"SEO isn't working" is a diagnosis, and it is almost always made by looking at the wrong number. Whoever says it is usually looking at organic sessions in Analytics or at the deals closed that month, which are the last two things to move in an organic search project.
Behind that sentence there are three situations that need to be kept apart, because the answer to each is the opposite of the others: the project is working and the results have not yet reached the place you are looking; the project is working on the wrong things; the project could never have worked, and you could have known before signing.
Below are the checks to work out which of the three you are in, with the numbers we use on the Italian projects we manage and those from international research, stating each time which market they were measured on.
First case: it is climbing where nobody is looking
Since early 2026 we have been working with an Italian company that develops solar farms on agricultural and industrial land. In mid-September we track 143 keywords on its domain. Thirty-four sit between position 11 and position 20, in other words on page two. Eight of them have climbed more than eighty positions in the last thirty days.
The clicks those eight keywords brought in over the same period: zero.
| Query (Italian, with translation) | Searches/month | Position | Positions gained in 30 days | Impressions | Clicks |
|---|---|---|---|---|---|
| vendere terreno per fotovoltaico (sell land for solar) | 320 | 11 | +89 | 12 | 0 |
| affittare terreno agricolo per fotovoltaico (lease farmland for solar) | 90 | 12 | +88 | 7 | 0 |
| affitto terreno per fotovoltaico prezzi (solar land lease rates) | 90 | 13 | +87 | 13 | 0 |
| quanto costa 1 ettaro di fotovoltaico (cost of 1 hectare of solar) | 50 | 14 | +86 | 3 | 0 |
| cerchiamo terreni per fotovoltaico (land wanted for solar) | 170 | 16 | +84 | 10 | 0 |
| terreno industriale fotovoltaico (industrial land for solar) | n/a | 19 | +81 | 3 | 0 |
| cerchiamo terreni fotovoltaico (land wanted solar) | 10 | 19 | +81 | 2 | 0 |
| aziende interessate ad affittare terreni per fotovoltaico (companies looking to lease land for solar) | 10 | 20 | +80 | 3 | 0 |
Anyone looking at Analytics sees zero. Anyone looking at rankings sees the best month since the project began. It is the same month.
Page two is where a project sits before it becomes visible, and it is also where it produces nothing. Between position 11 and position 10 there is not one position of difference: there is a step, because nobody scrolls below the tenth result. That is why a project seven months in can have gone very well and look stuck. And that is why the first number to check when someone says SEO isn't working is not traffic, it is the distribution of rankings.
Across all the Italian projects we track today, 278 keywords in total, the picture is this: 122 on page one, 59 on page two, 39 on page three, 12 between pages four and ten, 46 outside the top hundred. 21% of the work done sits in a zone that does not yet produce a single click and that the client does not see in any traffic report.
Further reading: if traffic was arriving and then fell, the problem is a different one and is isolated in a set order. We covered it in website traffic drop: how to find the cause.
How long a new page takes, in numbers
Ahrefs tracked one million URLs seen by its crawler in September 2023 and followed them for twelve months. 1.74% reached the top 10 within a year; filtering for non-empty English-language content only, 6.11%. In the same study, on a sample of 1.3 million US keywords, the page in position one is on average five years old and 72.9% of pages in the top 10 are more than three years old (Ahrefs, 2025, US market).
This is US data, and smaller markets are less crowded almost everywhere, which shortens the timescales but does not change the order of magnitude: a page published in March is not competing with pages published in February, it is competing with pages that have been there for years.
Then there is the part nobody puts in the proposal. Ahrefs again, analysing around 14 billion pages in its index, found that 96.55% get no clicks at all from Google and another 1.94% get between one and ten a month (2023 study). Publishing and ranking are two different activities, and the first on its own yields close to nothing.
Google, for its part, does not promise timescales. Its starter guide says that "some changes might take effect in a few hours, others could take several months" (Google Search Central). If your agency gave you a date, that date did not come from Google. On real timescales measured on our projects we wrote how long it takes to see SEO results.
"I want to be number one on Google"
This is the expectation that breaks more projects than any other, and it has two problems.
