How to position against bigger incumbents

1 The bottom line

Pick one specific opening the incumbent cannot or will not close, such as price, speed, an off-ramp or a neglected customer, and make it dramatically better. Copying the leader feature for feature gives buyers no reason to switch.

2 Key takeaways

• Start with a customer who is blocked, not a copy of the leader. A buyer who is not blocked will not switch for something better or cheaper.
  • Rob tried a Linear-style better UX in HR software and it failed because the buyer did not care about UX Rob Snyder, Lenny's Podcast1:24:20.
• Attack the incumbent where its model hurts customers: fees, friction or lock-in. Customers already complain there, so the opening is real.
  • One in seven new beehiiv writers now come from Substack Tyler Denk, beehiiv27:30.
• Make the switch obviously worth it, judged by what current users would cheer. Good enough does not move people.
  • Words with Friends reached 14 million DAUs while Scrabble itself did not Mark Pincus, Zynga8:30.
• Against AI labs and general chatbots, be opinionated and own what they cannot. A blank chat page gives the average answer.
  • SuperMe tested its expert-sourced answers against ChatGPT and Claude and found them better Casey Winters, SuperMe19:50.
  • People can already track travel in ChatGPT but still choose Flighty Jeff Morris, Tinder42:43.
• Protect the opening early, because simple products get copied. Once a wedge works, others clone it.
  • Competitors copied Zero's UI pixel for pixel, sometimes even the name Mike Maser, Zero24:23.

You could give me a 100x better printer for my house, and I won't buy it cuz I'm not blocked.

Rob Snyder, Lenny's Podcast1:24:20

2.1 Find the opening

Look for a customer the leader leaves blocked, not a gap you can fill by matching its features.

• Do not build the same product as a visible winner. Source: Rob Snyder, Lenny's Podcast1:24:20
  • Ask whether you have a theory of a different customer who is unblocked.
  • Better or cheaper is not enough on its own.
  • Rob calls copying a winner without that theory a common tar pit.
• Avoid the parity trap. Source: Claire Vo, ChatPRD10:51
  • Competitors talk to the same customers, pull the same insights and reach the same obvious conclusions.
  • Aim for something surprising that competitors could not imagine.
  • Vo's product graph matched competitors feature for feature. She felt it belonged in the trash because it was competitive but not differentiated.
• Aim at people the incumbents are not reaching. Source: Kenneth Schlenker, Opal6:07
  • Target people who say they spend too much time on their phone but will not download a blocking app.
  • Offer no setup and coaching instead of restriction.
  • Eight out of ten people asked say they spend too much time on their phone, yet few would download an app to manage it.
  • Copycats limit themselves by copying existing features.
• Reinvent an entrenched category with AI before the incumbents do. Source: Jacob Eiting and David Barnard, RevenueCat36:20
  • The first category definer rarely stays on top.
  • In niches like meditation, someone has to build the AI-first version, and incumbents must be willing to break their own eggs.
  • Claude's consumer app rose from hundreds of thousands a month to almost a million dollars a day, even though ChatGPT seemed to have won (App Figures data).

2.2 Pick your wedge: price, speed or an off-ramp

Flat fees beat a take rate as publishers grow

	Substack	beehiiv
Pricing model	% take rate on revenue	Flat $49 / $79 / $159 a month
Take rate	Grows with your revenue	0%
Cost as you grow	$10k–15k a year in fees	$79 a month
Big publishers	Leave as they scale	Time, Newsweek switched
Direction	Social platform	Enterprise software for prosumers

Choose the one thing the incumbent's business model stops it from fixing, and build your pitch on it.

