Marketing analytics: measuring what your marketing actually does
Most marketing reporting measures activity, not impact. It counts clicks, sessions, and impressions because those are easy to pull, then quietly avoids the harder question of what actually drove revenue. Analytics done right answers that question, or at least narrows it honestly.
This page is a working reference, not a sales pitch. It covers how measurement breaks in the real world: attribution that lies, tracking that silently stops firing, and dashboards full of numbers nobody acts on. If you want the paid-media companion, see the PPC field guide; for organic, the technical SEO guide.
The honest truth: perfect attribution does not exist, and any tool that promises it is selling something. The goal is not certainty. It is a measurement setup trustworthy enough that you can make better decisions than the person guessing.
Attribution: giving credit where it is due, imperfectly
Attribution is how you assign credit to the touchpoints that led to a conversion. The default in most accounts is last-click, which hands all the credit to the final step and ignores everything that warmed the customer up. That is why last-click quietly kills top-of-funnel channels: they never get credited, so they look like waste, so they get cut, so the pipeline dries up a quarter later. The fix is not a magic model, it is understanding what each model over- and under-values. Start with how attribution models work in GA4, and for paid specifically, attribution for paid search.
GA4 and event-based measurement
GA4 replaced sessions-and-pageviews with an event model, which is more flexible and less intuitive. The common failure is treating it like the old Universal Analytics: importing the same reports, ignoring events, and missing the conversions that matter. Set up the handful of events tied to real business outcomes, mark the right ones as key events, and stop drowning in the rest.
Server-side tracking and the cookieless shift
Browser-based tracking is degrading: third-party cookies, ad blockers, and privacy features quietly erase a growing share of your data. Server-side tagging moves measurement off the browser and onto your own server, which is more durable and gives you control over what gets shared. It is more work to stand up, but for accounts that depend on ad-platform signal it is becoming the practical foundation. See server-side tracking in a cookieless world.
First-party data: the audience you actually own
Rented audiences on ad platforms get more expensive and less reliable every year. First-party data, the contacts and behavior you collect and own, is the asset that survives a platform changing its rules overnight. Building it is a measurement and a strategy problem at once. See building a first-party data asset.
Marketing mix modeling and incrementality
When user-level tracking is not available or not trustworthy, marketing mix modeling estimates each channel's contribution statistically, from spend and outcomes over time, without tracking individuals. It used to require an enormous budget and a data-science team. It no longer does. See marketing mix modeling on a small budget. The companion question is incrementality: not what a channel gets credited for, but what would not have happened without it.
What to actually measure
Most dashboards track what is easy, not what matters. Tie every reported metric to a decision someone will make because of it. If a number would not change any action, it is decoration. Measure the few things that map to revenue and retention, and treat the rest as diagnostics you pull only when something looks wrong.
Frequently Asked Questions
What is the best attribution model?
There is no single best model. Last-click is simple but starves top-of-funnel; data-driven and position-based models spread credit more fairly but need enough conversions to be stable. The right choice depends on your sales cycle and volume. The point is to understand what your model over- and under-credits, not to find a perfect one.
Do I need server-side tracking?
If your business depends on ad-platform performance and you are losing signal to cookie loss and ad blockers, server-side tracking is worth the setup. For a small site with light tracking needs, standard GA4 is usually enough to start.
Can I measure marketing without cookies?
Yes. First-party data, server-side tagging, and marketing mix modeling all work without third-party cookies. The cookieless shift makes owning your own data and measuring at the channel level more important, not impossible.
Why do my analytics and ad-platform numbers never match?
They never will exactly. Different systems attribute conversions differently, count across different windows, and lose data at different points. Pick one source as your system of record for decisions and use the others as directional checks.