How to Measure SEO Performance
Almost everybody measures the wrong thing. They do it because the wrong thing is easier to count. Rankings are visible and traffic is a number on a screen. Neither one tells a business owner whether the money is working.
What You Are Actually Trying To Find Out
You are not trying to find out whether rankings moved. You are trying to find out whether the business is getting more of the enquiries it wants. Every measure in this series is useful only insofar as it helps answer that. A great many of the measures in circulation do not help at all.
Why the framing matters. It sorts everything.
Once the question is fixed, deciding what belongs in a report becomes straightforward. Anything that cannot be connected to it comes out, however impressive it looks.
What the question is not. Whether the work happened.
Effort and outcome are different things. A month of genuine work that produced nothing is still a month that produced nothing. A report should be capable of saying so.
Who this series is written for. The person paying.
Not for marketers. For an owner who commissioned this, is receiving reports and wants to know whether to continue.
The uncomfortable part. The answer can be no.
Measurement that can only produce good news is not measurement. If the arrangement is not working, the point of measuring is to find that out early.
What follows. Structure before metrics.
Why the wrong things get measured, then the three layers everything sits in, then which measures earn a place for your particular business.
Why Everybody Measures The Wrong Thing
Rankings and traffic dominate reporting for a simple reason. They are easy to count, they are available without asking the client for anything and they usually move in a direction that looks like progress. None of that makes them the outcome anybody is paying for.
Why rankings lead. Everybody understands them.
A position is the thing people outside the industry think this work is. Reporting one requires no explanation, which makes it the path of least resistance for both sides.
Why traffic follows. It is one number.
A single figure that went up or down, available instantly. Reporting it takes seconds and interpreting it properly takes considerably longer, so the first happens and the second frequently does not.
What both share. Availability.
They can be gathered without any involvement from the business. Everything that actually matters requires the client's own records, which is harder. That is the whole explanation.
The consequence. Accurate and useless.
A report can be entirely correct, well presented and tell a business nothing about whether its money is working. That combination is extremely common and it rarely involves anybody acting in bad faith.
How the situation persists. Nobody objects.
The client cannot easily say what is missing. The supplier is not being asked for it. The arrangement holds until somebody wonders why the enquiries have not changed.
What to do about it. Ask for the third layer.
Block three sets out the structure. The layer almost every report stops short of is the one that answers your actual question.
The Three Layers
Measurement works through three layers in order. Whether you appear at all. Whether the right people click and stay. Whether anything happens as a result. Most reporting covers the first thoroughly, touches the second and never reaches the third.
Layer one. Visibility.
Are you being shown for the searches that matter. This is where impressions, positions and the number of different searches you appear for live. It is also the easiest layer to measure.
Layer two. Engagement.
Are the right people arriving. Does the page do its job when they do. Clicks, the searches that produced them and what visitors do next.
Layer three. Outcome.
Did anybody get in touch, book or buy. This is the only layer the business is actually paying for and the only one that cannot be gathered without the business's help.
Why the order matters. Diagnosis.
A problem in one layer looks completely different from a problem in another. Plenty of visibility with no engagement is a different fault from no visibility at all.
Where reports stop. Layer one.
Because it is available without asking anybody for anything. That single practical fact explains the shape of most reporting in this industry.
What the layers do for you. A checklist.
Read any report you are given and ask which layers it covers. If it never reaches the third, you are being shown activity rather than results.
The Right Indicators Differ By Business
There is no standard set of measures that suits everybody. A report applying the same list to every client is a template rather than an analysis. A plumber, an online shop and a care home make money in genuinely different ways, so they need genuinely different indicators.
The test for any measure. Would it change a decision.
If a figure moved substantially and nothing about your plans would change, it does not belong in the report. That single test removes most of what usually appears.
A local trade. Calls, mostly.
The valuable outcome arrives by telephone, frequently within hours of the search. Anything that cannot connect to a ringing phone is secondary.
An online shop. The full chain.
Visits through to baskets through to purchases, with value attached. The measurement is more complete here and the risk is measuring it thoroughly while ignoring where it breaks.
A considered service. Slower and softer.
Where the decision takes weeks and involves several people, first contact matters more than immediate conversion. The useful measures sit earlier in the chain.
What everybody shares. One anchor.
Enquiries attributable to search. The form differs and the question does not, which is why it appears in every version of this.
When to agree them. At the start.
Measures chosen before work begins prevent the argument later. Choosing them afterwards means choosing the flattering ones.
Where The Numbers Come From
Three sources between them contain everything worth knowing. Each one sees a different part of the picture. This series covers what the numbers mean rather than how to operate anything, so the mechanics live elsewhere.
What happens in the results. Search reporting.
Which searches showed your pages, how often, whether anybody clicked and roughly where you sat. Free, plus covered properly in our Google Search Console guide.
What happens on the site. Analytics.
What visitors did once they arrived, where they came from and whether they reached anything that counts. Our Google Analytics guide covers the tool itself.
What happens afterwards. Your own records.
Which enquiries turned into work and what that work was worth. This lives in your books rather than in any tool. It is the layer nobody can gather for you.
Why we describe sources not products. A programme position.
We name no paid tool anywhere on this site. Google's own free tools are named because they are the sources rather than somebody's product recommendation.
What none of them contain. The judgement.
Whether the work is worth continuing is a decision. No source answers it. They supply the evidence and somebody still has to read it.
The practical order. Third source first.
Getting your own outcome records into shape does more for measurement than any amount of work on the other two.
No Single Source Tells You Everything
The three sources will not agree with each other. Expecting them to is the single commonest cause of confusion in this subject. They are not measuring the same thing, so their totals were never going to match.
