Digital Marketing Audit: A 7-Step Guide You Can Run Yourself

A free 30-minute marketing audit is not an audit. It is a sales call with a spreadsheet attached. By the time the agency tells you what is wrong, they have already decided what they want to sell you to fix it.

A real audit starts the other way round. You write down what you are trying to achieve. Then you look at what your marketing is actually doing. The gap between those two is the audit. Everything else is decoration.

This guide is the seven steps Mira runs when we audit a business' digital marketing for the first time. You can run it yourself with a couple of free tools and a quiet afternoon. If you do, you will know whether you have a marketing problem, a measurement problem, or a strategy problem before you let anyone sell you a fix.

Why most audits are theatre

Three reasons.

The first is the free-audit format. To pack the diagnosis into 30 minutes, the auditor has to skip the bit where they understand your business. They look at the channels in isolation. Your Google Ads account looks "underspent". Your SEO is "missing key terms". Your social is "not posting often enough". All of these can be true. None of them tells you whether your marketing is working.

The second is the channel-by-channel structure. Most audits report channel performance against channel benchmarks. Your CPC is low. Your CTR is high. Your impression share is up. The problem is that channel benchmarks have no opinion about your business. A 1.5% conversion rate is good for a £900 sofa and catastrophic for a £20 candle. Without the business context, the numbers are just numbers.

The third is the absence of a position. A useful audit tells you what to stop doing. Most audits cannot bring themselves to do this because stopping work is not a service the auditor wants you to buy.

Mira's audits take a position. So should yours, even if you run it yourself.

Step 1: Write down what you are trying to do

Before you open a single analytics tab, write down the answers to three questions.

What does success look like in 12 months? Be specific. "Grow the business" is not an answer. "£1.4m revenue, 30% gross margin, 60% of revenue from repeat customers" is.

What is the one number you would change if you could only change one? For most businesses this is either revenue, gross margin, customer acquisition cost, or repeat purchase rate. Pick one.

What are the constraints? Budget, headcount, time, technical debt, supplier capacity. The audit cannot recommend things that the business cannot afford or operate.

The point of this step is to anchor everything that follows. When you look at a channel, you are not asking "is this channel performing". You are asking "is this channel moving the one number that matters, given the constraints". That is a much sharper question.

Step 2: Audit the measurement layer first

This is the step everyone wants to skip. Don't.

You cannot audit a marketing programme through a broken measurement layer. If your conversion tracking is wrong by 30%, your channel report will be wrong by 30%, and your audit will recommend the wrong things.

The four checks:

GA4 is collecting cleanly. Open Realtime, trigger a key event from your own machine, see it appear. Look at the last 30 days of conversions and check the trend looks plausible. If you have a sudden cliff or a sudden spike that does not match a business event, something is broken.

Conversion definitions match the business. A conversion in GA4 should be the action that closes the customer, or the action that reliably predicts they will close. For ecommerce, that is "purchase". For lead-gen, that is "form submission" minus the spam. If you have ten conversions defined and none of them is the one that pays the bills, you are reporting on noise.

Server-side tracking is in place where it matters. For ecommerce sites running paid social, the Conversions API (CAPI) is no longer optional. If you are still relying on browser pixels, your match quality is degraded and your channel reports overstate cost per acquisition by 15 to 40%.

Search Console is connected and clean. Open Performance. Look at the last 90 days. If impressions are flat and clicks are dropping, your SEO has a problem. If the property is not verified, fix that first.

If any of these checks fail, stop. Fix the measurement layer before continuing the audit. Anything else you find before then is provisional at best.

Step 3: Channel mix and where the money actually comes from

Once measurement is clean, open GA4 → Acquisition → Traffic acquisition. Look at the last 90 days.

Two things to note.

First, the percentage of revenue by channel. Most businesses are surprised by the answer. Many ecommerce sites who think of themselves as "Meta-led" are actually 60% organic and direct, with paid social acting as a brand amplifier rather than a primary acquisition channel. Many lead-gen businesses who think of themselves as "SEO-led" are actually pulling most of their qualified leads from referral and direct.

Second, the cost-to-revenue ratio by channel. If you are spending £8k a month on Google Ads and it is producing £12k of attributable revenue, that is not "working". A £4k margin on £8k of spend is not paying for the agency, the platform fees, or the operator's time. The channel is alive but not load-bearing.

Write down two things for each channel: percentage of revenue contribution, and unit economics. If both are weak, the channel is a candidate to be cut. If one is strong and one is weak, the channel needs a brief, not a budget change.

Step 4: Paid media health

If you run any paid media (Google Ads, Meta, LinkedIn, TikTok), this is where most audits spend 80% of their time. We spend about 30%, because by step 3 you already know whether the paid channels are pulling their weight at the business level. Now you are just trying to work out why.

Five checks, ordered by how often they catch something.

Account structure. A Google Ads account with 70 campaigns and no naming convention is a red flag. Most accounts have 2 to 5 campaigns doing 90% of the work and the rest are noise. Pull a campaign-level report sorted by conversions over 90 days. The shape will tell you whether the account is being managed or just kept on.

Wasted spend. In Google Ads, look at the search terms report. Filter to terms with cost over £50 and zero conversions. If the list is short, the negative-keyword work is being done. If the list is long, you are paying for unrelated searches. In Meta, look at audience overlap. If three of your active campaigns are targeting overlapping audiences, you are bidding against yourself.

