SEO
How to tell if your SEO is working
Not from a monthly report full of green arrows. From four numbers you can read yourself, in an order that tells you whether anything is actually moving.
THE SHORT ANSWER
Telling whether SEO is working means reading the numbers in the order they move: first the number of queries the site appears for at all, then impressions, then clicks, then enquiries. Progress shows up at the top of that chain months before it shows up at the bottom, which is why judging early work by leads alone will make genuinely successful work look like failure. Average position and rankings for a handful of chosen keywords are the two metrics most often reported and the two that tell you least.
The chain, and why order matters
Search performance moves through four stages, and they move in sequence. Reading them out of order is the most common reason people cancel work that was about to start paying, or keep paying for work that never started.
- Queries you appear for at all. The count of distinct searches where your site showed up even once. This moves first, and it is the earliest honest evidence that new work exists as far as Google is concerned.
- Impressions. How often you appeared. This follows the query count and tells you whether the searches you have entered are ones anybody performs.
- Clicks. Whether people chose you. Lags impressions, and depends heavily on which positions you hold and how the results are presented.
- Enquiries. The only one that pays. It lags everything, it is affected by things SEO does not control, and it is the worst possible early indicator despite being the one everyone watches from week one.
Where to look, for free, this afternoon
Google Search Console is the source. It is free, it is your data, and if an agency has not given you access to your own property, that is the finding.
Open the performance report, set the date range to compare the last three months against the previous three months, and turn on all four metric tiles. Then read the query count first and the clicks last. Almost everybody does the reverse.
One setting matters: look at the totals rather than filtering to your favourite keywords. Filtering to the terms you care about tells you about those terms. It tells you nothing about whether the site is growing, and a site can be growing substantially while every term you personally picked stands still.
Metrics that mislead, and how
These are not useless. They are routinely presented as evidence of things they do not evidence.
- Ranking reports for a chosen keyword list. The list was chosen by the person reporting on it. Results are also personalised by location and history, so a screenshot of a ranking is close to unfalsifiable. Ask what share of total traffic those terms represent.
- Average position. It is an average across every query you appear for, so it gets worse when you start appearing for more things. A site that expands into thousands of new searches will usually see average position fall while clicks rise. Falling average position with rising clicks is normal. Rising average position with flat query count often means the footprint shrank.
- Click-through rate on its own. Also an average, and it moves for the same structural reason. New queries typically enter at lower positions and get clicked less, which pulls the site average down even when nothing got worse.
- Traffic, undifferentiated. Branded searches, people who already knew you, and direct visits all inflate it. The question is whether people who did not know you are finding you.
- Pages published. A count of work done, not of results. Ask how many of them have ever received an impression.
How long each stage honestly takes
Nobody can give you a date, but the shape is predictable enough to plan around.
New pages get discovered and crawled in days to weeks. Appearing for new queries at all follows within weeks. Accumulating enough impressions to judge whether those queries matter takes a couple of months. Clicks and then enquiries follow after that, and the pages that eventually perform are frequently not the ones anyone predicted.
The practical consequence: judge the first sixty days on the query count and impressions, not on leads. If those two are flat after two months of real publishing, something is wrong and you are entitled to ask what. If they are climbing while leads are not yet moving, the work is doing what it should and it is too early.
The questions that expose an agency doing nothing
None of these require you to know anything technical, and all of them are hard to answer with a slide.
- "Can I have access to my own Search Console property?" There is no good reason for the answer to be no.
- "How many distinct queries did the site appear for this quarter, versus last?" One number, no interpretation, straight from Search Console.
- "Which pages published in the last six months have never received a single impression?" This is the number nobody volunteers, and on a large build it is the whole story.
- "What have you removed or merged?" An answer of "nothing" means nobody has looked at what is not working.
- "What would make you tell me this is not working?" An agency without an answer has not defined failure, which means it cannot detect it either.
What we publish about our own site
It is easy to write an article about measurement. It is harder to be measured.
We publish our own search performance publicly, refreshed daily, against a baseline frozen before we rebuilt this site. It reports what moved, what did not, and where the data is incomplete, and it is deliberately held to the standard we would apply to a client report rather than to a marketing page. It will not fill a missing value with zero, will not print a percentage against a zero baseline, and will not claim the rebuild caused anything it cannot support.
You can read it at /seo-results. It is worth looking at whether or not you ever work with us, because it is a reasonable model for what an honest report looks like: the numbers that went the wrong way are on it too.
Common questions
How long before SEO shows results?
The stages move in order and on different timescales. New pages are typically crawled within days to weeks, start appearing for new queries within weeks, and accumulate enough impressions to judge within a couple of months. Clicks and then enquiries follow after that. Judging the first sixty days by leads alone will make genuinely successful work look like failure.
Why did my average position get worse while traffic went up?
Because average position is an average across every query the site appears for. When a site starts appearing for many new searches, those new queries typically enter at lower positions, which pulls the average down even though nothing got worse and total clicks went up. It is the expected shape of a growing footprint rather than a problem.
Is a ranking report good evidence?
Weak evidence at best. The keyword list was chosen by whoever is reporting, results are personalised by location and search history, and a position for a term nobody searches is worth nothing. The useful version is what share of your total impressions and clicks those terms actually represent, which is a question a ranking report does not answer.
What is the single best indicator that early SEO work is landing?
The count of distinct queries the site appears for, compared quarter over quarter in Search Console. It moves before impressions, well before clicks, and it is very hard to inflate without doing real work. If it is flat after a couple of months of genuine publishing, ask why.
KEEP READING
The report you should be getting is the one you could check.
Every number that matters here is in your own Search Console, which means any claim made about your SEO is verifiable by you in about ten minutes. If a report cannot survive that, the problem is the report.