These three are sold interchangeably and produce completely different things. An audit inspects a site against a standard and reports what’s wrong. An analysis investigates one specific question and reports why something is happening. A strategy decides what to do and in what order. Buying one while expecting another is the most common reason businesses feel an SEO engagement didn’t deliver.
In plain English
Think about a building you own.
A survey is somebody walking the whole building against a checklist. Roof, wiring, damp, structure. You get a report listing what’s wrong and how serious each thing is. The surveyor fixes nothing and doesn’t tell you what to do with your budget.
An investigation is different. The top floor is cold and nobody knows why. Somebody comes and works out why. They aren’t checking the wiring, because that isn’t the question.
A renovation plan is different again. You now know what’s wrong. The plan decides what gets done, in what order, with what money, and what you’re prepared to live with.
All three are legitimate. None of them substitutes for another. And the biggest waste of money is commissioning a survey when what you needed was an investigation.
Why this matters for your site
The commercial version of this problem is specific and expensive.
A business loses traffic and buys an audit, because that’s what agencies sell. The audit comes back with 140 findings, most of them true and none of them the cause of the drop. Three months later they’ve implemented forty recommendations and traffic is unchanged, because the actual cause was a redirect map that broke during a migration and no checklist was going to surface it as the priority.
The opposite happens too. A business with no idea of its condition buys a strategy. The strategy is well written and rests on assumptions nobody checked, because nobody looked at the site first.
Matching the work to the question is most of the value.
The three, precisely
An audit
The question it answers: what condition is this site in?
What it looks at: the whole site, against a standard. Technical health, indexing, structure, content, links, and how it compares to competitors.
What it produces: a list of findings, each with a severity. Google’s guidance adds a useful test: an audit should give realistic estimates of improvement and an estimate of the work involved. A list of problems with no sense of what fixing them is worth or costs is only half an audit.
When you need one: you’ve inherited a site, you’re considering a rebuild, you suspect general neglect, or you’re deciding whether to invest at all.
An analysis
The question it answers: why is this specific thing happening?
What it looks at: whatever is relevant to that one question, and nothing else. It might be very narrow and very deep.
What it produces: an answer, the evidence behind it, and usually a recommendation about the specific thing.
When you need one: something changed and you don’t know why. Traffic fell. A competitor overtook you. One section stopped performing. Sales dropped while traffic held.
The tell: you can state the question in one sentence. If you can, an audit is probably the wrong purchase.
A strategy
The question it answers: what should we do, in what order, and why that order?
What it looks at: your business goals, your resources, your market, and whatever you already know about the site’s condition.
What it produces: a sequenced plan with priorities, owners, and a rationale for what isn’t being done.
When you need one: you already know what’s wrong, or you’ve just had an audit, and you need to turn a hundred possible actions into a plan somebody can execute.
How they fit together
The ordinary sequence is audit, then strategy, with an analysis whenever a specific question comes up.
But the sequence isn’t compulsory, and forcing it wastes money. If you know exactly what’s wrong, skip to strategy. If you have one specific question, an analysis is faster and cheaper than an audit that will contain the answer somewhere on page thirty.
The one thing that should never be skipped is knowing which one you bought.
What this looks like in practice
A retailer’s traffic falls 30% over a month. They approach three providers.
The first proposes an audit. The second proposes an audit and a strategy. The third asks when the drop started, what changed on the site around that date, and whether the Manual Actions report is clean, then proposes a two-week analysis of the drop specifically.
The third is the right shape for the situation. If the analysis finds a single technical cause, the audit was never needed. If it finds general decline, an audit is the sensible next purchase and it’ll be better scoped for having done the analysis first.
What people usually get wrong
“An audit will tell us why traffic dropped.” It might, incidentally. It isn’t designed to. An audit surveys condition, it doesn’t trace a specific event.
“A strategy is just a longer audit.” An audit describes. A strategy decides. Deciding requires knowing your budget, your team, and what your business is actually trying to do, none of which appears in a site inspection.
“More findings means a better audit.” A hundred and forty findings ranked by nothing is a way of transferring the hard decision back to you.
“We had an audit last year, so we know the situation.” Sites and competitors change. An old audit is a historical document.
What you can do about it
Before buying anything, write down the question you want answered in one sentence. That sentence usually names the right piece of work.
Ask what the deliverable is, literally. A document listing problems, a document answering a question, and a document setting out a plan are three different things.
Ask whether prioritization is included. It often isn’t, and it’s the part that makes the rest usable.
Agree how the result will be judged before the work starts, and take a baseline reading at the same time. All three pieces of work are easier to evaluate against a number recorded beforehand than against memory.
At the audit stage, grant read access to Search Console only, not write access. This is Google’s own recommendation.
Google’s guidance on hiring an SEO covers what to ask and what to be wary of, and takes about five minutes to read.
Words you will hear
Scope. What is and isn’t included in a piece of work. The most important paragraph in any proposal.
Deliverable. The actual thing you receive at the end.
Prioritization. Ranking findings by impact and effort so somebody can act on them.
Baseline. A record of the current state, taken before work starts, so change can be measured.
Discovery. The initial period where a provider learns your business before recommending anything.
For which numbers to judge the work by, see How SEO is measured. For the causes an analysis usually investigates, see Why website traffic drops.
