Your Conversion Tracking Is Lying to You. Fix It Before You Spend Another Pound

Before we touch bids, budgets or campaign structure on a new account, we check one thing: is the conversion tracking actually telling the truth? More often than not, it isn't — and that single fact explains why the account has been so hard to scale.

It's simple logic. Google's bidding algorithm optimises toward the conversions you report. If those conversions are wrong, double-counted, or missing their value, then every automated decision the system makes is built on a lie. You can have the best structure and the sharpest creative in your category and still lose, because the machine is aiming at the wrong target.

The usual suspects

When tracking is broken, it usually breaks in a few predictable ways:

• Double counting. A purchase fires the conversion tag more than once — on page reload, on a thank-you page revisit, or because both Google Ads and GA4 import the same event. Suddenly the account "earns" far more than your Shopify dashboard shows, and bidding scales toward a number that doesn't exist.

• No conversion value, or the wrong one. The tag reports a conversion but not the order value — or it reports a static value for every sale. Now the algorithm can't tell a £20 order from a £400 one, so it optimises for volume of cheap orders.

• The Ads-versus-GA4 gap. Google Ads and GA4 attribute differently and count differently. Teams waste weeks arguing about which is "right" instead of picking one source of truth and committing to it.

• Consent and cookie loss. With consent mode and tracking prevention, a meaningful share of conversions never get observed at all. If you're not modelling or recovering them, you're underpaying for channels that actually work.

How to audit it in an afternoon

You don't need a six-week project to know whether your data is trustworthy. You need a few honest checks:

1. Reconcile to the source of truth. Line up Google Ads conversions against your Shopify (or backend) order count for the same window. If they're more than a few percent apart, stop and find out why before you change anything else.

2. Confirm dynamic values. Place a test order and check that the actual order value flows through — not a hard-coded number, not zero.

3. Hunt for duplicates. Look at conversions per click. If a single click is producing multiple purchase conversions, you've found a leak.

4. Check enhanced conversions. For most e-commerce accounts, enhanced conversions meaningfully improve measurement in a privacy-safe way. If it's off, that's free accuracy you're leaving on the table.

Build toward one source of truth

The goal isn't perfect data — perfect data doesn't exist anymore. The goal is one trusted number that everyone optimises against. For most brands that means:

• Server-side tracking to survive browser restrictions and reduce loss.

• Enhanced conversions to recover and verify observed sales.

• A clear, documented decision about which platform is your reporting source and which is just a signal feed.

Get there and something quietly powerful happens: the bidding algorithm starts learning from clean data, optimisations stop fighting noise, and the improvements you make actually stick.

Why this comes first

Founders want to talk about scaling, new channels, and creative. We get it — tracking is the least exciting topic in performance marketing. But it's the foundation everything else stands on. Rebuilding bids on top of broken data is like renovating a house on a cracked slab.

Fix the measurement first. Then every pound you spend after it is a pound you can actually trust.