For those aspiring to become leaders in their field, it is advisable to align with companies that have the potential to become iconic. While some companies may be successful and generate substantial revenue, they might not achieve that iconic status which provides substantial professional leverage and opportunities. It’s crucial to make informed decisions about where to invest your time and talents.

And once you’ve worked at an iconic company, I think a lot of people under-weight the problems with switching to a non-iconic company that may offer them better comp or a better title. Once you’re working at B-list or worse companies, it becomes harder for iconic companies to consider you for their openings.

Ultimately, making smart career decisions involves assessing whether the company has the potential to evolve and succeed over a longer timeline. Even if things don’t work out as planned, recruiters and future employers will look favorably on the decisions you made based on the available information. They will analyze whether they think you made a good decision at the time. So if you join a hot AI company like say Eleven Labs or Glean and they cool off, that will still not work against you in the market. If you however joined Getaround or Nextdoor and they don’t turn around, people will think the weird decision you made with your career reflects negatively on the decisions you might make for their company.

Most gen one companies during technology shifts are not durable. Sure Amazon did well as an early internet company, but 99% died. Sure Instagram did well as an early mobile company, but foursquare and Urban Spoon and thousands of others faded away.

I think what is also critical is joining a company that is changing the way it works to stay current with AI. If none of the engineers are using coding assistants like Cursor, and people aren’t prototyping with tools like Bolt, maybe beware that they will be sunk by companies that are leveraging those tools to be more productive. You can’t sit on the sidelines when the world is changing this fast.

Especially for early-career individuals, investing in developing skills that will be crucial at prominent companies in the future is essential. Identifying and acquiring these skills will better prepare you to excel in interviews and positions with top companies looking to innovate and lead in their sectors.

So, to recap, don’t just look at current growth. Project out can the company remain elite five to ten years from now. Think about current or future iconic company potential, and don’t let short term optimizations push you away from the best companies. And make sure the way they work and you work align to what the job market will value today and in the future.

Currently listening to my 2004 playlist on Spotify.

First, it's essential to understand the principal motivations for engaging in M&A. Generally, there are three main reasons. One, to achieve greater scale, which involves acquiring a competitor to boost market share and enhance profitability by centralizing certain operations. Two, to increase scope, which allows a business to enter new markets or product lines, thereby enhancing the range of capabilities and offerings. And three, to acquire capabilities, such as skills, expertise, or technology, which can enable a company to innovate and potentially launch new business lines without starting from scratch.

Traditionally, marketplaces have utilized M&A primarily to scale by acquiring competitors, thus consolidating market shares. Grubhub’s merger with Seamless was a strategic move to cement our position as the leading food delivery player by expanding our reach in New York and other markets where we already competed against each other.

For scope, we acquired CampusFood to secure a foothold in college towns, which was essentially an extension into new geographic territories while ensuring national coverage faster than new market development would have allowed.

At Eventbrite, when we acquired Toneden, we acquired it for their marketing capabilities so we could launch a martech suite in the future.

When thinking about what to do with the acquired company, it's crucial to orchestrate the integration process thoughtfully. Many executives tend to rush integration which may lead to unnecessary complications. Here are three strategies for integration: full integration, partial integration, and maintaining operational independence. For example, when we acquired CampusFood at Grubhub, we opted for full integration by retiring their brand and assimilating their markets into the Grubhub brand. However, with Seamless, we retained their branding online to maintain SEO rankings, thus avoiding competition increases from platforms like DoorDash or Uber, which could have gained ground otherwise. We can debate whether either of these was the right strategy.

Timing and methodology in integration should be strategic too. Integrating too quickly and fully, as we observed with Eventbrite's acquisition of Ticketfly, led to customer and employee dissatisfaction due to misalignment in product capabilities and market needs. Robinhood did a great job leaving their acquisition X1 largely alone until the product was closer to being available in market, operating it as a separate business unit.

Finally, it is essential to evaluate whether M&A is the best option for your strategic goals. There might be cases where partnerships or internal development could be more advantageous considering the cultural and technological implications. A thoughtful evaluation can prevent costly mistakes and ensure the acquisition fulfills its intended objectives.

Whether choosing to go through with an M&A or opting for an alternative strategy, careful analysis and deliberation are paramount. I hope these perspectives provide some guidance as more organizations navigate the intricacies of M&A in their growth journeys, because when done well it can be an extremely profitable growth strategy.

Currently listening to my Instrumental Hip-Hop playlist on Spotify.

