TLDR: How to identify and avoid startup idea space that are traps.
Recently YC partners Dalton Caldwell and Michael Seibel hosted a startup school session: Pivoting out of a tarpit idea. The session gave an VC perspective on common pitfalls in searching idea spaces with a focus on tarpit ideas: ideas that are are much more difficult to pull off than they appear and are very difficult to pull out once in.
I think this is a hot-take: the conclusions are quite opinionated, drawn mostly from observations and heuristics with limited (or none) supporting fundamental theories. Nevertheless I appreciate and respect YC’s openness in sharing their thoughts that may seem raw and unpolished. I’d also like to share my take-away and comments:
Characteristics of tarpit ideas
YC shared the follow common categories and characteristics of tarpit ideas:
ideas that have survivor bias: user problems that may look solvable from a posterior, results oriented perspective. For example, discovery problems that help people find out new restaurant to try, new places to go.
My comments: 1. I would argue that discovery is a problem being solved by TikTok For You, Instagram Reels and Google Discover; However, being able to witness first hand how Google’s increasing magical discover experience; I tend to agree that it’s not an idea space for start ups. A good solution requires huge amount of user data and deep machine learning recommendation technologies, either prerequisite favors large aggregators and plays against startups. 2. It’s also worth call out this idea space appears to be easy to solve due to mental bias, which is something a founder should try to avoid as much as possible. Take the discovery as an example: sure there are always new fancy restaurants coming up, new places worth to go, and it’s happening every day. However on the other hand a founder needs to be aware those survivors are only a fraction of new restaurants coming up, new businesses created. Only a small head of the long tail would success and that’s why large amount of user signals and deep computing powers are a must to discover the winners. The same applies to the creator economy example YC gave, that any services toward serving successful content creators is difficult to build because only a tiny fraction of the creators would be really successful and the vast majority would not make it. 3. I do want to add that being in a tarpit idea space should not intimidate well-crafted problem solution pairs that are mindful of mental bias. Finding gold in gold-rush is difficult but that does not stop Levis becoming a widely adopted consumer brand.
ideas that are second-order on widely adopted innovations. Innovations that have step function improvements bring many low-hanging second order opportunities. Yet the cost of entries will be increasingly high as time and adoption go on. In the web2 era triggered by smartphone, a social network might be a good idea but in a post-web2/pre-web3 era, social network idea space is becoming a red ocean.
ideas that are more difficult for startups to scale. This specifically refers to eCommerce and hardware startups. Scaling a business in the areas require continuous investment in operations that also favors large companies. Startups usually becomes a free R&D for large companies and usually the best outcomes SuSopportunities
YC made multiple disclaimers that some ideas could be tar pits temporary and there have been examples of great companies solving tar pit ideas. However, the change is predicated on fundamental/core technology changes that resolve blockers. E.g. smartphone for Uber. Cloud computing and large data for Neural network.
My comment: If we apply the “good ideas do not have competition” theory by Peter Thel to generate ideas and filter them by the extra dimension of “is tar pit”, there are two distinct types of ideas worth mentioning: 1. Ideas that are seemingly difficult but are actually great business opportunities (e.g. ideas with high barriers of entry). 2. Ideas that are difficult because they were tar pit ideas (i.e. seeming appeal to solve, but turned out to be difficult). The second type of idea are sometime great business opportunities (or can transform to type 1 ideas) if there are mature technology break throughs that remove the underlying “tar”.
How to tell if an idea is “another tar pit idea”?
Dalton and Michael shared their rule-of-thumb: for consumer products, good ideas should easily have immediately, explosive organic growth. Here a consumer product is defined as services that any random people can consume, with an very large intial user base (100 millions, e.g. netflix, facebook). Examples of good ideas by then (tarpit by now:
Facebook: started with exponential growth from the very beginning: took over Yale in one week.
Twitch: on the edge of bankruptcy at growth stage, not because of user traction, but because of services overload.
There are a few naunances for dicussion. First, I’m sure that not all consumer products have a user base on a scale of 100s millions, and I can give numerous examples of vertical companies that serves a specific consumer segment (health care, sports, review sites etc.) and are doing well. My take is that YC has a strong preference to invest in consumer products targeting millions of users that organically grow exponentially, and that this may be true for all VCs. I can see one rationale behind this investment strategy: consumer products space are very competitive due to its low barrier of entry. To maximize investment return VCs need to tighten up the bets and focus on startups with the highest predicted/projected success rate.
Second, I want to highlight that consumer products are becoming easier to build, as the building blocks/tools are becoming mature: cost of cloud services are dropping, building an app is not possible with low-code/no-code tools; and of course the better than ever UI tools (figma). I reckon that successful consumer products probably donot need VC funding until ready to scale.
How to pitch VC a “tar pit” idea?
YC is suggesting against pitching tarpit ideas. The only alternative is to show that you have done extensive research of the “tar pit” and all the fossiles. Be prepared to answer why us and why now.
Reference
Twitter threads on the session by Tanmay Dhote, Jake Stevens and Alex Scott.
Session notes by Sam Shapiro.
Excellent job thanks
Nelson