SEO for a new website is the work that takes a domain with no history, no links and no Search Console data into Google's index and onto the first searches. You start on an uphill slope: in SE Ranking's study of 100,000 US results pages, collected in April 2026, domains under two years old made up only 2.03% of those in the top 10.
Yet the same study concludes that domain age is not a direct ranking factor.
A new site isn't penalised for being new.
It's weak because it doesn't yet have links, content with a track record or a brand that people search for by name. Three things you build one page and one link at a time.
The trouble is that almost every new site starts out getting the same things wrong:
- the development noindex left switched on after launch, and Google seeing nothing for weeks;
- the "services" page with everything in it, which doesn't answer any search well;
- the first report checked on the category keyword, the one a new site reaches last.
This article starts from there. And from a test we hadn't seen anyone run.
At Sunscape, which develops solar farms and looks for land across Italy, a new structure with one page per region, started in March 2026, took keywords on page one from 1 to 38 out of 143 in six months, and organic impressions from 329 to 5,809 a month.
At Macropix, an LED screen manufacturer, getting from position 88 to position 2 for "monitor pubblicitario" (advertising monitor) took five years.
Six months against five years?
It depends on which search you choose to win first. That's the core of what you'll find here:
- the decisions to make before launch, while fixing things is cheap;
- the six checks on launch day, in one table;
- our test on 12 UK results pages: which searches leave room for a domain with no links;
- what to measure while clicks are still zero, month by month.
Let's start.
Is there a sandbox for new sites? (No, and the numbers explain why it feels like there is)
The sandbox is the idea, around for twenty years, that Google holds back newly launched sites for a trial period. Google denies it: in a 2021 video reported by Search Engine Journal, John Mueller said the idea of Google "keeping things back to prevent new pages from showing up" is "not the case".
With new content Google has little data and has to make assumptions. Then the page settles: sometimes a bit lower, sometimes a bit higher.
So why do new sites struggle so much?
Because the disadvantage is real, it just isn't a punishment.
Semrush followed 28,000 new domains in the United States from July 2021 to July 2022: of those that stayed in the top 100 for the whole period, about 41% were in the top 10 after six months, but only 4.2% held at least one keyword in the top 10 for all 13 months.
Ahrefs, in the version of its study updated in 2025, found that only 1.74% of newly published pages reach the top 10 within a year.
All three studies look at US keywords (we haven't found a UK equivalent). The UK figure that matters is about who's already in the field: according to the government's UK Business Data Survey 2024 (3,911 businesses interviewed between October 2023 and February 2024), 93% of small businesses, 97% of medium-sized ones and 99% of large ones already have a website. Almost every competitor with staff has been online for years, with years of links behind it.
We've laid out the timings stage by stage, from indexing to enquiries, in our guide to how long SEO takes to show results. Here we're interested in what to do during those months.
What to decide before launch (the choices that cost redirects later)
Before launch you decide the things that can later only be fixed with redirects and changing URLs: domain, site structure, keyword-to-page map and Search Console. Moving them afterwards means scattering your history exactly when it's thinnest.
- The domain. Short, easy to say, consistent with the brand. How much a keyword in the name really counts, we measured in our article on domains and SEO.
- The structure. One page for each service or category that matches a distinct search. At Sunscape the choice was one page per region: in six months, keywords on page one went from 1 to 38. What Google looks at in a site's structure is in our guide to website architecture.
- The keyword-to-page map. Each page gets its main search before anyone writes the copy. Two pages on the same search, on a domain with no authority, split what little Google gives.
- Search Console before launch. According to the Search Console help, data collection for a property starts as soon as anyone adds it to their account, even before verification. While you're at it, register the site on Bing Webmaster Tools.
Now, the mistake we see most often...
The development site closed to search engines with a noindex, or with the WordPress option "Discourage search engines from indexing this site", and the block going live untouched.
The site is online. The owner sends it to every customer. And for Google it does NOT exist.
What to check on launch day (six checks in one table)
Launch day needs six checks: robots.txt, meta robots, sitemap, canonical, redirects and the live test in Search Console. Ten minutes that save weeks of invisibility.
| Check | Where | What you should see |
|---|---|---|
| robots.txt file | yourdomain.co.uk/robots.txt | No Disallow: / left over from staging (the test environment) |
| Meta robots | Source code of the home page and service pages | No noindex |
| XML sitemap | Search Console, Sitemaps report | Submitted and read without errors, final URLs only |
| Canonical | Source code | Points to the https version on the final domain |
| Redirects | Browser | http, https, with and without www all lead to one version |
| URL Inspection | Search Console | Live test passed for the home page and service pages |
How to write the file is in our guide to robots.txt; what search engines actually use from a site map, in our article on the sitemap.