• Counter-position against the leader's most hated pricing. Source: Tyler Denk, beehiiv27:30
  • Find the top complaint on X about the leader. For Substack it was fees of $10,000 to $15,000 a year as publishers grow.
  • Charge flat SaaS fees of $49, $79 or $159 a month by size, with a 0% take rate. Passing money through Stripe is not value worth extracting.
  • A percentage take rate looks harmless at small scale but becomes meaningful as publishers grow, so the biggest ones leave Tyler Denk, beehiiv30:00.
  • Time and Newsweek switched for more features at about 40% of the cost. ESPs charge from $100,000 a year for what beehiiv offers at $79 a month.
  • The platforms also diverge. Substack is building a social platform, while beehiiv brings enterprise-level software to prosumers.
• Win on speed and simplicity for a price-sensitive audience. Source: Daniel Heintzman, Cal AI34:45
  • Cal AI targets a younger, more price-sensitive audience that wants fast logging. MyFitnessPal serves serious users who accept friction for more features.
  • Simplicity opened the category to people who had never tracked calories.
  • Cal AI later added barcode scanning and a food database to improve accuracy.
  • Phil Carter ties the $50M+ ARR in two years to Brian Balfour's four fits: product, market, channel and model.
• Offer an off-ramp where the incumbent profits from dependence. Source: Mike Maser, Zero40:00
  • Telehealth GLP-1 incumbents earn $100 to $500 a month by keeping customers on the drug with no exit.
  • Zero plans to prescribe the drug, guide users through it, then offer a home base for maintaining weight and metabolic health after they stop.
  • Mike calls this a disarming position that he has not seen done elsewhere.
  • He says the most common problem after stopping a GLP-1 is that the weight comes back along with muscle loss.

2.3 Make the switch worth it

An opening only works if the product behind it is clearly better, and better is defined by users, not by you.

• Build something dramatically better, not merely good enough. Source: Marty Cagan18:39
  • Solving problems customers love and that work for the business is not enough.
  • To get people to switch, the solution must be dramatically better than the competition.
  • Cagan says to prepare for the competitive open market as a full contact blood sport.
• Define better as a polished step that 10 out of 10 existing users would cheer. Source: Mark Pincus, Zynga8:30
  • Think small, polished increments such as free, no download, or smoother mobile play.
  • Judge it by what current users want, because your own idea of better is really just new.
  • Words with Friends was Scrabble with mobile polish and social built in.
  • Power users notice polish most.
• Invest in design and craft, since specialized apps win on them. Source: Jeff Morris, Tinder42:43
  • Jeff expects the design bar to rise, with an anti-slop mood favoring high quality and thoughtful products.
  • A strong point of view on interface and workflow separates specialized apps from general-purpose super apps.
  • The winners are still the people who think about product, experience and craft Rik Haandrikman, RevenueCat27:30.
  • Survivors also use game mechanics, like a Pokedex-style collection for house plants.

2.4 Beat general AI with focus and data

When the incumbent is a general AI model, win with a dedicated job, trusted expertise and outcomes the labs cannot see.

• Sell a coach for one job, not a blank chat. Source: Adam Turaev, Praktika23:09
  • Build context, an opinionated experience and product taste around the job, such as a better job through English.
  • Praktika offers a coach with a CEFR-level assessment and a safe, judgment-free environment, not a sparring partner.
  • In high-stakes areas like career, people want a dedicated product that solves one problem. You need an extremely opinionated interface that delivers dense information in an engaging way Tomas Hoyos, Apt53:40.
• Bring expertise and frameworks a chatbot cannot fake. Source: Casey Winters, SuperMe19:50
  • General chatbots return the average of what the internet thinks. Source only the best people in growth, product or engineering.
  • Show each expert's background and the companies where they applied the advice, because trust rises.
  • A famous name plus content is not meaningfully different from asking ChatGPT to act like that person. Purpose pairs Mark's tone with scientifically validated frameworks in its Quest feature Raj Singh, Purpose40:20.
• Build where the model companies lack outcome data. Source: Koby Conrad, Sunflower26:36
  • Track outcomes like relapse, days sober and loneliness, then optimize the whole system around them, including tooling and function calls.
  • You do not need to train a new model. Build where the agent is only part of the solution.
  • OpenAI and Anthropic do not know whether users are less depressed or staying sober, and they carry legal liability. Koby says OpenAI people told him in a YC talk that this kept them out of behavioral health.
• Differentiate on emotional connection, not agentic task features. Source: Ajay Mehta, Tolan49:50
  • Build a friendly guide that learns about your life, not a transactional assistant.
  • Meet the user differently and ask about different things each day, since the capability floor will rise.
  • Tolan's onboarding forms a bond in the first 60 seconds, which raises activation and retention. Across AI apps, revenue per payer is 41% higher but churn is 30% faster Rik Haandrikman, RevenueCat27:30.

