Promoting a new site from scratch: the first 90 days

The first three months of a new site rarely look like growth. Pages enter the index unevenly, impressions come in tens and clicks in ones, and it is easy to conclude that something was done wrong. Usually nothing was. A new site has no links, no history and no behavioural data, so search engines have nothing to go on yet. What follows is a week-by-week plan for the first 90 days: the technical groundwork before launch, getting indexing started, building the keyword core and the first content, finding the first links, and what numbers count as normal by the end of the quarter.

What to expect from the first three months

Start with what the search engines themselves say about timing. Yandex's help page on adding a site to search says that once the crawler has visited a site, its pages can appear in search within two weeks (yandex.ru/support/webmaster, checked 15 October 2026). The same page notes that the crawler can also find a site through links from other sites, but that this takes time and does not guarantee every page you want shown will be crawled.

Google's SEO Starter Guide (checked 15 October 2026) says some changes might take effect in a few hours while others could take several months, and suggests waiting a few weeks before judging whether your work had an effect. Its page on asking Google to recrawl says crawling can take anywhere from a few days to a few weeks.

Both engines are talking about getting into the index. Neither promises traffic within those timeframes. Indexing within the first month is realistic. Rankings for queries with real search volume usually are not, so the plan should aim to have the foundations in place by the end of the quarter: clean technical setup, indexed pages, a keyword core split into clusters, a handful of strong pages and the first links. Traffic arrives later, on top of that.

Neither Yandex nor Google documents a special filter for new sites. What people call the sandbox is easier to explain by the lack of signals: the engine has nothing yet to tell it that a new domain deserves more trust than ten competitors with years of history.

Weeks 1–2: technical groundwork before launch

Anything you can fix before the crawler's first visit is worth fixing before it. A mistake the crawler has already seen can take weeks to clear, because the fix only counts once it comes back.

• One version of the site. The site should answer at a single address, with https, and either with or without www. Every other variant returns a permanent 301 redirect to it.
• robots.txt and meta tags. Before launch, check that a Disallow: / or a template-wide noindex has not come across from the staging server. It is the most common reason a new site "won't index". The difference between blocking crawling and blocking indexing is covered in robots.txt vs noindex.
• Sitemap. sitemap.xml lists only canonical URLs that return 200, with no noindex pages and no redirects. Reference it from robots.txt.
• Status codes. A URL that does not exist should return 404, not the homepage with a 200.
• Canonicals. Every page carries a rel="canonical" pointing to itself or to the main version.
• Mobile and speed. Pages should work properly on a phone; the full order of checks is in the technical SEO audit checklist.
• Reachability. The server has to answer crawlers reliably from different networks.

We learned the last point on our own site. Until 10 October 2026 gbseo.ru was served from a single server with no CDN, and Google's crawler could not always reach it: on 4 October Search Console reported robots.txt as unreachable. The site has since moved to a CDN. If a new project lives on a cheap server in one country, look at the crawl stats report in Search Console at least once and make sure there are no connection errors.

In the same two weeks, add the site to Yandex Webmaster and Google Search Console and verify ownership. Yandex's help says plainly that adding a site to Webmaster does not by itself guarantee it will appear in search, but without it you cannot see what is happening. If the site runs a Yandex Metrica counter, link it in Webmaster: Yandex lists crawling by counter data among the ways its crawler learns about pages.

Weeks 2–4: start indexing and confirm it is happening

Yandex's help lists four ways to tell the crawler about pages: a sitemap, crawling pages that carry a Metrica counter, the Reindex pages tool in Webmaster and the IndexNow protocol. For a new site with a few dozen pages the first three are enough. In its IndexNow FAQ, Yandex itself points small sites to the Webmaster reindex tool and leaves the protocol for automated change notifications on larger projects.

The reindex tool works like this (Yandex help, checked 15 October 2026). URLs marked "In queue" are visited with priority on the crawler's next visit, which takes no more than three days. "Request processed" means the crawler has been to the page and the data will update within two weeks; it does not mean the page was indexed. There is a daily limit on the number of URLs, so send the important pages rather than everything.

