How to do an online marketing audit: a step-by-step guide

What an online marketing audit is

An online marketing audit is a systematic pass through everything a company uses to win customers online — the website, content, search, paid campaigns, social media and measurement — with a single goal: to find where money and attention are leaking. It is not a list of mistakes for their own sake. It is a comparison between what you believe about your marketing and what the data says about it.

It differs from an ordinary report in two ways. A report describes what happened last month; an audit looks for causes, and it is independent of whoever runs the campaigns. That is why it is done either by someone from outside or at least by a person not responsible for the results of the audited channels — otherwise it easily turns into a defence of the work done so far.

What it is not matters too. It is not a one-off fix, it is not buying a new tool, and it is not a website redesign. The output is a decision-making document: a list of findings with an estimate of impact and effort, from which you can say what you will do this month and what can wait.

When an audit makes sense and how often to run one

The best moment is before you add budget, not after. If you keep raising ad spend and the number of enquiries stands still, another euro will not solve the problem, it will only make it more expensive. An audit fits equally well before a redesign, before changing agency, after replacing whoever owns marketing, or when you notice you are starting not to trust the numbers in your own reports.

How often? For a company spending hundreds of euros a month on online marketing, once a year is enough, plus a short measurement review after every larger change to the website. For an e-shop running daily campaigns, a full audit once a year and a smaller account check each quarter makes sense. More often usually does not — you will not have had time to implement the recommendations and measure their effect.

There is a bad time for an audit too: the middle of a seasonal peak. The data will not be comparable and nobody will have the capacity to act on the findings. The ideal moment is the start of a quieter period, so you can get the changes live before the next wave.

What to prepare before you start

Most of the time in an audit does not go into analysis but into chasing access. Get this part out of the way up front and you save a week. This is the minimum, without which the audit is guesswork:

Step 1: The goal and the numbers you start from

Before you open the first tool, write down one sentence: what marketing is supposed to bring the company over the next twelve months. Not "more traffic", but "forty enquiries a month at up to twenty euros each". Without that sentence an audit is only a list of observations — there is nothing to measure them against.

Then write out four to six numbers that describe this goal, month by month for the last twelve months. Usually they are: organic search traffic, enquiries or orders, conversion rate, ad spend, cost per conversion and average order value. A spreadsheet is enough; it does not have to be a dashboard.

Use those numbers as your only yardstick from here on. When you find a slow website in step two, the question is not "is this slow?" but "how many of those forty enquiries is it costing us?". An audit that cannot get back to the numbers from this step ends up as a wish list.

Step 2: Auditing the website and technical SEO

The website is where every channel meets, so a mistake here multiplies the damage across all of them. Start with what the search engine sees: open Search Console, go through the indexing report and look at how many pages Google knows about, how many it has indexed and on what grounds it excluded the rest. A page that is not indexed cannot bring a single visit, however well it is written.

Then measure speed, on a phone and on a real connection, not on the office wifi. What you care about are the Core Web Vitals: LCP (when the main content appears), INP (how quickly the page responds to a tap) and CLS (whether content jumps around under your finger). An extra second of loading shows up in people leaving before the visitor even sees what you sell.

Finally, walk the site as a customer on a phone, from the ad to the submitted order, and write down every place where you had to stop and think. That list tends to be short and extremely valuable — the mistakes on it are cheap to fix and show up in the conversion rate within two weeks.

Step 3: Auditing content and keywords

Open the Search Console performance report for the last twelve months and sort queries by impressions. That gives you a picture of what Google already shows you for. Now compare it with what you actually sell. The gap between "what people find us for" and "what people pay us for" is the most common and the cheapest finding to fix in the whole audit.

Next, look for three things: pages with many impressions and a low click-through rate (good content with a bad title), queries in positions eleven to twenty (content that needs very little to reach page one), and two of your own pages competing for the same query. The last is called cannibalisation, and it usually comes from good intentions — you wrote about the same topic twice.

Only then think about new content. Rewriting an existing page sitting in position twelve is almost always faster and cheaper than writing a new one that starts from zero.

Step 4: Auditing paid campaigns

With campaigns, do not start from the account structure but from the search terms report. That is exactly where you see what you are really paying for — not the keywords you entered, but the queries the system matched to them. In accounts where the negative list has not been topped up for a while, a fifth of the budget typically goes on queries with no purchase intent at all.

Then break the results down by campaign and compare cost per conversion with the value a conversion brings you. A campaign with a nice cost per click and an expensive conversion is a worse deal than one that looks expensive but brings orders. And be careful with automated bidding strategies: they only work when they have something to optimise for, meaning when conversions are measured correctly — which you only verify in step six.