2.5 Say the opening out loud

State the opening as a clear premise, frame it against what users already tried, and be bolder than feels comfortable while you are small.

• Name the underserved group and the incumbent's flaw in one contrarian line. Source: Weber Wong, FLORA12:57
  • FLORA launched with the premise that current AI creative tools are built by non-creatives for non-creatives, and that it would build one for professionals.
  • The launch went viral among professionals, because there had been no overlap between AI model users and professionals before.
• Test messaging that contrasts you with the alternative users already tried. Source: Raj Singh, Purpose17:17
  • Purpose A/B tested top-of-funnel messaging and found "self-help books don't text back" resonated.
  • Many users had already tried self-help books, therapy or other DIY options, so the line frames the product against what they know.
• Punch above your weight early with bold design and extreme focus. Source: Casey Winters, SuperMe46:00
  • With few users, put maximum design into a minimal feature set so people pay attention.
  • Brands get less divisive as they scale, so go bigger when you are early.
  • At GrubHub, offline ads on Chicago trains cost about one hundredth of what Verizon might spend. They used construction paper design and cursing characters showing the irrationality of hunger.

2.6 Defend the opening

A working wedge gets copied, so add stickiness, ecosystem and brand, and watch the competitor that is not in your category.

• Add distribution, community or stickiness to a simple product before copycats arrive. Source: Mike Maser, Zero24:23
  • A dead simple product is easy to duplicate, and AI makes UIs even easier to clone.
  • Simplicity without distribution, community or something else that keeps users sticky leaves a compromised moat.
• Build a moat from speed, ecosystem and brand, not model quality alone. Source: Jack McDermott and Tanmay Jain, ElevenLabs53:00
  • Large labs will ship good-enough voice models, so keep doubling down on your focus.
  • Add platform depth, such as custom evals, workflows and integrations in voice agents, and compound earned media.
  • Advantages accrue at the application layer and in brand. People name ElevenLabs first when they think of voice.
• Treat all screen time as your competition and make the product fun. Source: Cem Kansu, Duolingo53:40
  • Duolingo's churned users mostly go to Instagram, YouTube or other fun phone activity, not other education apps.
  • Learning takes more effort than passive scrolling, so invest heavily in making the app fun.
  • Duolingo is already far larger than any education app.

2.7 Where they disagree

Sources split on how big the improvement must be, whether rivals are enemies, and whether AI apps can live beside general chatbots.

• Small polish or a surprising leap. The right size of improvement is contested.
  • Pincus favors small, polished steps that all existing users would cheer Mark Pincus, Zynga8:30.
  • Cagan wants dramatically better, and Vo wants something competitors could not imagine Marty Cagan18:39, Claire Vo, ChatPRD10:51.
• Rivals as enemies or as market educators. The stance depends on how many people have yet to adopt the category.
  • Cagan calls open markets a full contact blood sport Marty Cagan18:39.
  • Opal says more screen time apps help educate the market, where only a few dozen million of 5 billion smartphone owners use one Kenneth Schlenker, Opal6:07.
• Coexisting with ChatGPT or being absorbed by it. The answer seems to depend on how much the product relies on the model alone.
  • Many Tolan users also use ChatGPT, using Tolan for brain dumps on the way to work or chatting on the couch Ajay Mehta, Tolan49:50.
  • Apps whose core function depends on a model API are being absorbed by general AI agents Rik Haandrikman, RevenueCat27:30.
• A price wedge does not fit every product. Free tiers are hard when every use costs money.
  • AI wrappers cannot afford free users because usage is paid API calls, so a beehiiv-style price attack is harder for them Rik Haandrikman, RevenueCat27:30.