Quality Score (Google) or relevance metrics (Meta). In Google Ads, a Quality Score below 6 on a high-spend keyword is costing you 30 to 50% more per click than you should be paying. In Meta, Quality Ranking, Engagement Rate Ranking, and Conversion Rate Ranking should all sit at average or above on any campaign with meaningful budget. If creative is the suspect (more often than not, post-Advantage+), the Meta Ads Library is the fastest free read on what your competitors are running and how your own work stacks up.

Bid strategy. Manual CPC on a Google Ads account that has more than 50 conversions a month is almost always leaving money on the table. Target CPA or Maximise Conversions with a target are better defaults. If the account is on Manual CPC, ask why.

Cost per acquisition vs lifetime value. This is the killer check. If your CPA is £80 and your average customer lifetime gross profit is £120, the channel is alive but the unit economics are thin. If your CPA is £80 and your LTV is £400, the channel is a growth engine. The audit's job is to surface this number, not just the CPA.

For more on what a good cost per click actually looks like in your category, and how to think about marketing efficiency ratio (MER), we have walked both of those out in their own posts.

Step 5: SEO baseline

Open Search Console.

Performance over the last 12 months. Is the trend up, flat, or down? A flat trend on a growing business is a problem. The internet got bigger; your share of it did not. A downward trend is an emergency.

Top 10 pages by clicks. Note them. These are the pages doing the work. If your top 10 includes a single product page and nine blog posts, your SEO is informational, not commercial. That is fine if it is intentional and a problem if it is not.

Top 10 queries by impressions, not by clicks. Impressions tell you what Google thinks you are about. Clicks tell you what users do once they see you. If you have a query at 50,000 impressions and 200 clicks, the SERP is showing you and users are not clicking. Either the title and meta need work, or the page does not match the intent.

Pages in positions 11 to 20. These are your quick wins. A page at position 14 for a 500-volume keyword is one good refresh away from page one. Most SEO programmes ignore them in favour of net-new content. Do not.

Coverage report. Look at "Indexed" and "Not indexed". A "Not indexed" count higher than your indexed count is a technical problem worth fixing. Common causes: duplicate content, thin pages, accidental noindex tags, blocked-by-robots.txt directories.

If your SEO baseline shows a flat or declining trend and you have not done any technical or content work in 12 months, that is the explanation. If it shows the same pattern despite consistent work, the work is the problem. We have written more on our SEO services if this is the bit that needs attention.

Step 6: Content audit

Pull a list of every page on the site that has been published in the last 24 months. For each, three columns: monthly organic traffic, last update date, conversions attributed (any goal).

Sort by traffic, descending.

You will see the same pattern almost every time. The top 20% of pages pull 80% of the traffic. The bottom 50% pull less than 5% combined.

Three actions:

Top 20%. These pages keep the lights on. Refresh them every 12 months. Add internal links to them from new content. Make sure they convert (a CTA, a relevant lead magnet, a phone number).

Middle 30%. These are candidates for consolidation. If you have three blog posts on related topics and none of them is ranking, consider merging them into one stronger piece and 301-redirecting the others. Topical authority is a function of depth, not page count.

Bottom 50%. These are candidates for noindex or deletion. A page that has not had a visit in 12 months is dragging down the site's average quality signal. Be honest. Cut.

This is the audit step most businesses cannot bring themselves to do. They published the content; they want it to count for something. It does not. The bottom 50% is hurting you.

Step 7: Creative and brand consistency

This is a soft step, but it is the one that pays the biggest compounding return.

Open your homepage, your top three product or service pages, your most-viewed blog post, your latest Meta ad, your latest Google ad, and your latest email. Lay them out next to each other.

Three questions.

Do they sound like the same company? Not "do they use the same logo". Do they make the same promise to the same kind of customer, in language that has the same shape?

Do they say the same thing twice? Many programmes have a different "main message" on every surface. The homepage says one thing, the ad says another, the welcome email says a third. The customer cannot consolidate the message. They will leave with no clear idea of what you do.

If a new customer landed on each surface in turn, would they make a buying decision faster, or slower? The honest answer here is what the audit produces.

Most "rebrand projects" are an over-correction for a problem that was actually inconsistency, not bad design. If you can fix the consistency without redesigning the brand, do that first.

What to do with what you find

The output of an audit is not a report. It is a list of decisions.

The format we use, and recommend you use:

Stop. What are you doing today that should be cut? Be specific. "Stop running the Google Display campaign that has cost £4,200 over 90 days and produced 1 conversion." Not "review display strategy".

Continue. What is working? Most audits skip this; it is the most important section. If you do not name what is working, you risk fixing it.

Start. What is missing? "Start a structured retargeting campaign in Meta segmented by funnel stage." "Start refreshing the seven posts in positions 11 to 20 of Search Console." "Start measuring lifetime value at the channel level."

A useful audit produces between three and eight items across those three buckets. If your list is longer than that, you are not auditing, you are wishlist-building.

When to bring in an agency

You can run this audit yourself. We have laid it out so you can.

The work that is harder to do yourself is the part after the audit. Translating "the Google Ads account needs restructuring" into "here is the rebuilt account, launched and bedded in over six weeks" is the bit that requires a team. Same for "the SEO programme needs a refresh strategy" turning into 30 refreshed posts that move impressions and clicks.

We work with brands that have run the audit (theirs or ours), know what is broken, and want a partner to fix it. If that is where you are, get in touch. We will tell you whether we are the right fit before we talk about scope.

A closing thought

Audits look like work. They are not. Audits are the cost of admission to good work. The audit is what makes the next twelve months coherent.

If you finish your audit with a clear stop, continue, and start list, you have done it correctly. If you finish with a 40-page report and no clear next move, you have just paid for a sales call. Mira-run or self-run, the test is the same.