An audit lands in your inbox. Forty pages. Two hundred issues, colour-coded, scored by priority. Every one of them is real. You read it twice and you still do not know the only thing you needed to know, which is whether to spend the money.
That is not a failing of the audit. It is what an audit is.
Three different acts
Diagnosing a local search problem involves three separate things, and they are routinely confused because they arrive in the same document.
Gathering. Sampling rank across an area, scanning a listing for risk, baselining competitors, checking whether the business is described consistently across the web. Observation. No opinion.
Grouping. Sorting what came back into what is structural, what is market, what is trust, and what is noise. Still no opinion — but a great deal of judgement about which pile a finding belongs in, because the pile determines whether anything can be done about it.
Judging. Deciding what it means for this business, with this money, in this market, and saying so plainly.
Software is now genuinely good at the first. It is passable at the second. It cannot do the third, and it does not know that it cannot, which is why the reports are so confident.
What software produces instead of judgement is a recommendation engine: two hundred findings, each with an action attached. Fix the titles. Build the citations. Publish more. Every action defensible in isolation. Nothing anywhere in the document asking whether the whole exercise is worth attempting.
The four questions
Judgement is not a mood. In our diagnostic process it is a defined step with four questions, and the report is not finished until a person has answered all four in writing.
Is Google structurally capable?
Can the search engine put this business in front of that customer, in that place, at all? Sometimes the answer is no, and no amount of work changes it. Local rank is a field, not a number — it decays with distance, and a business at the wrong edge of a saturated city may be permanently out of reach of the far side of it. We wrote about that at length: checking your own ranking measures nothing.
If the answer here is no, the remaining three questions barely matter. Nothing downstream is worth discussing.
Is spend rational?
Not “would this work”. Would this work for the money, against the value of a customer, at the rate this business wins them, within a period the business can survive.
A tactic that produces results in fourteen months is a good tactic and a bad recommendation for a business with four months of cash. A campaign that costs more per acquired customer than the customer is worth is a failure that shows up as a success on every dashboard, because the dashboard measures rankings and traffic and neither of those is money.
What expectation must die?
Almost every business arriving at a diagnostic carries an expectation that will not survive contact with the data. That it should rank first everywhere it operates. That the competitor above it is doing something clever, rather than being nearer. That last year’s enquiry volume is the baseline and this year is a failure, when the whole market has contracted underneath it.
Naming the expectation and killing it is the most valuable thing in the report and the least welcome. It is also the part no software will ever write, because it requires someone to tell a paying customer that a belief they hold is false.
What risk must be protected?
Most businesses have one thing that is quietly working. A listing that ranks well for the term that actually converts. A page that has accumulated trust over a decade. A review profile that is doing more work than the website.
Aggressive change endangers it. A recommendation to restructure a site, consolidate pages or rebuild a listing can be perfectly correct in the abstract and catastrophic in this specific case, because it puts the one working thing at risk to improve six things that never mattered. Someone has to identify what must not be touched, and say so before anybody starts.
Why a person has to answer them
Look at what those four questions have in common. Each one requires somebody to be willing to be wrong in public, and at least two of them require telling a customer something they are paying not to hear.
Software will not say the ground is unwinnable, because saying so ends the engagement. It will not say the spend is irrational, because it exists to justify spend. It will not kill an expectation, because it has no relationship to preserve and no reputation to lose. It will recommend, always, because a recommendation is what it is for.
A person answering these questions is accountable for the answer. That is the entire difference. Not intelligence, not access to better data — accountability. The judgement is worth something precisely because someone can be held to it afterwards.
The uncomfortable part
The honest answer to “is spend rational” is sometimes: not with us, not now, not on this.
An audit that concludes the money should not be spent has done its job perfectly and earned nothing beyond its fee. That is an awkward commercial position and it is the one worth occupying, because the alternative is a business model that requires every diagnosis to end in a sale. Once that is true, the diagnosis is decoration.
Anyone can produce the data now. The scarce thing was never the data. It is somebody willing to read it and tell you the answer is no.
What to ask
Before you buy any audit, ask who answers the four questions, and ask to see the answers in writing rather than as a list of actions.
If the deliverable is a report of findings with recommendations attached to each one, you have bought gathering and grouping. Nobody has judged anything. You are still holding the decision, and you now have two hundred reasons to spend.
See where you stand
The free reality-check produces a proximity grid run against your Google Business Profile, across the area you actually serve. A person reads it.
If the ground is winnable, you will be told what stands between you and it. If it is not, you will be told that, and you will not be sold anything.