Once the checks pass, request indexing of the main pages with the URL Inspection tool. Once.
Why once?
Because Google says so: crawling can take anywhere from a few days to a few weeks, and requesting a recrawl of the same URL several times won't get it crawled any faster. Clicking the same button every morning only uses up your quota.
About to launch a new website?
The noindex left over from development is the mistake we find most often in first checks. Let's go through the six launch checks and the keyword-to-page map together before the site goes live.
Book a strategy consultationWhich searches can a domain with no links win? (Our test on 12 UK results pages)
A domain with no links can mostly win long-tail searches, where page one leaves room for weak sites. In our test on 12 UK results pages, results from domains with fewer than 100 linking domains made up 15.7% on long-tail searches and 9.6% on generic ones.
"Start with the long tail" is advice you read everywhere. Almost always without any proof.
So we measured it. On 10 October 2026 we pulled Google UK's organic results on mobile with DataForSEO for six pairs of searches: one generic and one long-tail on the same topic. For each domain in the results we counted the domains linking to it (referring domains, the closest measure of a site's strength you can take from the outside). We had run the same test on Google Italy three days earlier, with similar results.
| Search | Organic results analysed | From domains with fewer than 100 linking domains | Social, UGC or forums |
|---|---|---|---|
| business management software | 9 | 0 | 0 |
| garage management software for mechanics | 8 | 6 | 0 |
| solar panels | 9 | 0 | 0 |
| solar panels on asbestos roof | 8 | 0 | 3 |
| payroll services | 9 | 1 | 1 |
| payroll service for nannies | 8 | 0 | 1 |
| van hire | 9 | 2 | 0 |
| van hire with driver for house removals | 10 | 0 | 1 |
| 3d printing | 8 | 1 | 1 |
| 3d printing classic car parts | 9 | 1 | 4 |
| spanish course | 8 | 1 | 0 |
| spanish course for adult beginners online | 8 | 1 | 0 |
| Group | Results | From domains with fewer than 100 linking domains | Social, UGC or forums | Median linking domains |
|---|---|---|---|---|
| 6 generic searches | 52 | 5 (9.6%) | 2 (3.8%) | 1,496.5 |
| 6 long-tail searches | 51 | 8 (15.7%) | 9 (17.6%) | 918 |
Source: Visilay test on Google UK (mobile) with DataForSEO, 10 October 2026. Referring domains of the root domain. Only organic results count: AI Overviews, local packs and video carousels are left out, which is why some results pages have fewer than ten results. UGC stands for user-generated content, material published by users. The median excludes YouTube, Facebook and Instagram.
On the long tail results from weak domains go up by more than half: from 9.6% to 15.7%.
And social, UGC and forum content goes from 3.8% to 17.6%.
What does that look like in names?
For "garage management software for mechanics", page one has motmanager.co.uk with 77 linking domains, mechmate.co.uk with 52 and autosage.co.uk with 29.
For "3d printing classic car parts", the first result is 3dcarparts.org, with 77.
For "solar panels on asbestos roof", Google shows a Reddit thread, two Facebook posts and an Australian blog post from 2012. A sign that nobody has yet written a decent answer to that question...
...while on "solar panels" there isn't a single result from a domain under 100 links, and page one includes Which? and the Federation of Master Builders.
The test has its limits too, and they need saying. Six of the eight results from weak domains come from a single pair, the garage software one. The median of linking domains drops (1,496.5 against 918), but strong sites are there on the long tail too; what changes is the room left between them. And two pairs go the other way: "payroll service for nannies" and "van hire with driver for house removals" have no results from weak domains, while "van hire" has two.
Twelve searches aren't the market. They show you what to look at on a results page before building a page for it, not where to set a threshold.
The method for finding them in your sector, and why the keyword difficulty in tools is often missing or wrong on small volumes, is in our guide to easy keywords to rank for.
Which pages to publish first (services first, then questions)
A new site should publish its service or category pages first, because they bring in enquiries, and then the articles that answer long-tail questions. The number of pages matters less than whether they get indexed: in Ahrefs' analysis of around 14 billion pages, 96.55% get no traffic from Google.
Publishing a lot, in other words, guarantees nothing.
Each article should link to the service page it relates to: that internal link tells Google which of the two pages is the one that sells. If you're on WordPress, the technical steps are in our guide to SEO for WordPress.
In the first weeks the Page indexing report shows two statuses that look like synonyms. They are NOT.
- "Discovered - currently not indexed": Google knows the URL but has postponed the crawl, usually so as not to overload the server. Look at the server.
- "Crawled - currently not indexed": Google has read the page and for now decided not to store it, and says there's no need to resubmit it. Look at the content.
