Home › SEO Guide › What SEO does and how long it takes to show
What SEO does and how long it takes to show
This conversation usually comes up around the third month, phrased politely: “so, roughly when should we be seeing something?” It is a fair question, and the answer “it depends” helps no one. This chapter tries to give an answer you can verify.
It is not one job, it is three
What is sold under a single name is in fact three different activities, which produce results at very different speeds. The confusion between them explains most of the complaints.
| What is done | When it shows | What changes |
|---|---|---|
| Technical repairs indexing, speed, URLs, errors | 2 to 8 weeks | Pages that were not being found start to be found. The effect is often visible, because you start from something broken |
| Content pages that answer real questions | 2 to 6 months | Each new page starts to bring in a little, then it adds up. It does not climb suddenly, it accumulates |
| Authority what others say about you | 6 months and beyond | The part that cannot be rushed with money and that decides the rankings on contested terms |
In the first month the work is almost entirely on the first category, because it is the one that blocks the rest. A site that Google cannot read properly has nothing to gain from new content.
The difference between repair and growth: repairs give you back what you already had and were losing without knowing it. Growth brings something you never had.
The first months usually look good precisely because they are repairs. That does not mean the pace holds, and it is only fair to say so from the start.
A real case, with what we know and what we do not know yet
In June 2026, our own site was losing ground and we could not work out why. Here is exactly what we measured, what we found and where we are now.
What the figures showed
Over an 87-day interval, measured in Search Console: indexed pages fell from 569, in the second half of May, to a low of 441 on 1 July. A fifth of the site had disappeared from the index, without anyone having changed anything visible.
What we found, after we looked
The first cause. The blog's listing pages were declaring themselves copies of the first page. A one-line-of-code detail: the canonical URL was generated by stripping the parameters, so page 2 of the blog was saying about itself that it was page 1. Google then devalues both the page and the links on it. Out of 202 published articles, only about 47 were still one click away from the blog, and 65 articles in Romanian and 46 in English sat in the “crawled, currently not indexed” state. All technically flawless: they responded correctly, they were in the sitemap, they explicitly asked to be indexed.
The second cause. An earlier migration had shortened the URLs of some articles without leaving redirects. Whoever landed on the old URL got an error, and so did Google.
What we did
We fixed the generation of the canonical URL and put in 15 redirects, each with its destination checked separately. We deliberately left as errors the URLs that truly no longer have a counterpart, because a redirect to something else would be a polite lie told to the search engine.
What the 14 August measurement shows
Indexed pages came back from the low of 441, reached on 1 July, to 561. Almost exactly the value it all started from, 569. The recovery came in two jumps, on 8 and 11 August.
Impressions, on the other hand, have not come back. The daily average was 1,410 before 8 August and is 1,321 after. That is, slightly lower.
This is the usual shape of a recovery, and the reason it must not be misread.
Indexing moves first, because it is a technical decision that Google makes quickly. Traffic comes afterwards, if it comes, because it depends on how well those pages answer real searches. Weeks can pass between the two moments.
There is one more thing that has to be said, even if it spoils the story. The jumps of 8 and 11 August happened before we fixed the canonical URL; the fixes only went in on 14 August. So that recovery is not ours, it is the effect of moving the site to another platform. The effect of the actual fixes does not show at all yet in these figures.
It would have been very easy to put the chart in a report and say “we recovered 120 indexed pages”. It would have been true and, at the same time, a false attribution. Chapter 12 is about exactly this kind of misreading.
What remains to be solved, from the same 14 August report: 206 pages are still in the “crawled, currently not indexed” state, and 36 URLs return an error. We will know in September whether the figure of 206 goes down.
Why it does not show the next day
The reason is mechanical, not mysterious. Google does not reread your site at every change. It visits each page at an interval that depends on how often it changes and how much trust it has in the site. For a small site, an ordinary page may be revisited once every few weeks.
So a simple chain: the change first has to be noticed, then evaluated, then the resulting position has to produce clicks, and the clicks have to be enough to stand out from normal fluctuation. Each link adds time.
Seasonality complicates things too. A 20% increase in December at a gift shop says nothing about the optimization work. The comparisons that matter are with the same month of the previous year, not with last month.
What you should see and when
- Weeks 1 to 4. You should not see results yet, but activity: a list of problems found, fixes made, pages repaired. If in the first month you receive a report showing growth, ask to be shown what exactly changed on the site.
- Months 2 and 3. Indexed pages and errors in Search Console start to move. This is where the repairs show.
- Months 4 to 6. The number of pages that bring in at least a few visitors should grow. Not the average position, the number of pages that are working.
- After 6 months. If the number of pages that bring in people has not grown at all, something is not working and it deserves a serious conversation, not another month of patience.
The signs that something is wrong
Three situations in which patience is no longer a virtue.
- The reports contain no actions, only figures. If you cannot tell, reading the report, what exactly changed on the site last month, probably not much changed.
- Only growth gets reported. Any real work has bad months, algorithms that change and things that did not work out. Their absence from the report is a communication choice, not a reality.
- Nobody asks you for anything. Optimization needs information from inside the company: what customers ask, what actually sells, what has changed in the offer. A provider who never asks you for anything is working on guesswork. That is what the next chapter is about.
Key takeaways
- Repairs show in weeks, content in months, authority in over half a year.
- The first results are usually recoveries, not growth. The initial pace does not hold.
- In the first month, ask for a list of changes, not charts.
- The most honest measure of progress is the number of pages that bring in at least a few people a month.
- Compare with the same period last year, not with last month.
- A result you cannot verify yourself in Search Console is not a result, it is a claim.
Sources
- The figures in the case study come from the Search Console page indexing report for goai.ro, interval 20 May to 14 August 2026 (87 days), export downloaded on 18 August 2026.
- Google Search Console documentation on the page indexing report and on the “crawled, currently not indexed” state.