Book a grown-up conversation
Tracking

How do we know what’s working and what needs to be fixed?

You track, you analyze, you optimize. Set tracking up before you launch, or you’ll find out in month three that the numbers were never there.

You can’t improve what you don’t measure

Digital marketing is a performance discipline, like sports. The goal is to get better, faster and cheaper over time, and you can’t do that without the numbers. The first version of any campaign rarely works the way you planned. That’s normal. What separates good programs from bad ones is what happens next.

Google’s former chairman Eric Schmidt put it simply: the internet would transform advertising because of its trackability, not its beauty. For the first time, you could see the steps between the ad and the sale.

Track, analyze, optimize

That is the core loop of digital marketing. Track what happens. Analyze what it means. Optimize what you do next. Most marketing barely does the first step, which is why so much of it can’t say what worked.

Set it up before you spend

Every place you send people needs tracking that works. The pixels from Google, Meta, LinkedIn and every other platform you advertise on have to fire. Every link needs UTM codes, so you know which channel and campaign each visit came from. Form submissions, calls and bookings have to register as conversions.

Then someone has to check. Don’t take “it’s set up” on faith. A campaign that runs for three months without accurate tracking is three months you can’t learn from.

Pick the “key” in KPI

Digital marketing produces an avalanche of numbers: cost per thousand impressions, click-through rate, cost per click, cost per lead, cost per acquisition, engagement rate, cart abandonment, average order value. It’s easy to drown.

Baseball tracks hundreds of statistics, but to know who’s winning you need three: the score, the inning and the outs. Your marketing needs the same discipline. Find the two or three key performance indicators that tell you whether you’re winning.

Acquisition, behavior, outcomes

A framework from Avinash Kaushik, Google’s former analytics evangelist, makes it simple. Pick one KPI for each stage. Acquisition: how efficiently you bring people in, with cost per acquisition as the measure every business should know. Behavior: what they do once they arrive, like engagement rate, pages viewed or emails opened. Outcomes: whether they convert in a way that supports the business goal.

These map onto reach, depth and relationship from the Tactics page. One KPI per stage is enough to steer by.

Know what a customer is worth

Cost per acquisition only means something next to price. Spending $50 to sell a $10 product makes no sense. Spending $20 to sell a $100 product, or to win a customer who buys every month for years, can be a bargain. Know the value of a customer before you judge what it costs to get one.

A metric is not an analytic

A metric is about tracking: a count of what happened. An analytic is about analysis: putting many sources together to understand how people actually move from first touch to purchase.

There is no single source of truth. Google Analytics, Meta and every other platform use different data models and each tells a slightly different story. The skill is in reading them together.

Find the O-Ring

The space shuttle Challenger was a complicated system that failed because of one small rubber seal. Marketing funnels work the same way, because systems run on multiplication. One step at zero, whether a button hidden below the fold on mobile, a form that doesn’t submit or a phone nobody answers, zeroes the whole chain. Fix the weakest step first.

Testing lives inside tracking

Once you can see what’s happening, you can test: version A against version B, one change at a time. Fail fast, learn faster. The first campaign rarely works as planned. Because you had a plan, you have something to compare against and adjust.

If only one visitor in a hundred buys, a sale alone gives you too little data. Track the steps before it, like a page visit, a sign-up or a request for information, the way a shoe store would value everyone who walks in, not just the people who buy.

Tracking when AI answers first

More buyers now get their answer from an AI tool without clicking anything. You can’t always follow someone through the funnel the way you once could. So track whether you are named and cited, and watch branded search and direct visits, alongside the clicks you can still measure.

Common questions

Why is this decision called Tracking and not Testing?
Because you can’t test what you aren’t tracking. Tracking covers the whole loop, track, analyze and optimize, and testing is one of the ways you optimize inside it.
What is a UTM code?
A short tag added to a link that tells your analytics which source, channel and campaign a visit came from, so you can compare what each one produces.
What should we track if very few visitors buy?
The steps before the sale: key page visits, sign-ups, downloads, calls and booking requests. They give you enough data to improve the funnel long before purchases do.
Can you track how AI talks about your brand?
Yes, by checking which AI tools name and cite you for the questions your buyers ask, and by watching branded search and direct traffic, which can rise when AI tools recommend you.
What KPIs should we track?
One for each stage of the journey: an acquisition KPI like cost per acquisition, a behavior KPI like engagement rate, and an outcome KPI tied to the business goal, like booked calls or revenue.
What is cost per acquisition?
What you spend in marketing to win one customer or conversion. It only means something compared with what that customer is worth.

Find your difference with your leadership team in the room.

See the Brand Truth Workshop Or get a free AI Brand Scan