Overview
Setup is seven screens and takes about three minutes — you're mostly confirming rather than typing. The moment you enter a domain, we read the site and draft the answers to the rest; your job is to correct what we got wrong. After setup, the Strategy Wizard row on your project rail reopens the same flow.
It's not a form. Each answer feeds the next screen, and the audit, the keyword map, the AI-search prompts, and the whole content plan are built from what you confirm here. Five minutes spent correcting it saves re-running everything.
Progress is saved as you go — leave and come back, and you'll resume where you stopped. Any step you've already reached is clickable in the left rail, so you can go back and change an answer before the analysis starts.
Step 1: Your website
The domain to analyse. No https:// needed.
Use the live public site. A staging URL, a page behind a login, or a site that blocks bots gives us nothing to read, so every pre-filled answer after this comes back thin or empty — and so does the audit. If the site can't be fetched, the next screen says so. Note that an unreachable site is itself an SEO finding — search engine crawlers hit the same wall.
Setup takes roughly three minutes from here. The analysis afterwards runs for another three to five.
Step 2: Your business
Three things, all drafted from your homepage while you were on step 1.
Target market. One to five countries. This selects the search database every number in your strategy comes from: search volumes, keyword difficulty, competitor rankings, all of it. Getting it wrong doesn't fail loudly — it gives you accurate numbers for the wrong country.
Pick where your customers search, not where your office is. This isn't a cosmetic setting: the same site measured against Singapore returned one usable quick win, and against the United States returned a 1,300/month term at difficulty 0. Each market is researched separately, so more markets means a longer analysis.
Business description. What you do, who you serve, and where. This field has the widest effect of any setup answer: it decides which keywords count as relevant, which competitors count as real, and what every generated article assumes about you. If the draft reads generic — "a digital marketing agency" — the keywords come back generic too. Say what you sell, not what you value.
Target audience. The types of customer you serve, as chips. Nothing else in setup captures who you sell to — the description usually only implies it. It steers keyword intent and the voice of everything written later.
If the whole read is off — you're B2B and we described a consumer business, or you stopped offering something years ago — use Improve this rather than editing four fields by hand. It sends your note back to the generator and re-drafts the description, the audiences, and the offerings together.
Step 3: What you sell
Two lists, side by side, both read off your homepage.
If your homepage can't be read (a firewall blocks us, or the page needs JavaScript to show anything), we read your other pages instead: your sitemap and your services, products and about pages. If those say nothing either, we use the searches Google itself links to your site. The screen says which one the suggestions came from, so check them before you continue.
What you sell seeds the keywords we research and the pages we plan. Add anything the read missed; select a chip to remove it.
What you don't sell matters just as much, and it's the half people skip. Every generator downstream treats it as a hard never-write rule. Without it, we plan content for adjacent services you don't take on — the cleaning company that starts ranking for pest control it has to turn away. Note that a term can't be in both lists; adding it to the exclusions removes it from the offerings.
Step 4: How this works
Nothing to fill in. Four cards explain what runs after setup: we research your market, map keywords to pages, plan the content, and you approve and publish.
The screen is also doing real work in the background. Competitor discovery starts the moment you leave step 3, and this screen is where it finishes — which is why the next step usually has its table ready by the time you arrive.
Step 5: Your competitors
We look at who actually ranks for the searches for what you picked in step 3, and for the searches your site already ranks for, then propose those, measured against your own site in one table: keywords, estimated traffic, and — when the Backlinks data is available — linking domains and authority. Your own row sits inside the ranking, so "where you stand" is a position in a list rather than a claim.
Expect surprises. The businesses you think of as rivals are often not the ones taking your search traffic.
The list is yours to edit, and leaving it alone is a valid answer. Remove anything that isn't a real competitor — directories, dictionaries, job boards, and news publishers are filtered automatically, but the filter isn't perfect. It matters because these sites are the corpus we mine for keyword ideas: one big out-of-industry domain left in the list pulls the whole keyword set sideways.
Every suggested competitor has to rank for enough keywords in your market to learn from: at least 30, or a quarter of your own count if that's higher (up to 200). A rival that ranks for fewer is still listed, unticked, with the reason. When too few strong rivals turn up, we look further on broader searches for the same offerings ("best ...", "... near me") before settling. This works for a brand-new site too: it needs what you sell, not rankings.
If your site looks like a different market from the one you picked in step 2, this screen offers to switch and re-run discovery.
If Google ranking data for your site isn't available during discovery, the competitors still come from what you sell, and the screen says so. If discovery fails or times out, you're not blocked: add competitors by hand, retry, or continue without any.
Step 6: Your quick wins
A fast first pass, before the paid analysis runs. We draft the terms your site could plausibly win, then check each one against real search volume and how contested it is. Only the terms that survive are shown, with:
- volume — searches per month in your market.
- KD — keyword difficulty, 0 to 100: how hard the current page-one results are to displace.
- Already ranking — you rank for this outside page one already, which is the fastest kind of win there is.
Terms with no measurable demand are dropped, and you'll see the count of what was dropped rather than having it hidden. If none survive, we say so plainly and show the real searches that are harder to win as longer-term targets instead. The full analysis searches far wider and will find the reachable ones.
Under the terms sit the first articles we'd write. Remove anything that doesn't fit; the full strategy follows your edits, at no extra cost.
Note that this is a preview, not your strategy. The real keyword set is chosen after the analysis, on Pick Keywords.
Step 7: Create your account
The only gate in the flow. Everything above ran without an account.
Enter an email and the account is created in place — no password, and for a new address, no inbox trip. If the address already has an account, we email a one-tap sign-in link instead, because a stranger typing your address must never get a session on your account. Either way, your answers are saved and the analysis starts as soon as you're in.
What happens after setup
Creating the account starts the analysis. From there the journey is Analyze, Pick Keywords, Pick AI Prompts, Keyword Map and Your Strategy → — five stages that turn what you confirmed here into the deliverables.
Troubleshooting
We couldn't read your site. Step 2 says so instead of drafting from nothing. Type a short description yourself and continue — the rest of setup works normally. If the block is a bot filter or a login wall, it's worth fixing: Google's crawler is hitting the same wall.
The description or the offerings are wrong. Fix them before pressing on, or use Improve this. Correcting them later means the keyword research already ran against the wrong business.
No competitors, or absurd ones. Usually the market in step 2 is wrong, or the description is too generic to search on.
The rail lists more steps than the counter. The counter counts the screens you fill in; the rail shows the whole journey, including the stages that run after the analysis. Both are correct.
You want to change an answer after the run. Business description, market, audiences, and keywords are all editable in the website's settings. Note that changing the market means the existing keyword data was pulled for the old one — re-run the analysis afterwards.