How Do You Measure the Return on Investment from SEO?
This is the question underneath every other question in this series. The arithmetic is trivial. Every single input to it is contested, which is why the subject is served so badly everywhere else.
The Question Underneath Every Other One
Every other measure in this series is a step towards one question. Was this worth the money. Everything about visibility, traffic and enquiries exists to make that question answerable. Most businesses paying for search work have never been able to answer it.
Why it goes unanswered. Each step loses something.
Appearances become visits, visits become enquiries and enquiries become work. Information is lost at every stage, so by the time you reach money the trail is thin.
What businesses do instead. Judge by feel.
Does it feel busier. That is not a foolish method and it is unreliable, since it responds to whatever happened most recently rather than to the year.
What this page assumes. The previous one.
That enquiries have been connected to their source, as far as that is possible. Our guide on leads covers that work and its limits.
What it adds. What the enquiry was worth.
Turning a count of enquiries into a judgement about money, which introduces four problems that have nothing to do with search at all.
Why it matters commercially. It ends the argument.
A business able to answer this stops debating traffic figures. The conversation becomes whether to continue, expand or stop, which is the conversation worth having.
What roughly is worth. A great deal.
Nobody needs precision here. An approximate answer built on stated assumptions supports a decision perfectly well. It is achievable by any business willing to spend an afternoon on it.
The Arithmetic Is Easy, The Inputs Are Not
What you gained, set against what you spent. That is the whole calculation and anybody can do it. Every difficulty in this subject sits in deciding what those two things are. Both are more contested than they look.
Why the sum is not the problem. Nobody disputes it.
No argument about return has ever been about the division. They are all about which costs were counted and what the gain was measured as.
The first contested input. The cost.
Most businesses count the invoice and nothing else, which understates it. Block three sets out what belongs in there.
The second. The gain.
Most use the value of the work won rather than what they kept from it. That is the commonest error in this whole subject and it is block four.
The third. The period.
Cost arrives immediately and return builds, so the window chosen decides the answer before any figures go in. Block six.
The fourth. The comparison.
Against what would have happened otherwise, which nobody can know and everybody ignores. Block seven.
Why these compound. Each one moves it.
Get all four wrong in the flattering direction, as most assessments do. The result then bears no relation to what the business actually experienced.
Why no worked example follows. It would mislead.
Substituting invented numbers into this would produce something that looks authoritative and describes no real business. Block eight explains why we refuse it.
What Counts As The Cost
The invoice is the visible part and it is rarely the whole of it. Leaving the rest out flatters the answer, which serves the supplier and misleads the business.
The fee. The obvious one.
What you pay whoever does the work, monthly or otherwise. The only element most businesses include.
What the website needed. Frequently substantial.
Development work, hosting changes, anything the site required to support what was recommended. Real money, spent because of this work, almost never counted against it.
Your own time. The invisible one.
Approving things, supplying information, answering questions, attending calls. That time has a value and in a small business it is the owner's, which makes it the most expensive time available.
What it displaced. The hardest to see.
Money spent here was not spent elsewhere. Judging this work in isolation from what the alternative would have returned is a decision rather than an oversight.
What not to include. Pre-existing costs.
Things you would have paid for regardless, such as ordinary hosting. Loading those in understates the return as badly as omitting the others overstates it.
The one that divides opinion. Content you would have written.
Where a business was publishing anyway, only the additional effort belongs in the figure. Decide which it is and record the decision rather than arguing about it later.
Why the fuller figure helps you. It settles arguments.
A business that knows its true cost can make a clear decision. One working from the invoice alone is deciding on incomplete information in a direction that favours continuing.
What Counts As The Return
Margin, not revenue. Using the value of the work won rather than what you actually kept from it is the single most frequent mistake here. It can make genuinely unprofitable work appear to be paying handsomely.
Why revenue is the default. It is available.
The value of a job is known immediately while the margin on it requires knowing your own costs. The easier figure wins, as it usually does.
What it does to the answer. Inflates it enormously.
In any trade where materials, labour or stock consume most of the ticket, revenue is a multiple of what was actually gained. The distortion is largest in exactly the trades that use this reasoning most.
Whose interest it serves. Not yours.
Revenue produces the better looking figure, so it is the figure that appears in proposals. Nobody has to intend anything by that for it to be the effect.
What to use instead. Your own margin.
Roughly what you keep on a typical job. It does not need to be precise, since a reasonable estimate transforms the calculation while an exact revenue figure ruins it.
The complication. Margin varies by job.
Different services keep different amounts. Where search brings mostly low margin work, the return can be poor while the enquiry count looks excellent.
What that implies. Enquiry quality again.
Knowing which services the enquiries were for matters as much as counting them, which is why our metrics guide insists on recording it.
Lifetime Value Changes Everything
Most businesses assess return on the first transaction. Most trades earn across a relationship. Assessing a long term channel on a single job's margin is one of the commonest ways a working arrangement gets cancelled. The business never finds out it was working.
What the first job represents. Frequently a fraction.
A customer who returns, buys other services or recommends you is worth a multiple of that first ticket. Counting only the first is counting the opening of a relationship as though it were the whole of it.
Which trades this affects most. Repeat and referral.
Anything with servicing, maintenance, renewals or word of mouth. The gap between first job and true value is largest exactly where local trades sit.
What it does to a decision. Reverses it.
Work that appears marginal on first transactions can be comfortably profitable across a relationship. The same figures, the same period and an opposite conclusion.
Why so few businesses do it. It needs records.
Knowing what a customer is worth over time requires having tracked customers over time. Plenty of businesses have never looked, which is a bigger gap than anything in their analytics.
