How long does SEO take: four clocks, not one deadline

This blog went live on 9 August 2026. Eight days later, 42 of its 50 URLs were indexed. In its first partial month it recorded 1,202 impressions and 3 clicks. Every one of those numbers is a result, they arrived on different dates, and the answer the whole web gives to this question collapses all of them into one range of months that nobody can trace to a measurement.

How long does SEO take to work?

Google documents no fixed timeline: some changes take hours, others months. On a new domain the first signals are indexing within days and impressions within weeks, while clicks lag by months; a site with existing authority moves in weeks; a consolidated one may wait for the next core update. Nobody can promise a date.

Quick answer

• Indexing is the fast clock. Google's own documentation puts crawling at "a few days to a few weeks", and on this domain 42 of 50 URLs were indexed inside eight days.
• Impressions come next, and they flatter. 92.3 % of this site's August impressions came from beyond position 30, which is page four and further.
• Clicks are the slow clock. Three of them, in the month that produced 1,202 impressions.
• No date is promisable. Google's guidance for hiring an SEO says plainly that no one can guarantee a first position.

Why this question has no single answer

The model behind the question is that "results" is one event with one date. It is four events, and they are separated by months. A site can be doing well on the first two clocks and look like a total failure on the third, which is exactly what a first quarter looks like from the client side of the table.

The confusion has a commercial cost. A retainer signed in January is judged in March against a metric that structurally cannot have moved yet, and the provider who explained the sequence in advance keeps the contract while the one who promised a range loses it. The economics of that sequence, and why serious engagements ask for six months, are in SEO cost for a small business.

There is a second reason the single answer fails. The word "SEO" covers removing a noindex tag, which is visible in days, and rebuilding a site's architecture, which is visible in quarters. Those are the two ends of what SEO actually is, and averaging their timelines produces a number that describes neither.

Four clocks, measured on this blog's own Search Console

This site is a live experiment, which is unusual for an article on this subject: almost everything published about SEO timelines is written by people describing other people's sites. The domain went live on 9 August 2026, publishes three articles a day, and has Search Console attached from the first deployment. Here is what each clock actually did.

Clock one: getting crawled and indexed

Google's documentation on requesting a recrawl says "crawling can take anywhere from a few days to a few weeks", and adds that "requesting a recrawl multiple times for the same URL won't get it crawled any faster". Both sentences were on the page on 5 September 2026.

The measurement matched. Of the 50 URLs inspected through the index inspection API on 17 August, 14 had been crawled on 10 August, the day after launch, and 42 were indexed. Seven had no crawl recorded at all. So the honest reading is not "indexing takes eight days": it is that most of a small new site enters quickly and a tail of it sits unvisited for reasons the report will not explain. The diagnosis for that tail is a separate job, covered in why is my page not indexed.

This is the only clock a new site can influence directly, and the influence is smaller than the guides suggest. A sitemap and internal links help discovery. Nothing accelerates the queue.

Clock two: the first impressions

August 2026, counting from the 9th, produced 1,202 impressions across the site. That sounds like the channel starting. Read one column further and it is something quieter: the median position of the pages producing them was 79.8, and 92.3 % of those impressions came from beyond position 30.

Position 79 is page eight. Those impressions are real, they are recorded, and almost nobody scrolls that far. What they prove is narrow and still worth having: Google has decided which queries this site is about. Which report shows that, and how to read it without flattering yourself, is in how to set up Search Console.

Impressions are the earliest honest signal available and the one most reports bury. They are also the easiest to misrepresent, because a chart of impressions climbing looks identical whether the pages sit at position 8 or position 80.

Clock three: the first clicks

Three clicks. A 0.25 % click-through rate, in the same window that produced 1,202 impressions.

The first three days of September produced another 527 impressions and zero clicks, at roughly three times August's daily rate. Impressions accelerating while clicks stay flat is not a malfunction: it is the normal shape of the gap between appearing and being chosen, and it is the stretch where most contracts die. September is a partial month and is not extrapolated here.

Anyone comparing this against a paid channel is comparing two different physics. An ad account spends money on Monday and reports clicks on Monday, which is the whole argument in SEO versus Google Ads.

Clock four: the first customer

This clock has no reading yet, and saying so is the point. Three clicks a month cannot produce an attributable client at any conversion rate anyone has ever published, so there is nothing to report and no honest way to model it forward.

That is the clock the business actually cares about, it is the last to move, and it is the one every timeline article quietly substitutes for one of the earlier three. A provider showing a rising impressions chart in month four is answering a question about clock two while the client is asking about clock four.

What the published studies really counted

The most useful third-party measurement on this subject is an Ahrefs study published on 15 May 2025, and it does not say three to six months. It took 1 million random URLs its crawler first saw in September 2023 and checked how many reached the top ten within a year. 1.74 % did. Filtered to pages with non-empty English content, on a separate sample of 2 million URLs created in October 2023, the figure rises to 6.11 %.

Two numbers from one study, and the denominator is the whole argument. Neither supports a timeline promise, because both describe an outcome most pages never reach.

The finding that changes the shape of the answer is the next one: of the pages that did reach the top ten, 40.82 % got there within one month. The distribution is not a slow climb toward an average. Most pages never arrive, and a large share of the ones that do arrive fast. A range of months describes the middle of a distribution that has almost nothing in its middle.

The same study measured age from the other direction, across 1.3 million random US keywords: 72.9 % of pages in Google's top ten are more than three years old, and the average first-position page is five years old. In 2017 that average was two. The competitive floor rose while the industry kept repeating the same range.

