Your SEO report looks great. Has it brought you any work?
There is a particular kind of document I have seen a lot of over the years. It arrives monthly, it is fifteen pages long, and it is full of numbers going up.
Domain Authority: 34, up two points. Visibility score: improved. Keywords ranking: 847, up from 812. Backlinks: 1,240, up 60. There are graphs. Everything is green.
And the business owner reading it has no idea whether any of it means anything, because the only number they actually care about is not in the report at all.
Did it bring us any work?
Where those numbers come from
Here is the thing nobody explains when they hand you that report.
Most of those metrics do not come from Google. They come from third-party tools like Semrush and Ahrefs, which are genuinely excellent bits of kit, and which I use myself. But it is important to understand what they are doing.
They are not reading your Google account. They cannot. They crawl the web themselves, sample search results, and build a model of what they think is happening. The numbers they produce are estimates from a simulation, not measurements of reality.
Domain Authority is the clearest example. It is a score invented by an SEO software company. Google has said repeatedly it does not use it. It is a useful rough proxy for comparing two sites, and it is not a thing Google is looking at when it decides where to rank you.
So when your report says Domain Authority went up two points, what has actually happened is that a third-party model’s opinion of your site changed slightly. Nothing occurred at Google. No customer did anything differently.
The two tools that are actually measuring
There are two sources that are not estimating, and both are free.
Google Search Console is Google telling you, in its own words, what happened. Which queries you appeared for. How many times. Where you ranked. How many people clicked. Not a model of what might be happening, but the record of what did.
Google Analytics tells you what those people did once they arrived. Did they read anything? Did they ring you? Did they fill in the form?
Between them you can answer the only question that matters: people searched for something, some of them arrived, and some of them became customers. That is the chain. Everything else is a proxy for it.
The month everyone’s visibility collapsed and nothing happened
Here is my favourite illustration of why these numbers deserve less trust than they get.
On 11 September 2025, Google quietly switched off something called the num=100 parameter. It was a technical shortcut that let rank tracking tools pull a hundred search results in one go rather than ten at a time. No announcement, no blog post. Google later described it as never having been an officially supported feature.
For the average business owner that sounds like an obscure bit of plumbing. Then people opened their reports.
Impressions fell off a cliff. Roughly 88% of sites saw total impressions drop, and around 78% saw their count of ranking keywords fall. Dashboards that had been climbing steadily for years suddenly looked like a disaster had struck.
Nothing had happened. Not one customer behaved differently. No site lost a single ranking.
What had actually been going on is worse, and funnier. A large slice of those impressions had never been human. They were rank tracking bots, hammering Google a hundred results at a time, and every one of those was being counted as somebody seeing your site. Estimates suggest something like 30 to 40% of reported impressions were this synthetic noise. When the shortcut disappeared, the bots went with it, and the numbers dropped to something closer to the truth.
There was an even stranger side effect. Average position improved overnight, because all those page-seven appearances stopped being counted. Sites that had done nothing at all suddenly looked like they were ranking better.
So in one week, a lot of businesses saw their visibility apparently collapse and their rankings apparently improve, and both were measurement artefacts. If your agency had wanted to, they could have taken credit for the second while blaming Google for the first.
That is the whole argument in one event. A number moved a long way, and reality did not move at all.
Why agencies report on the other stuff
I want to be fair here, because there are two reasons and only one of them is cynical.
The honest reason: the third-party tools genuinely are useful, and I would not work without them. But they are useful for diagnosis, not for measuring results.
Let me give you the clearest example I have. Back when Google released the Panda and Penguin updates, a lot of sites got hammered overnight. I spent a long stretch recovering sites from both, and those tools were how I found the problem. Crawling a link profile to see what was pointing at a site, spotting the patterns, working out which pages were thin, that is exactly what they are built for and they were invaluable.
But here is the part that matters. Once the tools had shown me what was wrong, none of the numbers in them fixed anything. What fixed it was writing genuinely good content, and getting rid of the piles of spammy rubbish links somebody had built years earlier. Slow, unglamorous work.
