What a Brand-New Website's First 7 Weeks in Google Actually Look Like
Most articles about a new website's SEO are written by people selling SEO. That creates a predictable distortion: the timelines are optimistic, the case studies are anonymous, and the numbers arrive after the good part has already happened.
So here are mine, unedited, from a site I launched in July 2026. I run a small web studio and I had every incentive to wait until the graph looked good. I'm publishing it at week seven instead, because week seven is where most people quit.
The numbers
Search Console, 16 July to 1 September, web search only:
- 6,251 impressions
- 31 clicks
- Average position 75
- 233 distinct queries
- Domain authority at launch: 1. Backlinks: 13, all nofollow.
Read that top line again. Six thousand times, this site appeared in someone's search results. Thirty-one times, somebody clicked. A 0.5% click-through rate.
That is not a failure. It's what position 75 looks like. Position 75 is page eight. Nobody goes to page eight. The impressions are real — Google genuinely served the page — but they're served to the small number of people who scroll past seventy other results, plus a lot of automated tools.
The shape of it, week by week
| Week | Clicks | Impressions |
|---|---|---|
| 27 Jul | 9 | 671 |
| 3 Aug | 14 | 713 |
| 10 Aug | 3 | 1,616 |
| 17 Aug | 2 | 1,430 |
| 24 Aug | 1 | 1,151 |
Impressions more than doubled in mid-August, when a batch of new pages got indexed. Clicks went the opposite way: 14, then 3, then 2, then 1.
If you only track clicks, mid-August looks like a catastrophe. If you only track impressions, it looks like a breakthrough. Neither is true. What actually happened is that Google indexed several new pages, tested them deep in the results, and hasn't decided about them yet. The early clicks came from a handful of low-competition queries; the new impressions come from higher-value queries where the site ranks eightieth.
This is the single most useful thing I learned from watching the data weekly instead of monthly: impressions tell you whether Google understands your pages. Position tells you whether it trusts them. Those are two different problems with two different fixes, and most people conflate them into one number called “traffic.”
Google understood the pages immediately
Within days of publishing, the site was appearing for queries that map precisely to what each page is about — maintenance pricing queries on the maintenance page, e-commerce build-cost queries on the shop page, and so on. Two hundred and thirty-three distinct queries in seven weeks, from a site with no links and no history.
That surprised me. The received wisdom is that Google takes months to work out what a new site is about. In my case it took days. What takes months — what is still taking months — is Google deciding the site deserves to be seen.
There's a practical consequence. If your new site is getting impressions on relevant queries, writing more pages will not fix your ranking. You've already demonstrated the thing that content fixes. Ten more pages get you ten more entries at position 80. The constraint has moved somewhere else.
Where the constraint actually is
Authority. Which in practice means: other websites linking to yours, and — if you're a local business — reviews and a complete business profile.
I don't love this answer, because it's the answer that can't be solved in an afternoon and can't be bought safely. Izood has written about guest post marketplaces and what Google does about them; I'd add only that I looked at a direct competitor with 466,000 backlinks, almost all from an obvious private blog network, and it ranks for nothing that matters in my market. The authority is nominal. Google isn't fooled, it just isn't in a hurry to say so.
What I'm doing instead is unglamorous and slow: credits in the footers of sites I've built, listings in regional business directories, a couple of local sponsorships, articles like this one where an editor decides whether the piece is worth publishing. Fifteen honest links over two months, not fifteen hundred.
Two smaller findings, both useful
Local search volume is much smaller than it appears. Google's Keyword Planner reports roughly 500 monthly searches for “website laten maken enschede” (a Dutch city term) nationally, but only around 50 from people physically in that region. Most people searching for a service “in Enschede” are not in Enschede. That destroys the case for local paid advertising — I ran the forecast and Google would accept a maximum of about €4 per day, projecting seven clicks a month at an implied cost per acquisition near €960 — while leaving the case for local SEO intact, because organic reaches all of them for free.
Nobody is searching for AI visibility yet. Half my industry is now selling “get found in ChatGPT.” Google's own data puts the Dutch query for it at zero searches per month. The equivalent conventional query — “rank better on Google” — sits at 500. My conclusion: build for AI retrieval now, because it will matter, but don't sell it as today's problem using vocabulary nobody types.
What I'd tell someone at week seven
Stop refreshing the clicks number. Open the queries report instead and ask one question: is Google showing my pages for things I actually do? If yes, your content is fine and your problem is trust, which is earned in months, one link and one review at a time. If no, your problem is the pages, and no amount of link building will rescue them.
Then set a date far enough out to be meaningful. Mine is 1 December — six months after launch, three months of link work. I'll publish that one too, whatever it says. If average position is still in the seventies, I'll have been wrong about something, and that's worth reading as well.
Reza Siavashiasl runs Studio 053, a one-person web studio in Enschede, the Netherlands, building websites, online shops and business photography for regional companies. He publishes his own search data monthly.