Lay down clean, compliant tracking
Without a tracking plan, your data is noise
People often install an analytics tool and hope "it measures". In reality, without prior thought, you collect a jumble of badly named events, conversions counted twice, lost acquisition sources. A month later, you can't answer a simple question, because the data wasn't designed. The tracking plan is that document — often just a spreadsheet — that decides in advance what to measure, how to name it, and where.
It's the least glamorous and most profitable step of the whole data stack: ten lines written before installing are worth a thousand corrections after.
The tracking plan, concretely
A tracking plan lists the events that matter and their properties, with a consistent naming convention:
| Event | When it fires | Properties |
|---|---|---|
page_view |
Page load | url, title |
signup |
Account created | source, plan |
purchase |
Payment validated | amount, product, currency |
trial_started |
Free trial begins | plan |
subscription_canceled |
Churn | reason, tenure |
Two naming rules: be consistent (always purchase, never purchase then buy then order for the same thing) and stay simple (a short, explicit name beats an obscure code). A clear convention from the start saves you months of uninterpretable data.
Attribution: where do your customers really come from?
Attribution answers "which channel deserves credit for this sale?". It's crucial: without it, you risk cutting the budget of the channel that brings your best customers. The basic tool is universal and free: UTM parameters.
A tagged link looks like this:
https://yoursite.com/offer?utm_source=newsletter&utm_medium=email&utm_campaign=june-launch
utm_source: the source (newsletter, google, instagram…)utm_medium: the channel type (email, cpc, social…)utm_campaign: the specific campaign
Systematically tag all your outbound links (emails, social posts, ads). Your analytics will then group visits and sales by source, and you'll finally know which channel actually works. Without UTMs, everything ends up in an unusable "direct / other" mush.
Beware the nuance: last-click over-credits the end-of-journey channel (often brand search) and ignores those that sparked discovery. For a solo, last-click is enough to start; you just need to know it underestimates the top of the funnel.
GDPR is not optional
In Europe, collecting browsing data is governed by GDPR and the ePrivacy directive. The principles every entrepreneur must respect:
- Prior consent for non-essential trackers (most analytics and advertising cookies). Concretely: a banner that lets users accept or refuse just as easily, not an "accept all" banner with no alternative.
- Purpose and information: clearly state, in a privacy policy, what data you collect and why.
- Minimization: only collect what genuinely serves a decision.
- People's rights: allow access, correction, and deletion of data.
Consent-management tools (CMP — Consent Management Platform) like Axeptio, Cookiebot, or tarteaucitron (open source) automate the banner and block trackers before consent.
The privacy / measurement trade-off
The more consent you ask for, the less you measure (many visitors refuse). Hence the value, seen in chapter 2, of cookieless analytics like Plausible or Matomo configured without personal data: depending on their setup and the relevant authority's guidance, some can be used without a banner, which restores full measurement while respecting privacy. It's often the best trade-off for an entrepreneur: a little less depth than GA4, but 100% of traffic measured and simple compliance.
graph TD
A[Need to measure] --> B{Analytics with cookies<br/>GA4, ads}
B --> C[Consent banner required]
C --> D[Some traffic refuses<br/>= partial data]
A --> E{Cookieless analytics<br/>Plausible, Matomo}
E --> F[Often no banner<br/>= full traffic measured]
The entrepreneur's tracking checklist
Before considering your measurement "clean": a written tracking plan with a naming convention, UTMs on all your campaign links, a CMP in place if you use consent-bound trackers, an up-to-date privacy policy, and an end-to-end test (make a real conversion and check it shows up correctly in the tool with the right source). Once these five points are cleared, your numbers become trustworthy — and reliable data is the condition for everything above.