How we made BrainDump easier to find on Google and in AI answers

We built BrainDump to keep up with racing thoughts, not to sit in a drawer unread. The same instinct should apply to our own blog. We have published more than 60 practical guides on focus, note-taking, and using AI to turn scattered thoughts into finished work. For a while, though, there was no real system behind any of it. We picked topics by instinct, published when we remembered, and had no idea whether the people asking ChatGPT "what's the best note app for ADHD" ever heard our name back.

That is the problem we set out to fix: show up when someone searches on Google, and show up when someone asks an AI assistant instead.

The two places people actually look

Search habits have split. Some readers still type a query into Google. Others open ChatGPT, Claude, or Gemini and ask a question directly, expecting a recommendation, not a list of links. A note-taking app for ADHD minds lives or dies on both. If we only optimised for one, we would be invisible to the other half of the people looking for exactly what we built.

We needed one system that handled both, not two separate projects running in parallel.

How we set it up

• Connected our blog's GitHub repository, the one that holds our markdown content files, to RankAscend.
• Let RankAscend crawl the existing 60-plus articles to learn our tone and map what we had already covered.
• Connected Google Search Console so the keyword queue would skip anything we already ranked for on page one.
• Picked our publish schedule and left most content types on manual review, with only our straightforward how-to posts set to auto-approve.

Closing the keyword gaps first

Once the crawl finished, RankAscend built a queue of keyword gaps: topics our audience searches for that we had not written about yet, each scored for opportunity. Checking that queue against our actual Search Console data mattered more than we expected going in. A chunk of our early topic ideas were things we already ranked for, and writing those again would have been a wasted week.

Publishing on a schedule instead of a whim

Every article in the queue now comes with a meta description, a heading structure, an FAQ section built from real reader questions, and JSON-LD schema markup that gets validated before anything is committed. We still review most drafts in the inbox before they go live. Every publish is a commit to our repository, so we can see exactly what shipped, when, and roll it back if something looks wrong.

Making sure the AI assistants notice too

The part we had never been able to measure before was the AI side. RankAscend runs a weekly check across ChatGPT, Claude, Gemini, and other models, asking the kinds of questions our readers actually type: "best note-taking app for ADHD," "distraction-free writing app," "how to capture thoughts quickly." It records which models mention BrainDump, where we land in the answer, and keeps a trend line over time.

Seeing that number move is a different kind of feedback than a rankings report. It is the closest thing we have to knowing whether the AI recommending apps to a stressed-out student has ever heard of us.

What changed

• Organic sessions to the blog: 37% over 4 months
• Keywords ranking in the top 10: 8 new keywords
• Articles published per month: from 1/2 (irregular) to 30 (consistent)
• Mention rate across monitored AI models: from 1% to 6%

None of this happened overnight, and a content queue does not rewire search rankings in a week. What changed is that we stopped guessing. We know what we have not covered, we know what is actually ranking, and for the first time we have a rough sense of whether an AI assistant would recommend us.

If you are running a blog like this

If you have a real backlog of content and no system behind it, the highest-leverage first step is not writing more. It is finding out what you have already covered, what you are already ranking for, and what a language model would say about you if someone asked right now. The writing is the easy part. Knowing where to point it is not.

We used RankAscend to handle the crawl, the keyword queue, and the AI mention tracking described above.