Analytics & automation

Analytics only matters if someone uses it to decide something. Automation only helps if the process underneath it is sound. My work in this area connects the two: measure the actions that matter, report them in a way people read, and automate the follow-up that your team keeps doing by hand.

Start from the decisions, not the tools

Most analytics projects begin with a tool and end with a dashboard nobody opens. I begin with a list of the decisions the business actually makes: where to spend the budget, which channel to cut, which product to promote, whether a campaign worked, whether sales is following up.

For each decision I ask what information would change the answer. That list becomes the measurement plan. Anything that does not feed a decision is optional, however easy it is to collect.

This also keeps reports short. A one-page view that answers five questions is more useful than forty charts.

Tracking: events, definitions and cleanliness

In GA4, everything is an event, and the useful work is deciding which events represent business actions: a form submitted, a call clicked, a booking made, a purchase completed. Each gets a clear name, the right parameters, and a marking as a key event where it matters.

I implement through Google Tag Manager so that changes are controlled and documented, and test each event in preview mode before it goes live. I also check for common faults: duplicate firing, internal traffic included, missing cross-domain handling, consent settings that block data without anyone noticing.

Where the website platform makes tracking awkward, I note the limits and suggest what is practical.

A short naming convention saves trouble later. Events, parameters and campaign tags should follow a consistent pattern so that anyone reading a report understands what it means. I agree the convention with you before building, and record it in the tracking plan. It sounds minor, but inconsistent names are one of the main reasons old data becomes unusable.

• A tracking plan listing every event, its trigger, parameters and purpose.
• Consistent UTM conventions so campaign sources are not scattered across spellings.
• Conversion tracking linked to ad platforms where used.
• Consent and privacy settings reviewed against your market and legal advice.
• Documentation so the next person understands what was built.

Auditing what you already have

If tracking exists, the first step is an audit instead of a rebuild. I compare what is being collected with what you want to know, and list the gaps, duplicates and unclear definitions. Often there is more data than expected, and none of it is trusted.

The outcome is a short list of fixes in priority order, with an explanation of which past reports may have been affected. I will not tell you that a historic number was correct if I have reason to doubt it.

This audit is usually the cheapest useful starting point, and a natural companion to a broader digital audit if wider issues are suspected.

I also look at who actually uses the data today and for what. A tracking setup can be technically sound and still ignored. Talking to the people who open the reports, or avoid them, tells me whether the issue is accuracy, relevance or habit, and the fix is different in each case.

Reporting people will read

A report should open with what changed and what it means, not a wall of metrics. I build reporting around a small set of numbers tied to the business goals, with context: compared to the previous period, to the same period last year where it is available, and to the plan.

Tools like Looker Studio work well for most small and mid-sized businesses, connecting GA4, ad platforms, search data and spreadsheets. The format matters less than the habit: a regular time when the people who decide actually look at it.

Where a number can be misread, I add a note. Attribution, for instance, will never be perfect, and pretending otherwise leads to bad decisions.

I also suggest a standing review meeting, even a short monthly one, where the report is discussed and actions are recorded. Reports that nobody discusses become wallpaper. The conversation is where numbers turn into decisions, and where odd results get questioned before they are repeated as fact.

Automating the follow-up

The second half of the service is the work that happens after the data. Enquiry arrives; who sees it, how quickly, and what is recorded? Customer buys; what is sent next? Lead goes quiet; who nudges them?

I map the current process first, on paper, with the people who do it. Then I identify the repetitive and error-prone steps, such as copying form entries into a CRM, assigning leads, sending confirmations, creating follow-up tasks. These are good candidates for automation. Judgement-heavy steps usually are not.

Implementation uses whatever suits your stack: native CRM workflows, email platform automations, or connecting tools through an integration service. I prefer the simplest option that is reliable and that your team can maintain.

Mistakes I see most

Automating a broken process just makes it fail faster. If leads are not being followed up because nobody owns them, an automated email will not change that. Another is over-engineering: a dozen interlinked automations that only the person who built them understands.

A third is trusting numbers that nobody has tested. I always verify a tracked conversion by completing the action myself and following it through to the report and the CRM.

Another common mistake is collecting personal data without a clear reason or legal basis, which creates risk and gives little benefit. I recommend collecting what you need and no more, and being clear on who can access it. Good practice here protects your customers and also keeps your own reporting simpler and easier to defend.

Privacy, consent and data quality

Collecting data comes with responsibility. Depending on where your visitors are, consent rules may restrict what you can track and when. I set up consent handling in line with your legal advice, and I explain what data is lost when visitors decline so the reports are interpreted correctly.

Data quality is a continuing task. Sites change, forms are replaced, tags break. I recommend a periodic check, even a simple monthly one, that confirms key events still fire and numbers still reconcile with your CRM or sales records. Tracking left alone for a year is usually wrong in at least one place.

Connecting systems without making a tangle

Most businesses use several tools: a website, forms, a CRM, an email platform, an ad account, a spreadsheet. I connect them where it removes real manual work, and I document each connection, what it does and who owns it.

I prefer fewer, well-understood links over many clever ones. If the tools offer a native integration that does the job, I use it. If a custom workflow is needed, it goes in a documented place with notifications when something fails, because a silent failure in an automation can cost leads for weeks before anyone notices.

What I need and when this is not the right move

I need access to your analytics, tag manager, ad accounts and the tools you want to connect, plus the people who run the current process. I also need an owner on your side for any decision about data or customer communication.

If your site has very little traffic and few enquiries, extensive automation is premature; a clean tracking setup and a simple enquiry workflow will be enough. If the main problem is the offer or the page, start with conversion optimisation instead.

Measure what you will act on, and automate only what you understand.