Why they differ. Different vantage points.
One is counting what happened in a results page. Another is counting what happened on your website. Those are separate events with separate opportunities to be missed.
What that means practically. Neither is wrong.
A gap between two figures is usually both of them being correct about different things rather than one being faulty. Our comparison guide covers exactly why.
The mistake to avoid. Reconciling them.
Hours disappear into making two numbers match. They will not. The effort produces nothing except a false sense that one has been fixed.
What to do instead. Label the source.
Every figure in a report should say where it came from. That prevents somebody comparing two numbers that were never comparable.
Which to trust for what. Depends on the layer.
Search reporting for visibility. Analytics for behaviour on the site. Your own records for outcomes. Each is authoritative for its own layer and unreliable outside it.
The one that matters most. Whichever the argument is about.
When two sources disagree about something commercially important, the answer is usually in the third one nobody consulted.
Measure Against Something
A number on its own means nothing. Visits last month is a fact rather than information. It becomes information only when set beside something. A previous period, an expectation or a starting point.
The three comparisons available. Each answers differently.
Against your own past, against what you expected and against where you began. Those answer three different questions and a report should be explicit about which it is showing.
The most reliable one. Yourself, earlier.
Your own previous performance is the only genuinely like for like comparison available, because everything else about the business is held constant.
The comparison that misleads. Industry figures.
Published averages combine businesses with nothing in common. Comparing a local trade against one tells you nothing, which is why we publish none.
The one everybody skips. A starting point.
A record taken before any work begins. Without one, every later conversation about what changed becomes opinion. our benchmarking guide covers taking one.
Why it cannot be done later. The moment passes.
Some of what should be recorded is not fully available retrospectively. An engagement without a baseline can never fully establish what it achieved.
The labelling rule. State the basis.
A percentage with no stated comparison is the commonest way a report misleads without anybody intending it to.
Reporting Is Not Fixing
A beautifully presented decline is still a decline. Measurement exists to trigger action, so an arrangement where the numbers are reported carefully every month while nothing changes is failing, regardless of how good the report is.
How the substitution happens. Gradually.
Reporting is visible, schedulable and demonstrably done. Fixing is uncertain and harder to show. Over months the visible thing quietly replaces the uncertain one.
What it looks like from the outside. Diligence.
Reports arriving on time, well made, full of charts. Nothing about the appearance signals that the underlying work has stopped.
The question that exposes it. What changed.
Ask what was actually done to the site this month and what it was expected to achieve. A report describing measurements and no actions is the whole problem in one document.
Why we state it as a position. It is ours to hold.
The measuring series is where this belongs, since measurement is the thing being mistaken for the work. Other guides on this site refer to it and it lives here.
What a good report contains instead. Three things.
What changed, why and what happens next. Our monthly report guide sets that out. The third one is what usually goes missing.
The client's part in it. Asking.
A supplier reporting without acting is frequently doing so because nobody has asked for anything else. Asking changes the arrangement immediately.
How Often To Look
Less often than you want to. These figures move constantly for reasons nobody controls, so checking daily produces anxiety about noise and tempts people into changing things that were working perfectly well.
What daily checking produces. Reaction.
A figure dips, somebody alters a page, the figure recovers on its own and the alteration takes the credit. That teaches the wrong lesson permanently.
What the right cadence depends on. The decision.
Look as often as you would act on what you find. For most businesses that is considerably less frequent than the reporting cycle they are on.
Why monthly is a habit rather than a principle. Invoicing.
Reports are monthly because fees are monthly. That is an administrative rhythm being mistaken for an analytical one. A month is frequently too short a window to see anything real.
What to compare against. Like periods.
The same span a year earlier, so seasonal shape does not masquerade as performance. Most trades have a shape to their year that will otherwise mislead twice annually.
Contact is not reporting. A useful distinction.
Regular contact from whoever does the work is a different thing from a formal report. We commit to contact every three weeks, which is about the work rather than about the numbers.
What to do between looks. The work.
Time spent watching figures is time not spent improving pages, which is the thing that eventually moves them.
What This Series Covers
Nineteen guides in five groups. Most people arrive with one specific question, so this is arranged to let you find yours rather than read the lot.
Getting started. Two guides.
This page, plus how to tell if SEO is working, which is the question most people actually arrived with.
The metrics. Five guides.
What each number means, including impressions, clicks and position, organic traffic, bounce rate, domain authority and the engagement measures.
Reading the data. Five guides.
Traffic drops, trends over time, which comparison to use, how seasonality distorts everything and how to take a baseline.
Reporting. Five guides.
Dashboards, what belongs in a monthly report, explaining it to somebody non-technical, which measures matter and why rankings alone mislead.
The money. Two guides.
Connecting search traffic to actual enquiries, then working out what the return was. The second cannot be written before the first.
Where to start. Depends why you came.
If you are wondering whether your current arrangement is working, start with the second guide in the first group. If you are trying to make sense of a report, start with the metrics.
A beautifully
presented decline
is still a decline.
Reporting is visible, schedulable and demonstrably done. Fixing is uncertain and harder to show. Over enough months the visible thing quietly replaces the uncertain one. Nothing about a well made report signals that the underlying work has stopped. Measurement exists to trigger action rather than to demonstrate effort.
What we hold ourselves to:
Ask what was actually done this month and what it was expected to achieve. A report full of measurements and no actions is the whole problem in one document.
Every guide.
One question.
What each number actually means, how to read the data without being misled, what belongs in a report and what does not, then how to connect any of it to the enquiries the business is actually paying for.
