A while back I did what a lot of indie makers do: I looked at the AI image boom and thought, "I'll build a generator too." It took about three weeks of staring at flat traffic to admit the obvious — a generic "AI image generator" is one of the most saturated, giant-dominated categories on the internet. You're not competing with other indie hackers. You're competing with companies that own the top of every SERP.
Here's the shift that turned it around, and the playbook I'd give anyone staring at a red ocean.
Stop selling the engine. Sell the specific pain.
Nobody wakes up wanting "an AI image generator." They wake up needing to remove the background from a product photo, restore a blurry photo of their grandparents, turn a selfie into an anime avatar, or get rid of a watermark. Each of those is a concrete job with its own search intent, its own keywords, and far less competition than the head term.
Build a matrix, not a monolith.
Instead of one homepage trying to rank for "AI image generator," I built one focused landing page per job. Same generation backend underneath, different copy, different URL, different keyword target on top. Each page does one thing and answers one query completely. The long-tail pages pull the traffic; the core generator is what monetizes it once people land.
Go horizontal first, then vertical.
Early on, breadth beats depth. Ship many different tools and see which ones get impressions and clicks in Search Console. Don't guess which niche will win — let the data tell you. Once a page starts pulling traffic, then drill vertical: spin up long-tail variants around that proven winner.
Let demand decide, not ambition.
The expensive mistake I see repeatedly (and made myself on another project): bulk-translating into a dozen languages or shipping 50 pages on day one to "scale." Thin pages dilute your whole domain's quality signal. Add a tool or a locale when the data shows real pull for it — not because the spreadsheet looks bigger.
None of this is glamorous. It's a lot of small, boring pages each solving one real problem. But it's the only approach that gave a solo maker a foothold in a category owned by giants. If you want to see the matrix in practice, this is the project I've been building it on: z-image-ai.run.
Curious how others here have approached saturated categories — niche-down-then-matrix, or did you find an angle the big players structurally can't copy?
