Bad attribution does not announce itself. It shows up as confident-sounding dashboards with numbers that do not match reality, budget decisions that feel right but do not move the needle, and a persistent sense that something is off but no one can say exactly what. If your paid acquisition feels harder than it should, attribution is worth examining before you change anything else.
Reported conversions substantially exceed actual revenue or signups — often the result of duplicate firing or misconfigured events. Conversion volume looks healthy but lead quality is poor — often means you are optimising towards the wrong event. Channel performance looks good in-platform but your overall acquisition cost keeps creeping up — often means channels are taking credit for each other's work. Campaigns that you turn off do not seem to affect results — a strong signal that those campaigns were not causing the conversions attributed to them.
Duplicate conversion tracking is the most frequent. Pixel fires and CAPI both fire for the same event without deduplication logic. The ad platform reports twice the actual conversions. Everything looks better than it is.
Optimising towards the wrong event is the second. Lead form submissions that rarely result in paying customers. Free trial signups with high churn. Page views counted as conversions. The algorithm finds more of what you tell it to find — if you tell it to find low-quality leads, it will be very good at that.
Conversion window mismatch is less obvious but common in B2B contexts. A 7-day click window means a conversion that happens on day 10 is invisible to the algorithm. For products with longer consideration cycles, this can cause meaningful under-reporting of the campaigns that are actually driving pipeline.
Cross-device attribution gaps occur because the same person uses a work laptop, a personal phone, and a home computer during an evaluation cycle. Each touchpoint looks like a different user. Your reported funnel shows multiple partial journeys that are actually one.
Start with an event audit. For every conversion event in your ad platforms, verify it is firing correctly, firing once, and that the event represents something meaningful. Pull a sample of attributed conversions and check them against your CRM or order system. If the numbers do not reconcile, find out why.
Set up server-side tracking if you have not already. Browser-based pixels lose signal from ad blockers, ITP, and cookie restrictions. A Conversions API or server-side Google Tag sends events directly from your server, bypassing browser limitations. It is more robust and gives the algorithms cleaner data to learn from.
Connect your CRM to your ad platforms for B2B. This is the most impactful single change for companies with long sales cycles. Optimising towards a deal closed rather than a form submitted changes what the algorithm finds you.
Use holdout tests to validate. The only way to know whether a campaign is truly driving conversions — rather than just being present when conversions happen — is to show a control group nothing and measure the difference. Run these periodically, especially on your highest-spend campaigns.
Attribution audits benefit from fresh eyes. When you are inside an account every day, it is easy to normalise things that are actually wrong. An outside pair of eyes, with no investment in the current setup, will find things you have stopped seeing.
This is a specific piece of work I do regularly. If you want one done on your account, get in touch.