Marketers have spent the past decade optimizing for frictionlessness. Whenever possible, we remove steps, pre-fill fields, and demonstrate value as quickly as possible. Any ounce of friction is assumed to be a place to lose potential customers.

This approach makes sense in a world where products compete solely on feature sets, and where a buyer moves linearly through stages of a funnel determined by a marketing team: "see an ad, read a case study, do a demo, buy a product." In that world, we want to move people as efficiently as possible from the top of the funnel (awareness) to the bottom (conversion).

But we are so far from that world today. Thanks to AI, it's becoming harder for companies to differentiate on features alone. We aren't choosing products based on features or functionality anymore. We're choosing based on how they make us feel, on whether they validate the version of ourselves we want to be, or want the world to see.

The friction worth keeping is cognitive: the work of learning a new vocabulary, building a new mental model, going down a rabbit hole on an unfamiliar concept. Not arbitrary gating, not FOMO-driven waitlists, not manual processes that exist because no one bothered to remove them. Buying behavior changes; human psychology doesn't. People still feel the pull of sunk cost after they've put real time into understanding something, and that effort is what turns a user into someone with actual affinity for the product.

That friction shows up in two forms. The first is literal product onboarding, or the experience of learning a new system. Take Claude as an example: to use it well, you have to set up connectors, learn how to prompt effectively, understand what an agent and a skill are, and know the difference between Chat, Cowork, and Code. Anthropic hasn't built an onboarding flow that holds your hand through setup, and I wonder if that's intentional. The learning becomes the product itself. Once you go through onboarding, you emerge with a shared language and a sense of belonging to a new group. 

The second form is conceptual onboarding. It's the rabbit hole someone goes down before they ever touch the product, most common in businesses like deep tech with long sales and education cycles. When the cycle is that long, and there's initially nothing to buy and no one to convert, conventional product marketing doesn't work. The work is educating the market, over years, on why your solution will be the right one when it finally exists.

Either way the mechanism is the same. Cognitive effort creates commitment, and commitment is the difference between a user and a champion.

That champion matters more than ever, because buyers no longer take the company's word for anything. They want to hear what a product is worth from someone they already trust. Social proof used to come from the company: logo walls, testimonials, badges, case studies. That kind has been collapsing for years, and AI has finished the job, since everyone now assumes anything on your owned channels was ghostwritten or fabricated. What's left is earned social proof: a Reddit post where someone shows what they built, a YouTube tutorial that teaches the product better than the company can, a Discord conversation between users who don't know the company is watching. The customer who fought through friction is the same one who is producing social proof in the public sphere to signal something about themself.