On a newly launched site, check the thinnest pages first: tags, archives, the three-line "about us" page. The difference between crawling and indexing is explained in our guide to indexing on Google.
How to get your first links (without buying them)
A new site's first links come from people who already know the business: suppliers, trade associations, customers, local press. Without links a domain stays on the margins even on the long tail, and how much they weigh today is in our guide to link building.
- Suppliers and manufacturers whose products you resell or install: they often have a partner or "where to buy" page.
- Trade associations, industry bodies and trade fairs you take part in.
- Customers you've completed a project with that can be written up.
- Local or trade press, when you have a real figure or piece of news to share.
- Your Google Business Profile, if you work in a local area: it isn't a backlink, but it's the first signal that the business exists. How to manage it is in our guide to local SEO.
What about an expired domain with lots of links, bought to get a head start?
Stop there. Google's spam policies call it expired domain abuse: an expired domain bought and repurposed mainly to manipulate rankings, with content that gives readers little or nothing. It's the "shortcut" that sends you back to below zero.
What to measure while clicks are still zero (a table for the first 12 months)
In a new site's first months you measure indexed pages, queries with impressions and positions on the long tail, not clicks. The data arrives early: according to the Search Console help it's normally available in 2-3 days, and for a newly created site it can take up to a week.
Clicks come last. At Sunscape, in the six months when impressions went from 329 to 5,809, monthly clicks rose from 46 to 195: the impressions curve moves long before the clicks curve.
| Period | What to look at | Where | Sign that something's wrong |
|---|---|---|---|
| Weeks 1-4 | Important pages indexed out of the total | Page indexing | Service pages still missing after a month |
| Months 1-3 | Number of different queries with at least one impression | Performance, Queries tab | No impressions on the searches in the keyword map |
| Months 3-6 | Position of the long-tail searches you assigned | Performance, query filter | Positions stuck beyond 50 on weak results pages |
| Months 6-12 | Clicks and enquiries from service pages | Search Console, GA4, CRM | Impressions growing and CTR flat: review titles and snippets |
The periods are working guidelines we use on our projects, not Google thresholds.
How to read each report is explained in our guide to Google Search Console.
When to add Google Ads (and what it tells you that SEO can't)
Google Ads is useful to a new site when the business needs enquiries straight away, because for months organic search won't bring in leads. A campaign on the same searches as the keyword map covers that period and gives you a figure SEO can't give at the start: which searches actually bring in quote requests.
The searches that convert in Ads become the priorities for the organic pages. Not the other way round.
If you'd like to set it up with us, start from our Google Ads service page; the differences and costs of the two routes are in our article on SEO and SEM. The other typical mistakes of the first months, from pages duplicated on http and https to copy lifted from manufacturers' product sheets, are in our guide to SEO mistakes.
Want to know which searches your new site can win?
In our test on 12 UK results pages, the long tail left room for domains with 29 linking domains. We'll run the same analysis on the searches in your sector and build the map of the first pages to publish.
Book a strategy consultationFrequently asked questions about SEO for a new website
It varies: according to Google's documentation, crawling can take anywhere from a few days to a few weeks. Submit the sitemap in Search Console and request indexing of the main pages with the URL Inspection tool once, because requesting a recrawl of the same URL several times doesn't speed it up.
No. Google's John Mueller explained in 2021 that Google doesn't hold back new pages to stop them showing up: at first it has little data and has to make assumptions, which is why positions move around in the first months before they settle.
Not if the aim is to exploit its old links. Google's spam policies treat an expired domain bought and repurposed mainly to manipulate rankings as abuse. It only makes sense when the name fits your brand and what the site will publish.
Long-tail searches with a weak first page: forums, social posts, sites with few links. In our test on 12 Google UK results pages on 10 October 2026, results from domains with fewer than 100 linking domains made up 15.7% on long-tail searches and 9.6% on generic ones.
It depends: it's worth having if your customers' questions have searches of their own on Google. Service pages come first, because they bring in enquiries; articles pick up long-tail questions and send internal links to those pages.
In short, before you launch
Search Console added before launch.
No noindex left over from development.
One page for each distinct search.
The first pages on searches where the results page is weak.
Impressions and queries measured before clicks.
And one thing the industry rarely says: the first figure an owner checks after launch is almost always the position for the category search they've had in mind for years, "solar panels" or "business management software". It's the search a new site reaches last, and looking at it first is the quickest way to conclude that SEO doesn't work. The searches that left room in our test for domains with 29 to 77 linking domains are the ones the first enquiries come from. And the first links, which you then need to climb on the others.
If you want to start on the right foot, see how we work on SEO and get in touch.
Until next time!