The tools diagnosed. The work cured. That is the correct relationship between the two, and it has not changed.
There is a bitter irony in Penguin in particular, which is worth sitting with. It punished sites for chasing exactly the sort of vanity metric your report is full of. People had spent years watching their backlink count climb, congratulating themselves, and then one morning Google decided most of those links were garbage and the whole thing collapsed. The number had been going up the entire time.
The less honest reason agencies report on this stuff: those numbers reliably go up. If you do almost any activity at all, you will accumulate keywords and links and your visibility score will drift upward. It makes a satisfying graph. Whereas enquiries are stubborn, they fluctuate, and they can go down in a month where you did excellent work.
A report full of green arrows is much easier to send than a conversation about whether the phone rang more.
The number that can go down is the useful one
Here is a thought that took me a while to come round to.
A metric that only ever climbs is not a metric, it is a decoration. Keyword counts, backlink totals, visibility scores: these drift upward if you do almost anything, and they rarely fall in a way that means much. That is precisely why they make such comfortable reporting. Nothing in that document can ever bring bad news.
Real measurement is not like that. Search Console will fall off a cliff. Enquiries will halve in a month. Your rankings for the terms that pay will drop three places overnight.
That is not a flaw in the measurement. That is the entire point of it.
A sharp drop is the most valuable signal you will ever get, because it is the only thing that tells you something has changed and you have a window to find out what. Both of the big recoveries I worked on started exactly that way: a graph falling off a table, and somebody ringing me in a panic. You cannot fix a problem you have no way of seeing, and no visibility score in the world would have shown it.
So when you look at your numbers, do not just check whether they went up. Check whether they could have gone down. If the answer is no, you are not being told anything.
But not every drop means trouble
The obvious caveat, and the num=100 business is the proof of it.
Sometimes the measurement changes rather than reality. A tracking change, a seasonal dip, a Google reporting quirk, a bank holiday. Panicking at every wobble is as useless as celebrating every rise.
The skill is telling the difference, and it mostly comes down to one question: can I see this in more than one place? If impressions dropped but clicks and enquiries held steady, something changed in the counting. If impressions, clicks and the phone all went quiet together, something changed in the world and it is worth looking into properly.
Which is another argument for measuring things that connect to actual customers. Two independent numbers moving the same way is evidence. One number moving on its own is a puzzle.
The metrics I would actually look at
If you are a local service business, here is the short list. It is boring, and that is the point.
Enquiries. Calls, forms, emails, whatever you get. Count them. Most businesses genuinely do not, and then wonder whether their marketing is working.
Where they came from. Even roughly. “Found you on Google” versus “your van was outside next door” is the single most useful question you can ask a new customer, and it costs nothing.
Search Console impressions and clicks, for the searches that matter. Not all searches. The ones where somebody is looking to buy.
Your position for the handful of terms that pay. Not 847 keywords. The five or six phrases that turn into work.
Google Business Profile calls and direction requests. For most local businesses this is where the work actually comes from, and it sits in a completely different place from the website numbers.
That is it. You could fit it on a postcard.
Why the number of keywords is a trap
One specific thing, because it catches everyone.
“You now rank for 847 keywords” sounds tremendous. But that number includes every phrase you appear for anywhere in the top hundred results, including hundreds nobody searches and dozens where you are sat on page seven.
Ranking eleventh for a phrase two people a month type is worth precisely nothing. Ranking third for a phrase forty people a month type, when those people are about to hire someone, is worth your entire month.
One good keyword beats four hundred irrelevant ones, and a report that celebrates volume is measuring the wrong thing on purpose.
Test things, then check whether they worked
Here is the shift I would encourage, and it is a mindset more than a metric.
Treat the work as a series of small experiments rather than a monthly ritual. Write the page about the specific problem people keep ringing about. Add the town you cover to the service page. Rewrite the homepage so it says what you do in the first line.
Then wait, and look. Did impressions rise for those searches? Did clicks follow? Did anybody actually ring?
