measuremy.site

Attribution: what it is and why it is hard

Close-up of a signpost in New South Wales pointing to global cities like New York and Seoul.
Photo by Dr Jorge Reyna on Pexels

Attribution is the process of deciding which marketing touch deserves credit when someone does something valuable on your site. That something might be a purchase, a signup, a download, a demo request, or reading a key page. For a small site, attribution usually comes down to one question: which link, campaign, search result, social post, or email brought this person here?

The idea sounds simple. In practice it gets messy quickly, because people rarely arrive once and convert on the spot.

What attribution actually means

Attribution assigns value to a touchpoint. A touchpoint can be an organic search result, a paid ad, an email link, a social post, a referral from another site, or a direct visit.

Most analytics tools answer attribution by looking at the source of the current session. If someone arrives from a UTM-tagged link and signs up, that link gets the credit. If they arrive from Google, the credit goes to organic search. If the source is unknown, it often lands in direct traffic or unassigned.

For small sites, the most common model is last-click attribution. The last source before the conversion gets all the credit. That is useful but incomplete.

Why attribution is hard

Several things break the simple story.

People rarely convert on the first visit. Someone may click an ad, leave, search for your site later, read a few pages, then sign up from a bookmark or a direct visit. The last click may be direct, but the ad did some of the work.

Attribution models disagree. First-click, last-click, linear, and time-decay models spread credit differently. The same set of visits can produce very different answers depending on the model you choose.

Cookies are less reliable than they used to be. Many browsers block third-party cookies, people clear or reject cookies, and privacy features limit tracking. A visitor may be counted as a new person on each visit, which makes a multi-visit journey hard to reconstruct.

Cross-device journeys hide the full path. Someone clicks an email link on their phone, then completes the signup later on a laptop. Without a login or a stable identifier, that looks like two unrelated visits.

Dark social is hard to trace. Links shared in WhatsApp, Slack, Discord, text messages, or private emails often arrive without a useful referrer. They show up as direct traffic even though a real person recommended you.

Ad blockers and privacy tools can stop tracking scripts from loading at all. The visit may still happen, but the source may not be recorded.

Redirects and apps can strip UTM parameters. A link may have parameters when it leaves your email, but a redirect, URL shortener, or in-app browser can drop them before the destination page sees them.

The gap between click and conversion can be long. If someone clicks a campaign link in a newsletter and signs up three weeks later, connecting those events requires a way to remember the original click across time.

Direct traffic is a catch-all. It includes bookmarks, typed URLs, email links without parameters, dark social, and any visit where the source is not passed. That makes direct traffic larger and less useful than it appears.

Offline influence does not show up in web analytics. A podcast mention, a conversation, a conference talk, or a recommendation from a friend can drive a visit that looks like direct or search traffic.

What a small site can actually do

You do not need perfect attribution. You need enough signal to make better decisions.

Use UTM parameters on every link you control. Put them in newsletters, social posts, ads, and partner links. Keep campaign names consistent so you can compare like with like.

Use tracked campaign links for actions that matter. A pageview is useful, but a signup or purchase is the real outcome. If your analytics tool supports it, credit later conversions to the original campaign link.

Track clicks on buttons, not just page loads. A data attribute on a button can tell you when someone clicks "Sign up" or "Start trial" even if the next page does not load as expected.

Build funnels from page paths and events. A funnel can show whether campaign traffic reaches the pricing page, starts signup, and completes it. That gives you a clearer view than one total conversion number.

Compare sources as directional evidence, not exact truth. If one campaign consistently brings signups and another only brings bounces, that difference is usually meaningful even if the numbers are not perfect.

Accept that some traffic will always be unknown. A small percentage of direct or unassigned traffic is normal. Do not spend hours trying to identify every visit.

Check your own links before a campaign. Use a redirect checker or click through your own links to confirm UTM parameters survive. If a platform strips them, shorten the final destination yourself or use a tracked link you control.

Where analytics fits in

A cookieless tracker like measuremy.site can still show pages, sources, countries, browsers and events. Tracked campaign links with UTM tags can credit later traffic and signups. That gives you a workable view of which campaigns lead to real outcomes without trying to follow one person across days.

If you connect measuremy.site to a coding agent over MCP, you can ask questions like "which campaign led to the most signups this week" and let the agent read the analytics and build a funnel for you. That can be faster than clicking through dashboards when you are trying to decide where to spend the next hour.

Attribution is hard because the web was not designed to tell you why someone arrived. It was designed to show pages. You can still get useful answers by tagging consistently, tracking the actions that matter, and treating the results as a guide rather than a precise ledger.

See it on your own site

measuremy.site is cookieless analytics your coding agent can read. 30-day free trial, no card. Plans from $9/month.

Start free trial See the live demo

Add measuremy.site analytics to this project: connect the measuremy.site MCP server at https://a.measuremy.site/mcp for all my projects (claude mcp add --scope user; it signs in with OAuth, no key, and one sign-in covers every site), then tell me how to sign in. Details: measuremy.site/llms.txt

Free tools

All free tools

More from the blog