Google states the first one on its page about choosing an SEO: "No one can guarantee a #1 ranking on Google", and it tells you to beware of SEOs that claim to guarantee rankings or allege a "special relationship" with Google (Google Search Central). This is not a courtesy formula: rankings depend on a contest you do not control, with competitors who move and an algorithm that changes several times a year.
The second problem is that position one is not the finish line it seems. On macropix.it, an Italian LED wall manufacturer we work with and whose numbers we have already published, the 12 September 2026 snapshot of the tracked keyword set gives a share of voice of 21.26%. Ahead of it is amazon.it at 26.89%, behind it visualled.com at 3.96% and dominodisplay.com at 3.48%. The site is first or second on the sector's main queries and a marketplace still takes part of the click share by putting product listings on those queries. The number to negotiate is not the position, it is the share of clicks across the keyword set.
Further reading: the full path of that project, from position 88 to position 2, is told in SEO for manufacturing.
Second case: it is working on the wrong things
Here the traffic is there and it is no use. It is the hardest situation to recognise, because the reports look good.
The American agency Grow and Convert published aggregated data from its clients: keywords with purchase intent convert at 4.78%, informational ones at 0.19%. For one software client, 15% of articles produced 80% of conversions; the sixteen purely informational articles generated 40 conversions in total, just over two each (Grow and Convert, data on US clients). The ratio is twenty-five to one, and there is no reason to expect it to reverse in the UK.
The check takes ten minutes. Open Search Console, take the twenty queries that bring the most clicks and ask yourself, for each one, whether the person typing it is buying or studying. If the top fifteen are definitions, guides and "how to", the project is producing readers. It is not a technical failure, it is a selection error upstream, and it is fixed in keyword research before it is fixed in the content. The key lies in how you read the search intent behind a query.
The other version of this case is the right traffic on a site that does not capture it. If a service page gets eighty visits a month from commercial queries and has no form, no visible phone number and no reason to get in touch, the problem is not on Google. It is downstream, and no ranking will fix it.
Third case: it could never have worked
Three conditions make an SEO project a losing bet from the start, and all of them can be checked before signing.
The demand does not exist. If adding up all the commercial queries in your sector gives you a few hundred searches a month in your country, ranking first on all of them is worth a few dozen visits. It happens often in very specialised B2B, where the buyers number a few dozen nationwide. The sum to do first is in our piece on whether SEO is worth it.
The margin does not hold up. A monthly retainer has to be covered by the margin on the customers it brings in, not by turnover. With a low unit margin and a short sales cycle, paid advertising gets there sooner and costs less for the same result: it is worth considering a Google Ads campaign as the primary channel and keeping organic as a second-tier investment.
The horizon is shorter than the technical timescale. A six-month contract on a new domain buys the work, not the results. Looking at the numbers in the previous section, six months on a domain with no history gets you a good page two. If the business needs leads within the quarter, SEO is not the channel to switch on first.
When the problem is not the SEO provider's
There is a fourth possibility that rarely comes up in discussions between client and agency, because it is awkward for both: the work is right and it never reaches the live site.
On Search Engine Journal, Bill Hunt, president of Bisan Digital, listed five structural reasons why SEO fails to deliver inside companies, and none of them concerns the competence of whoever does it: nobody at leadership level is accountable for organic visibility, departmental incentives are misaligned, content goes out without a strategy, the CMS or the IT department blocks changes, and there is no stable way of deciding what gets published (Search Engine Journal, July 2025). The analysis is written for large American companies, but in small and mid-sized businesses the same thing takes a smaller and more lethal form: technical changes wait for the web developer, copy waits for the owner's approval, and three weeks of work sit in a folder.
An honest check takes half an hour: take the list of changes delivered in the last three months and mark how many are live. If less than 70% is live and the rest is waiting on someone inside the company, the project is not failing, it is waiting. This also changes how you read a report: the indicators to look at in that case are the ones in how to tell if your SEO agency is working.