In Google, you submit the sitemap in Search Console and individual pages through URL Inspection. Google's help (checked 15 October 2026) warns that there is a quota for submitting individual URLs and that requesting a recrawl of the same URL repeatedly will not get it crawled faster. Submitting the homepage and five to ten key pages makes sense; resubmitting the same page every day does not.

By the end of the first month, check that both engines know the site exists and have started crawling it. In Webmaster that means Pages in search and the crawl statistics; in Search Console, the Page indexing report. How to read those reports, and why a sitemap can show zero indexed pages for a long time, is covered in how to check whether your site is indexed. If after a month neither engine knows a single page, the cause is almost always technical, and it is back to the previous section.

Weeks 3–6: a keyword core for a site with no history

The core for a new site is built the same way as for an established one: seed terms, expansion through Wordstat, competitors, cleaning, search volume, intent tagging. The order and the tools are in building a semantic core. The difference is which queries from that list you work on first.

A new site has no chance on a results page where the top ten are sites with thousands of links and years of history. So the first wave goes to queries where the results are weaker. You can see that weakness by eye:

• the top ten includes forums, Q&A threads, small blogs and pages that only partly answer the query;
• the query is long and specific, with a condition or a scenario in it ("how to move a site to a new host without losing rankings" rather than "site migration");
• the pages in the top ten have few external links when you look them up in a backlink tool.

Each of these queries rarely brings hundreds of impressions a month, but there are a lot of them, and a new site can reach the first positions for some within the quarter. The high-volume queries stay in the core. The site structure is designed around them from the start, and work on ranking for them begins later, once the site has links and a history of its own.

The output of this stage is a core split into clusters and a map of which cluster leads to which page. Each page on the site answers one group of queries. If two clusters want the same page, or one cluster wants two, settle that before the pages are written.

Weeks 5–10: the first content and the order to publish it

Pick the first pieces from the core on two criteria: the query is reachable in the coming months, and you can make something better than what currently ranks. Publish first the pages the rest will link to: catalogue sections, service pages, broad overview articles. The narrower pieces follow, each linking to its section.

How much to publish depends on who is writing. One author who spends half their working time on content can realistically ship two or three strong articles a month; how to work out the load and the queue is covered in a quarterly content plan. For a new site that means six to nine articles by the end of the third month, plus the main commercial pages. Thirty pages written in a month for the sake of volume are worse than ten written properly.

On length, Google's Starter Guide is direct: there is no magical word count target, minimum or maximum (checked 15 October 2026). Yandex's help page on what separates a quality site from a poor one lists, among the signs of a problem site, rewriting information from other sites, low-quality machine translation and automatically generated content with no value to users. The same page says the algorithm's restrictions can affect the ranking of the whole site or its sections, so weak pages pull the strong ones down with them.

Add internal links when you publish, not afterwards. A new article links to its section and to two or three related pieces, and at least one already indexed page links to it. Yandex's help, in its section on why pages take a long time to appear, says exactly this: pages should be reachable through links from pages of the site that are already indexed.

Weeks 6–12: the first links and mentions

For a new site, the first links matter mainly for discovery: the crawler finds pages through them and comes back more often, and the search engine gets its first outside signals that someone knows the site. Google's Starter Guide calls word of mouth, where people who know your site tell others about it, one of the most effective and lasting ways to promote it.

Where the first links come from when there is no budget for buying them:

• your own presence: business directory and map listings, social profiles, pages on industry platforms where you already have an account;
• partners and suppliers that have a partners or clients page;
• professional communities: useful answers on forums and in discussions where a link belongs, rather than one dropped in for its own sake;
• your own data: a small study, a compilation or a calculator that people link to because it exists nowhere else;
• unlinked mentions: once someone has already written about the project, asking them to add a link usually costs nothing. How to find them is covered in working a backlink profile with no budget.