How to estimate it roughly. Two questions.
How often does a customer come back. How often does one send somebody else. Approximate answers from experience are enough to change the assessment materially.
The caution attached. It can be abused.
A supplier can inflate lifetime value until any cost looks justified. Use your own figure, arrived at independently, rather than one supplied to you.
Where it matters most. The decision to stop.
This is the calculation that most often reverses a cancellation. Doing it before deciding is worth an afternoon of anybody's time.
The Timing Problem
The spending starts immediately and the return builds afterwards. That means the window you choose decides the answer before any figures are entered. A window short enough to feel decisive is short enough to be wrong.
What an early assessment shows. A loss, always.
Cost has accumulated and return has barely begun. That is not evidence of failure. It is the shape of the thing, which is when many arrangements are abandoned.
Why the effect compounds. Two delays.
The work takes time to affect search. Then customers take time to decide. Those delays stack rather than overlap.
What we will not tell you. How long.
No timescale for results appears anywhere in this series, because it varies by trade and competition. Our guide on judging progress gives the sequence instead.
What survives the uncertainty. Direction.
Whether the assessment is improving between periods is readable long before the absolute answer is. That is what to watch while waiting.
The other half of timing. Return persists.
Work stopped does not stop returning immediately. A site continues producing enquiries for a period afterwards, which cuts the other way and is equally overlooked.
What that means for stopping. A delayed answer.
Cancel and the effect appears gradually rather than at once, so a business that stops frequently concludes the work was never doing anything.
What Would Have Happened Anyway
Some of those enquiries would have arrived without any of this. Nobody can say how many, which is precisely why the question gets ignored. Ignoring it means treating every enquiry as something the work created.
Why it cannot be answered. No comparison exists.
You cannot run your business twice. There is no version without the work to set the results against, so this is genuinely unknowable rather than merely difficult.
What gets attributed wrongly. Your own name.
People searching your business name were going to find you regardless. Counting those as produced by the work is the clearest version of this error and the easiest to correct.
The second source. A rising market.
If demand in your trade grew, some improvement was arriving anyway. Judging entirely against your own past absorbs that into your results.
The third. Everything else you did.
Other marketing, a busy season, a competitor closing. All produce enquiries that arrive through search and were caused elsewhere.
The version that cuts the other way. Decline avoided.
If competitors improved while you held position, the work prevented a fall that never appears in any figure. That is a genuine return nothing will show you.
What to do about it. Discount deliberately.
Assume some proportion would have come anyway and assess on the remainder. A conservative assumption you chose is better than an implicit assumption of none.
Why it is worth doing. It protects you.
A business that survives its own conservative assessment can commit properly. One relying on flattering arithmetic is exposed to the first quiet quarter.
Why We Publish No Figures
You will not find a return multiple, a payback period or a client result anywhere on this site. That is deliberate. This is the page where the reason matters most, because it is the page where every competitor publishes one.
What a published figure is. Somebody else's business.
Different margins, different value per customer, different competition, different starting position. Nothing about that number transfers to yours.
How it gets used. As an expectation.
A figure read once becomes what somebody expects. When their own business produces something different, they conclude their supplier has failed rather than that the figure never applied.
What is usually behind one. Selection.
Published results tend to be the best case a firm could find. That is ordinary marketing rather than deception. It makes the figure useless as a guide.
The version to distrust most. A single memorable multiple.
A number circulated widely enough to sound like an industry fact. Trace it back and it usually rests on a narrow sample from unlike businesses.
What we offer instead. The method.
What to count, what to exclude and where the traps are. You supply your own figures, which produces an answer about your business rather than about somebody else's.
What that costs us. A persuasive page.
A headline multiple would sell better than this does. We would rather be useful to somebody assessing their own arrangement, including one they have with somebody else.
What to ask a supplier. About you.
Ask what return they think is achievable in your trade at your margins. Then ask what it rests on. Our guides on choosing an agency cover what a reasonable answer sounds like.
What A Reasonable Assessment Looks Like
Nothing above should suggest this cannot be done. It can be done roughly. Roughly is sufficient. Six characteristics separate an assessment worth acting on from one that flatters whoever produced it.
It uses margin. The first.
What you kept rather than what you invoiced, even as an approximation.
It counts the full cost. The second.
The fee, the site work and your own time, rather than the invoice alone.
It looks past the first job. The third.
Some allowance for repeat custom and recommendation, using your own figure.
It covers a long enough period. The fourth.
Long enough to contain both delays, which is longer than most people choose.
It discounts what would have happened anyway. The fifth.
A deliberate conservative assumption rather than treating every enquiry as created.
It states its assumptions. The sixth.
Every one of the above written down, so somebody else can disagree with a specific choice rather than with the conclusion.
What it produces. A range.
A defensible range with stated assumptions, rather than a single confident number. That is what a real answer to this question looks like.
How often to do it. Rarely.
Once a year is plenty, since the window has to be long. Assessing return every month guarantees the answer describes the delay rather than the work.
Where everything sits. The hub.
All nineteen guides are on the measuring performance guide.
You are judging
a relationship
on the first
job's margin.
Most businesses assess return on the first transaction while most trades earn across a relationship. A customer who comes back, buys something else or sends somebody your way is worth a multiple of that opening ticket. Counting only the first treats the start of a relationship as the whole of it. Work that looks marginal on first jobs can be comfortably profitable across a relationship. Same figures, same period, opposite conclusion.
What a defensible assessment contains:
We publish no return multiple and no client result, here or anywhere else on this site. A figure from somebody else's business tells you nothing about yours.
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.