Seven expectations to put in writing before signing
Almost every breakdown we have seen starts from one of these seven, never written down anywhere and taken for granted by both sides. The right-hand column is the number to agree in place of the sentence.
| The expectation | What really happens | The number to agree |
|---|---|---|
| "I'll see results within three months" | At three months rankings and impressions move, not clicks | How many keywords must reach the top 20 by month three |
| "I want to be number one on Google" | Nobody can guarantee it, and position one does not take all the clicks | The share of clicks across the agreed keyword set |
| "Two articles a month is enough" | 96.55% of published pages get no clicks from Google | How many pages are needed to cover the keyword set, not how many per month |
| "You'll take care of everything" | Changes stall with whoever manages the site and whoever approves the copy | Internal response time and who signs off approvals |
| "Once it's fixed it stays fixed" | Competitors move and the algorithm changes several times a year | How often pages that already rank are reviewed |
| "The audit score has to reach 100" | Tool scores are not ranking factors | Which errors block indexing and rendering; the rest are ignored |
| "It costs too much for what it is" | The retainer buys hours of work, not rankings | Monthly hours, what they produce and how they are reported |
The row on the audit deserves a note. The score tools give a site is an arbitrary aggregate decided by whoever wrote the tool, and raising it from 68 to 94 moves nothing if the problems fixed were cosmetic. We gathered the evidence on what really counts in ranking factors, and the most common beliefs in SEO myths. How to read an audit without chasing the score is in SEO audit.
How to get a project back on track halfway through
If the contract is already signed and the feeling is that it is not going well, the conversation to have is not "is it working or not". It is this one, in four steps, each ending with a number.
- Take the last six months from Search Console against the previous six, and look at impressions before clicks. Rising impressions with flat clicks mean page two, the first case in this article.
- Count the keywords in the agreed set that sit between position 4 and 20. That is the reserve: they can turn into traffic over the next three months without writing anything new.
- Look at the twenty queries that bring the most clicks and separate commercial from informational. If fewer than five are commercial, the problem is selection, not execution.
- List the changes delivered and not published. If they are the majority, the meeting needs to happen inside the company before it happens with the provider.
Four numbers, and the discussion stops being one opinion against another. If after these checks the project really is stuck, the choice between renegotiating, changing provider or bringing the function in-house is simpler: the terms of comparison are in in-house SEO or agency, and the list of things to demand is in what to expect from an SEO agency.
When we take over someone else's project, we spend the first month redoing exactly these four sums before touching anything. It is also how we set up the SEO projects we manage: goals written in rankings and share of clicks, not in promises. Results measured on other clients are in our projects.
Frequently asked questions
Not before six months, and only if in those six months neither impressions nor rankings have moved. If rankings rise and clicks stay flat, the project is working: it means the pages are on page two, where traffic is close to zero. On one Italian project we manage, 34 keywords out of 143 currently sit between position 11 and 20, and eight of them climbed more than eighty positions in thirty days, with zero clicks.
No. Google says so in its own documentation for anyone choosing an SEO: nobody can guarantee a number one ranking, and you should beware of anyone who promises rankings or claims a special relationship with Google. A serious contractual goal is written as the number of keywords that must reach a band of positions by a given date, not as individual positions.
It depends on which queries bring that traffic. If they are definitions and guides, the problem is keyword selection: according to data published by Grow and Convert on US clients, queries with purchase intent convert at 4.78% against 0.19% for informational ones. If the queries are commercial and enquiries still do not come in, the bottleneck is the landing page, not Google.
Not on its own. Analysing around 14 billion pages in its index, Ahrefs found that 96.55% get no clicks at all from Google and another 1.94% get between one and ten a month. Publishing volume only matters once you have chosen a set of queries with real demand and intent that matches what you sell.
No, and chasing it wastes time. Those scores are arbitrary aggregates decided by whoever wrote the tool and they are not ranking factors. The only errors that always need fixing are the ones that stop Google from crawling, rendering and indexing pages; the rest are judged by how much traffic the page in question is worth.
It makes sense if the unit margin is low, the sales cycle is short and you need leads within the quarter. In that case the paid auction gets there first. Stopping organic entirely has a deferred cost, though: the rankings you earned are lost over a few months and rebuilding them takes more work than it took to hold them.
Something the industry rarely says: the wrong expectation almost never comes from the client. It comes from the proposal they read six months earlier, where someone wrote "number one on Google" to close the deal. Whoever sells SEO by promising rankings is buying a client who in eight months will say SEO isn't working, and will be right about what they were sold. The fastest way to make a project work is to write the four numbers above into the contract, and to accept losing the deals where those numbers are not welcome.