The timeline everyone attributes to Google

Search the query in English and the same sentence appears on most of the first page: Google says SEO needs four months to a year. It has a real source, and the source is older than most of the people quoting it.

It comes from Maile Ohye, then at Google, in a video on how to hire an SEO, reported on 16 February

1. She left the company weeks later. The line was accurate about 2017 and it has been recycled in the present tense for nine years, through two index rebuilds and the arrival of generated answers above the results.

Google's current documentation does not contain it. Checked on 5 September 2026, the page on whether you need an SEO carries no timeline at all, only the warning that "no one can guarantee a #1 ranking on Google". The starter guide says something narrower and more useful: "some changes might take effect in a few hours, others could take several months", and "in general, you likely want to wait a few weeks to assess whether your work had beneficial effects".

That second sentence is a measurement instruction, not a forecast. It says how long to wait before reading the report. It does not say when the report will look good, and the difference between those two statements is the entire disagreement of month five.

What moves the date, and what does not

Four variables account for most of the spread between one site and another, and only two of them respond to money.

Prior authority. A domain with years of history is crawled often, so its changes are evaluated sooner. A new one waits, and that wait is the constraint behind everything in SEO for startups.

The difficulty of the queries chosen. A site's timeline is the weighted average of the timelines of the terms it went after, which makes selection the cheapest accelerator available. Picking reachable terms before writing is the work described in free keyword research.

Technical debt. A site that cannot be crawled completely has not started the clock at all, and its first months are repair rather than positioning.

Direction of travel. Clocks run backwards too, and faster. On a site of mine the median position of its pages roughly tripled across six months of 2026, which is less time than earning those positions took; the measurement, and the three times my own tooling handed me the comfortable answer, are in why organic traffic drops.

Budget buys parallelism, not speed. Doubling the retainer doubles the pages reviewed in a month and changes neither the crawl frequency nor the reevaluation cycle.

When a timeline is the wrong thing to negotiate

When the business needs revenue this quarter. Search does not arrive in time, and pressing for a date produces exactly the promises that Google's own guidance warns about.

When nobody searches for what is being sold. A timeline for capturing demand that does not exist is a schedule for nothing, and no budget creates the queries.

When the site does not convert what it already receives. Adding traffic to a page that does not sell is paying to confirm a problem that predates the contract.

When the fall coincides with a core update. Google confirmed four updates during 2026, with rollouts running from under a day to three weeks, and diagnosing inside a rollout produces conclusions that expire before the rollout does.

Mistakes that repeat

• Judging the channel in the first quarter. That is the stretch where the spend has happened and the third clock has not moved.
• Reading impressions without reading position. 1,202 impressions at a median position of 79.8 is not the same event as 1,202 impressions at position 8.
• Presenting an accumulated total as a monthly rate. It inflates the result and destroys the next report.
• Requesting indexing repeatedly. Google states it does not make the crawl any faster.
• Quoting a 2017 figure in the present tense. The four-months-to-a-year line is not in Google's current documentation.
• Changing provider in month four. The replacement restarts the diagnosis, which resets the only clock that was actually running.

Which of those a given agency is committing is checkable from outside, and the five signals that separate slow work from no work are in is my SEO agency working.

Data and transparency

That crawling can take anywhere from a few days to a few weeks, and that requesting a recrawl repeatedly does not speed it up, comes from Google's page on asking it to recrawl your URLs, opened on 5 September 2026. That some changes take effect in a few hours and others several months, and that you likely want to wait a few weeks before assessing the effect, comes from Google's SEO starter guide, opened the same day. That no one can guarantee a first position comes from Google's page on whether you need an SEO, also opened that day, which carries no timeline of any kind.

The four-months-to-a-year figure is attributed to Maile Ohye, then at Google, in a video on hiring an SEO reported on 16 February 2017. It is cited here as an example of a number that aged in public, not as current guidance: it does not appear in the documentation linked above.

The ranking-timeline percentages come from Ahrefs, published 15 May 2025: 1.74 % of 1 million random URLs first seen in September 2023 reached the top ten within a year, 6.11 % on a filtered sample of 2 million URLs created in October 2023, 40.82 % of the pages that did reach the top ten arrived within a month, and the age figures come from 1.3 million random US keywords.

The site figures are for alexgonzalezgo.com, which is mine and therefore named with exact numbers. Indexation is from the index inspection API on 17 August 2026, over the 50 URLs of the sitemap. The monthly series is from the Search Console API, consulted on 5 September 2026 over a window closing on 3 September: August 2026 recorded 1,202 impressions and 3 clicks with a median page position of 79.8, and 1 to 3 September recorded 527 impressions and no clicks. September is incomplete and is not extrapolated. This series begins after 27 April 2026, so it falls outside the window in which, according to Google's data anomalies page, a logging error affected impressions, CTR and average position; clicks were never affected. One site is not a sample, and this curve is offered as a shape, not as a magnitude. Across the sites I operate and audit, Search Console records more than 300 million impressions a year. Verified as of September 2026.

What this changes

The useful replacement for this question is not a better estimate. It is an agreement about which clock is being read and when.

Named in advance, a bad month three stops being a crisis and becomes a checkpoint that behaved as predicted: impressions moving, clicks flat, indexing complete, revenue absent. Most relationships in this channel do not break because the work failed. They break because two people were watching different clocks and only one of them knew it.

Frequently asked questions