Do not forget overlap. If three campaigns target the same audience, you are competing with yourself and pushing your own prices up. The same goes for branded queries: you would win some of them for free from organic results and some you would not — and the difference can be measured by switching them off briefly.

Step 5: Auditing social media

With social media the most useful question is also the most uncomfortable one: how much time a month do they cost you, and how many people arrive on the site from them. Count the first honestly, including content preparation; you will find the second in Analytics among referral sources. For most small companies the answer is that two channels done properly bring more than five done out of obligation.

Judge the content by what it does, not by likes. Sort the last six months of posts into three groups — reach, interaction, clicks through to the site — and look at what the best ones have in common. It is usually the format and the topic, not the posting time, even though posting time is what gets written about most.

And check the basics everyone forgets: whether the profile is complete, whether it links to the right address, whether the opening hours are correct, and whether anyone answers messages. An unanswered message from three weeks ago costs more than a whole month of content.

Step 6: Auditing measurement and analytics

This is most often the weakest link, and the one that devalues every step before it. If the measurement lies, so do the decisions built on it. So it is worth running one simple test: take the number of orders from accounting for last month and compare it with the number in Analytics and with the number in Google Ads. If they do not agree to within ninety percent, you have the first priority of the whole audit.

The differences usually have dull causes: the conversion is counted on a thank-you page anyone can reach directly from Google; internal traffic is not excluded; consent mode blocks part of the measurement; or the same event is wired up twice. Each of them is fixed in hours, not weeks — you just have to know about it.

Finally, check whether you are measuring what the business actually cares about. A surprising number of accounts measure clicks on a phone number and miss a submitted form entirely, or count every conversion the same, even though one is worth ten euros and another two thousand.

Step 7: Competitors and the market

We do not audit competitors in order to copy them, but to find the gap. Take three to five companies you actually meet in tenders — not the biggest names on the market — and look at which queries they show up for, which topics they have covered, and how they communicate price, delivery and guarantees.

Look for two things: a query everyone covers except you (that is your nearest piece of work), and a query nobody covers (that tends to be a cheap opportunity, if anyone is really searching for it). A frontal attack on a query with five well-funded players is usually the most expensive route available.

And compare yourself outside search as well: how fast they respond to an enquiry, what is on their pricing page, what their first email looks like. Part of a competitive advantage is not created in marketing but just behind it — and an audit brings that to the surface.

How to rank the findings by impact

After seven steps you will have forty to a hundred observations. Unsorted, that is a list nobody will act on. So put every finding through three questions: how much money it costs or brings, how large the expected effect of the change is, and how long the fix will take. A three-point scale is enough; nobody expects accuracy to the euro here.

Then split the findings into four groups: quick wins (large effect, little work), projects (large effect, lots of work), maintenance (small effect, little work), and the ones you will not do (small effect, lots of work). Do not delete that last group from the report — a written reason why something will not be done saves repeated discussions later.

Spend the first two weeks after the audit exclusively on quick wins. They create both the room and the trust for the larger projects to go through — and since their effect tends to be measurable, they hand you a number to argue with.

What to do with the results

An audit that ends up as a PDF attached to an email has no value. So settle three things right at the end: who owns each finding, by when it should be resolved, and which number will tell you it worked. That number has to be one of the ones you wrote down in the first step.

Agree a check-in six to eight weeks later, too. Not another audit, just half an hour over the same spreadsheet. That is when you find out whether the changes worked and whether the measurement you fixed now says the same thing as accounting.

And expect some of the findings to be unwelcome — a channel you have invested in for years may come out as the weakest. That is fine. The point of an audit is not to confirm the decisions already made, but to make the next ones cheaper.

When to run the audit yourself and when to call someone in

You can handle most of this guide on your own, and we recommend it — someone who has once been through their own numbers never looks at marketing the same way again. If you have one website, one ad account and somebody who understands them, your own audit will give you eighty percent of the value for the price of two days of work.

Calling someone in from outside pays off when at least one of these holds: the campaigns are run by a supplier whose work you are meant to assess; the findings are to serve as the basis for a decision about budget or about changing agency; or you simply do not have the two days it takes in the team. Independence is not a slogan — it is the only way to avoid an audit that confirms whatever its author already believed.

However you do it, keep to the order of the steps. The most common mistake in a self-run audit is not a lack of knowledge, but starting with the campaigns and never getting as far as measurement — the part that would have given the rest of the audit its meaning.

Frequently asked questions