Docs · AI

Finding a niche

The niche you choose matters more than anything else about a new blog. The Niche Finder takes you from "I'm interested in coffee" to a specific sub-niche a new blog can win, with its articles planned from real searches, ready for AI Autopilot to build.

1. Tell it what to research

Open AI Autopilot → Help me choose a niche and fill in a short form:

Niches you'd like to research: 1 to 5, broad is fine ("home coffee", "dog training"). Or switch on Surprise me and pick a few broad areas.
Time you have each week: AI writes the articles; your time goes on checking them and adding your own touches.
How you'd like to earn: ads, affiliate products or both. This sets the mix of articles: "best…" and review articles earn affiliate commission, how-to articles bring the traffic ads need.
Country and language: who the blog is for. Searches, spelling and affiliate programmes differ.
Topics to avoid, if any.

The Niche Finder form: niches to research (budget travel, hiking), Surprise me, time each week, how to earn, country and language, and topics to avoid — The brief: a minute to fill in, and the cost is shown before you start.

2. Pick a niche

You get up to 10 niche ideas, best first, each with a scorecard:

Line	What it tells you
Demand	Do enough people search for this?
Competition	Can a new blog realistically rank?
Earning	How valuable the readers are to advertisers, and products to recommend
Room to grow	Enough topics for 100+ articles?
All year round	Steady interest, or seasonal?
Beginner-safe	Free of health, money and legal advice that needs credentials Google trusts
Fit	Suits your time and earning goal

Green is good, amber is OK, red is weak, each with a sentence saying why. Tick Compare on two or three to see them side by side. The scores are estimates from the AI's knowledge of each niche.

Scored niche ideas: Interrail and European rail travel recommended at 81, Budget hiking holidays in the UK at 77, Weekend city breaks by train at 72 (saved for later), each with seven scored lines, a reason, and Choose this niche and Save for later buttons — Niche ideas, best first, each with a scorecard and a plain reason. Save for later keeps one for growing into later.

3. Narrow it to a sub-niche

Broad niches are hard for new blogs: "dog training" competes with huge sites, "training rescue dogs" can be won. Choosing a niche shows 5 to 8 sub-niches, scored the same way, with the recommended one first. Research its keywords when one feels right, or Go deeper if it still feels broad.

Sub-niches of Interrail and European rail travel: Interrail for first-timers recommended at 84, then Night trains in Europe (saved for later) and Interrail on a budget — Sub-niches, scored the same way. Research one, go deeper, or save it for later.

4. Keyword research

fastCMS collects what people really type into Google (or Bing) about your sub-niche: it asks the search box's suggestions with question words ("how to…", "can…", "best…"), modifiers and every letter of the alphabet. That usually turns up a few hundred real searches. It's free and takes a minute or two.

The AI then groups searches one article should answer together, so no two of your articles compete for the same search, and marks each article as:

Pillar: the big, complete guide for a category, which the other articles link to.
Money or Traffic: money articles ("best…", reviews, comparisons) can earn affiliate commission; traffic articles (how-to, questions) bring readers.
Easy, Medium or Hard: how hard it is to rank. Specific questions are usually easiest for a new blog.

5. Your niche report

The report explains why the sub-niche is a good choice, who your readers are, how the blog can earn (with affiliate ideas), what to expect over the first months, things to watch out for, and your first steps. Below it is the whole article plan in publishing order: quick wins first, then each category's pillar guide, then the rest, with money articles spread through.

The niche report for Interrail for first-timers: summary and why, scorecard, readers and what they search for, the Build this blog card, the niches saved for later and what to expect — The niche report, with Build my blog and your saved niches on the right.

The article plan in publishing order: 45 articles, 6 pillar guides and 13 money articles from 406 searches, each with its main search, category and Traffic or Money, Easy to Hard badges — The article plan: quick wins first, then the pillar guides.

Save niches for later

Often two or three of the ideas would complement each other: night trains and budget Interrail alongside Interrail for first-timers, say. Click Save for later on any niche idea or sub-niche card (it turns into Saved; click again to un-save). Saved niches are listed on your report.

Nothing is paid for until you use one: once your blog is built, open AI Autopilot → Grow this site → Research new keywords first and pick it. Its keyword research runs then, so the searches are fresh, and the new articles fit into your existing categories.

6. Build my blog

Choose how many articles to build first and click Build my blog. You land on AI Autopilot's review screen with everything filled in: check the site description and image style, untick anything you don't want, choose images and Co-pilot or Autopilot, then build. Each article is written for its real main search and also answers the related searches grouped with it.

If your site already has articles, choose Start a new site with it (clears them, with a restore point) or Add these articles to this site.

The article ideas you didn't build aren't wasted: Grow this site uses them first when you plan more articles later.

Costs

The research uses your planning AI model (Settings → AI): about the same as planning a blog, shown before you start, and counted in your monthly AI budget. Collecting search suggestions is free.

If something goes wrong

"This step didn't finish": usually a busy AI provider. Click Try again; your earlier choices are kept.
"Search suggestions weren't available": Google and Bing didn't answer, so the AI suggested the searches itself. The plan still works; the searches just aren't confirmed.