Some things will work and some will not, and you learn more from the ones that do not. But you can only do this if you are measuring something real. If the only feedback you get is a visibility score, you cannot tell which of the twenty things you did was the one that mattered.
The question to ask your agency
Not an aggressive one. Just this:
“Which of these numbers connects to somebody actually contacting us?”
A good agency will welcome that question, because it lets them show you the chain: this search, these impressions, these clicks, these enquiries. That conversation is far more interesting than a slide of green arrows.
If the answer wanders, or you are told the results are longer term and the leading indicators are strong, that is worth noticing. It might be true. SEO genuinely does take time and anyone promising instant results is lying to you.
But “it takes time” and “we are measuring things that cannot be connected to your business” are different statements, and only one of them is acceptable.
Which is why I would rather have a plan than a hack
All of this is really an argument for something simpler: pick a sensible strategy, stick to it, and measure it against real outcomes.
The alternative is what most of this industry does, which is to chase whatever appears to be working this month. There is always something. A new tactic doing the rounds on social media, a clever shortcut somebody has spotted, a way of getting cited that nobody has caught onto yet.
I have written elsewhere about why I leave the clever new tactics alone, so I will not repeat the whole argument. The short version is that anything working by gaming the system gets dealt with eventually, and the damage is rarely confined to the guilty.
What is relevant here is what it looks like on a report while it is happening. Every one of those sites I recovered from Panda and Penguin had numbers going up right until the moment they collapsed. The metric was not lying exactly. It was measuring something real. It was just measuring something that was about to be worthless.
That is the danger of a number you cannot connect to a customer. It cannot tell you the difference between progress and an accident waiting to happen.
Paid citations, and the “best of” lists
The current version of this is worth naming, because it is being sold hard right now.
AI leans heavily on “best of” listicles when deciding who to recommend, which people have noticed. So there is now a trade in getting yourself onto those lists, sometimes by paying for the privilege.
It is also a good example of a metric that flatters you while telling you nothing. Being cited on more pages is easy to count and easy to celebrate. Whether those citations came from anywhere a customer would trust is a different question, and it is the one that decides whether the work survives.
So think about what you are actually buying.
If a directory has genuine criteria, checks them, and lists you because you meet them, that is a real credential and paying a listing fee for it is ordinary business. Trade bodies and proper vetted directories work like this.
If somebody is simply selling positions on a list dressed up as an editorial recommendation, you are buying a fake endorsement. That is a problem on three fronts. Google’s guidance names inauthentic brand mentions directly. Undisclosed paid endorsements are a consumer protection issue in the UK, not merely a search one. And those pages are precisely the sort of thing that gets targeted when the next update lands, at which point everyone on the list finds out together.
The test I use is simple: did somebody independently judge that this is true, or did I pay for the judgement? If it is the second one, I leave it alone, however well it appears to be working at the moment.
The boring alternative
So the plan I would rather have is unglamorous and it does not change much month to month.
Get the foundations right. Write genuinely useful content about the things customers actually ask. Earn reviews properly. Be the obvious answer to the questions in your trade. Measure whether the phone rings.
None of that gets penalised in a future update, because there is nothing to penalise. It is just being a findable, credible business. And the great advantage of a plan you can defend is that when Google does change something, you spend the week reading about it rather than repairing your site.
What I do about it
I will be straight about my own practice, since it would be a bit rich otherwise.
I use the third-party tools, and I would not be without them. Fifteen years in some of the most competitive corners of search, including pulling sites back out of Panda and Penguin, has left me with a healthy respect for what they are good at. I use them to work out what is wrong and what to do next.
But I report on Search Console and on enquiries, because those are what actually happened. And I would rather tell you a month was flat than dress it up with a metric that went up on its own.
That is not me being noble. It is that if we cannot both see what is working, neither of us can make it work better.
Want to know which of your numbers actually matter? Get your free audit, or email hello@inkon.co.uk. I read every message myself, and I will tell you straight.