A small, free passport-photo tool, the day I realized search had quietly split into two games, and why being "the answer" is not the same as winning.
I build a free tool that makes passport and visa photos in your browser. No signup, no watermark, nothing leaves your device. It is a small thing. I made it because I had overpaid for a drugstore photo one too many times, and because the problem is more annoying than it looks once you count how many countries have their own millimetre rules.
For the first months I did what everyone tells you to do. I wrote the pages, I filled in the meta tags, I checked the titles, I waited for Google. And Google did what Google does to a young domain with no history: almost nothing. I sat far down the second page for the terms I cared about, watching sites that were older and better-linked sit comfortably above me.
Then I looked at where my actual visitors were coming from, and the number stopped me. The single largest source was not Google. It was ChatGPT. People were asking an assistant "where can I make a passport photo for free" and the assistant was sending them to me, by name, in its first answer. Not after a follow-up question. In the first answer, for a whole cluster of the queries I cared about most.
I want to be honest about what that felt like, because the honest version is more useful than the triumphant one. The first feeling was not pride. It was confusion. I had done nothing that looked like "AI marketing." I had no growth hack. I had a clean, boring, accurate tool. So why was the newest channel rewarding me when the oldest one was not?
That question is the whole essay, so let me try to answer it.
Search has quietly split into two different games, and they reward different things.
The old game, the Google game, mostly rewards authority. Who links to you, how long you have existed, how much the rest of the web has already decided you matter. This is sensible and it is also brutally slow for anyone new. You can have the best page on the internet for a topic and still wait months, because authority is a reputation, and reputations are built on time and on other people, not on your own effort alone.
The new game, the assistant game, rewards something different. An assistant answering a question in real time is not asking "who is the most famous." It is asking "who is the clearest, most trustworthy, most machine-legible answer to exactly this question, right now." Those are not the same question. And the second one, it turns out, a small careful builder can actually win.
I did not win it with a trick. I won it by being genuinely easy for a machine to understand and genuinely safe to recommend. When an assistant points a stranger at a tool, it is putting its own credibility on the line. It will reach for the option least likely to embarrass it: the one that is free with no nasty surprise, that does not harvest your data, that actually does the specific thing asked, that states its facts plainly enough to be quoted. I had spent my effort making the tool honest and the information precise, not because it was a strategy, but because that was the product I wanted to exist. The assistant rewarded exactly that.
So the first lesson is almost embarrassingly old-fashioned: in the assistant era, being trustworthy is a distribution strategy. The qualities that used to be "nice to have" - no dark patterns, plain claims, doing one thing well - became the qualities that get you named.
The second lesson is the one I keep having to relearn, and it is why I am wary of the word "won."
Being the answer an AI gives is a real achievement and a fragile one. It is a layer, not a finish line. My tool is the first recommendation for a set of queries on one assistant. That is wonderful. It is also not the same as being found by everyone, everywhere, for everything. Other engines weigh authority more heavily, and there the old slow game still applies to me, exactly as it always did. A win in one layer does not hand you the others. It just tells you the building is possible.
I have watched founders, myself included, take an early, narrow victory and quietly relabel it as the goal. "We are #1 on ChatGPT for X" is a sentence that feels like arrival. It is not. It is a strong signal that you have something worth compounding, which is a different and better thing, because it points forward instead of letting you stop.
The third lesson is about what to do with a young domain while you wait for the slow game to turn. The honest answer is: be specific. Broad, generic terms belong to the giants for now. But the long tail - the precise, real, slightly awkward questions that actual people actually type - is wide open, and a careful tool can own it one exact question at a time. There is more durable traffic in being the unmistakable best answer to a narrow real need than in fighting for a head term you will lose for another year. Specificity is the young builder's advantage, not a consolation prize.
If you are building something small and new, here is what I would take from this, stripped of any hype:
Make the thing genuinely good and genuinely honest, because honesty is now legible to the machines doing the recommending. Write so a stranger, human or model, can understand and quote you without guessing. Do not wait for the slow authority game to bless you before you exist; go win the precise questions you can actually win today. And when an early layer rewards you, enjoy it for an afternoon, then treat it as a signal to compound, not a reason to coast.
I am not going to pretend I have this solved. I have one tool, one good surprise, and a long way to go before "people can find this" is actually true at the scale I want. But the surprise taught me something I did not expect to learn from a passport-photo app: the web did not just get a new search box. It got a new set of rules about what deserves to be recommended. And for once, the rules quietly favour the careful over the loud.
That is a web I am happy to keep building for.
I am the founder of IDPhotoSnap, a free passport and visa photo tool that runs entirely in your browser. This essay was originally published on Medium.
Top comments (0)