I’d like to talk about scaling marketplaces today and how the underlying platforms make us rethink good marketplace growth strategy. Traditionally, marketplaces have thrived by scaling both supply and demand, and it’s crucial to recognize that elements such as retention, lifetime value, and unit economics are often enhanced by scaling. Many notable marketplaces have historically built their success on platforms like Google while navigating various acquisition strategies, including robust investments in paid channels such as Google Ads, Facebook, and even job boards in the cases of Uber, Lyft, and DoorDash.

As we reflect on the landscape of 2024, it becomes evident that the once-stable avenues of customer acquisition are experiencing unsettling shifts. These trends, emerging through technological strides, notably artificial intelligence, and impending disruptions from market newcomers, press marketplaces to adapt. Entering the current era, marketplaces face a demanding challenge: they must excel in their customer acquisition channels earlier in their development. Previously, firms could afford a more moderate competency in paid advertising and SEO. Today, however, they require superior mastery, possibly even at the seed stage, and undeniably by Series A funding. For instance, Google's strategy shift from merely ranking aggregators toward embracing marketplace features themselves means a decline in organic impressions for marketplaces.

Marketplaces now must strive for top placement as Google's algorithm evolves to balance visibility among smaller businesses, startups, and established entities. This trend necessitates an agile approach where businesses anticipate platform directions and strategically maneuver. Consider businesses operating within shopping-related sectors. Google has pivoted more of these impressions to its own shopping interface. These marketplaces must recalibrate strategies to synchronize with Google's product direction, such as emphasizing individual product listings in lieu of aggregated product pages that were previously successful. In addition, marketplaces may need to more radically change their product to align to these platform changes to grow sustainably whereas the last generation of marketplaces could just focus on geo or category expansion alone. It feels weird to say, but platform changes from Google and Meta can change whether you have product market fit at all.

Challenges arise as these new marketplaces, targeting more niche markets than their historical predecessors in travel, transportation and food, need core product value expansion earlier than their predecessors to unlock growth and a venture scale TAM all while the underlying acquisition channels are also changing. Think about how Uber might be different if they needed Uber Eats to launch and work in year three vs. year seven of the business. And during that same year what if Apple changed targeting options for Facebook ads with the app tracking transparency changes. That is kind of what is happening.

A dual focus on maintaining user satisfaction and sustaining scalable customer acquisition becomes critical. As channels grow more rigid, adeptly refining product-market fit to unlock more user acquisition from these channels is necessary. This could be optimizing for better SEO ranking in Google’s new surface areas or exploiting social media visibility and figuring out how to get them to convert onto your site. This necessitates growth teams to rethink their methodologies: instead of merely enhancing SEO, virality, conversion rates, or onboarding practices, they need to understand where these channels they rely on are growing and work with core product teams to change their product market fit to be able to leverage these channels better moving forward.

The stakes are elevated—marketplaces must now operate with precision and adaptability akin to pitching a perfect game. We used to say product market fit is a hell of a drug that can mask a lot of operating issues. And it did for the last generation of marketplaces like Airbnb, Uber, DoorDash, Instacart, et al. In fact the best companies were much of the time the poorest run companies, because they are the only ones that can survive being poorly run because of their underlying product market fit. But the next generation will need to operate a lot more flawlessly to have that type of success. Marketplaces can still win big, but execution has gotten exponentially harder.

Currently listening to my 2024 playlist on Spotify.

Let's examine some key case studies. If we go back to the first era of marketplaces, we can observe examples like the online travel portals. In these early battles, two principal players emerged: Expedia and Booking.com. Today, Expedia is valued at approximately $19 billion, which is pretty nice. Yet, if you compare this with Booking which is worth nearly $140 billion, there's a substantial gap in valuation.

Another classic example is seen in the real estate market with Zillow and Trulia. Though they eventually merged, Zillow's valuation was about double that of Trulia's at the point of acquisition. This pattern isn't restricted to the early days of marketplaces; it carries over into the next generation from 2010 onward.

Consider Lyft and Uber in the transportation market. While Lyft holds a significant valuation of $5 billion, Uber dwarfs it with a valuation of over $150 billion. The same pattern appears in food delivery; Postmates, which sold for 2.5 billion, and DoorDash, now valued at $60 billion, both started around the same time with innovations in food delivery methods.

Through these examples, one glaring insight emerges: the concept of being slightly better at compounding over time creates a vast disparity in company valuations. This subtle yet constant advantage in growth can make you about ten times more valuable compared to your closest competitor. To achieve this, you need to excel in aspects like having an astute growth team and efficient supply expansion, among other factors that enhance network effects.

In conclusion, when examining competitive marketplace spaces, remember this: even if there are two similarly sized competitors, always staying slightly ahead will pay off profoundly over the years—far more so for marketplaces than SaaS. If you are part of such a marketplace, I hope you have a good growth team.