Bought links stand out on a new site, because there is no natural background for them to blend into. The first item under link spam in Google's spam policies is buying or selling links for ranking purposes (policies last updated 28 August 2026, checked 15 October 2026). Yandex's quality page counts, as a sign of a poor site, external links placed not as the author's recommendation to visit a resource but to deceive the search engine.

A realistic goal for the quarter is modest: ten or twenty links and mentions from sites that have something to do with your subject. That is enough for the crawler to visit more often, and to give you something to build on six months from now.

Yandex and Google: how the start looks different

Almost every new site goes through the same thing in its first weeks: some pages are already in one engine and not yet in the other. That is normal, and comparing the two engines on the first two weeks tells you nothing. The difference lies in what each one relies on.

Yandex gives you more levers for speeding things up: reindexing with a daily limit, IndexNow, crawling by Metrica counter. If the site serves a specific city or region, set that in Webmaster at launch. Google offers fewer ways to ask for a crawl and leans more on links: the Starter Guide says Google primarily finds pages through links from other pages it has already crawled. The broader differences between the two are covered in Yandex vs Google.

For reference, here is what our own blog shows. gbseo.ru is not a new domain, but its blog was relaunched in September 2026. On 15 October we checked eight September articles in Search Console. Six are indexed, one is "Discovered – currently not indexed", and Google does not know the last one at all, although it was published on 19 September and is in the sitemap. All six indexed articles show the same morning, 4 October, as their last crawl: Google picked them up in one batch, 15 days after the earliest was published and a week after the latest. Across the whole domain, the last 28 days brought 158 impressions and 2 clicks. Indexing is under way, traffic barely exists, and that is expected: nothing links to the blog yet.

What to measure at 30, 60 and 90 days

Traffic in the first quarter is a poor measure of the work, because there is too little of it to base decisions on. It is more useful to track what happens before traffic does.

If fewer than half the pages are indexed by day 60, stop publishing new ones and find out first why the existing ones are not getting in. Yandex's "Low-value or low-demand page" status and Google's "Crawled – currently not indexed" usually point to quality and structure rather than to anything technical.

Mistakes in the first three months

• Launching blocked. A staging Disallow: / or noindex survives into production and the first weeks are wasted.
• A hundred pages on day one. A catalogue of a thousand product pages with no descriptions, or fifty short posts published at once. The crawler works through them slowly, and Yandex's help ties the chance of appearing in search to how many pages useful to visitors the crawler finds. An empty product page is not one of them.
• Betting on high-volume queries. Content is written for queries a new site has no chance on for a year, while the reachable long-tail queries get no pages.
• Resubmitting every day. The same URLs go into reindexing and URL Inspection daily. Google's help says outright that this does not speed up crawling.
• Changing the structure in month two. URLs change because the new ones look nicer and the redirects are forgotten. Everything that was indexed starts over.
• Buying links in bulk. A new site has nothing to hide them behind.
• Judging by traffic after a month. The decision that "SEO doesn't work" gets made before there is anything to see.

Frequently asked questions

How long until a new site appears in search?

Yandex's help says that once the crawler has visited, pages can appear in search within two weeks. Google says crawling takes from a few days to a few weeks. Being in the index does not yet mean traffic.

Is there a sandbox for new sites?

Neither Yandex nor Google documents such a filter. A slow start is better explained by the lack of links, history and behavioural data.

Should every new page be submitted for recrawling?

On a small site it makes sense to submit the main new pages through reindexing in Webmaster and URL Inspection in Search Console. Resubmitting the same page achieves nothing, and both tools have limits.

How many articles should I publish in the first three months?

As many as you can write well. For one author working half time on content, that is six to nine pieces in the quarter, plus the site's main pages.

Which queries should a new site target first?

Long, specific queries where the results include forums, small blogs and pages that only partly answer the question. High-volume queries go into the site structure from the start but are worked on later.

When should I judge the results?

Indexing and the absence of technical errors can be judged after a month. Rankings for the first wave of queries make sense to discuss by the end of the third month, traffic later still.