Teach your ad platform what actually makes you money

Introduction

Most ad accounts are set up to chase whatever's easiest to measure: purchases, form fills, revenue. The trouble is that's rarely what the business actually wants more of. A £50 sale at 60% margin and an £80 sale at 8% margin look identical to a platform optimising for revenue. A form fill from someone who becomes a customer and one that goes nowhere both count as "conversion." The platform can't tell the difference unless someone tells it.

That's the problem Signal Engineering solves: making sure the number your bidding algorithm is chasing is actually the number that matters to your business, rather than whatever happened to be easiest to send it.

Why conversions and value aren't the same thing

Google Ads and Meta will get you more of exactly what you ask for. If you send purchases, you get more purchases, including the low-margin ones. If you send form fills, you get more form fills, including the ones that never turn into anything. If you send revenue, the algorithm will chase your priciest products regardless of what they actually make you.

This isn't the platform doing anything wrong. It's doing what it's told. The problem is that "conversion" is only ever a stand-in for what you actually want, and stand-ins stop working the moment there's a gap between what's easy to track and what's actually valuable.

Connecting the real outcome back to the platform

Signal Engineering closes that gap by getting whatever system holds your real outcome talking to the platform that's placing your media. What that involves depends on where the real outcome lives:

CRM data is the most common starting point. If a form fill is your tracked conversion but a closed deal is what actually matters, connecting your CRM (HubSpot, Salesforce, or similar) back to your analytics lets the platform learn which leads actually became customers, not just which ones filled in a form.

A lot of value also happens away from the browser entirely, phone calls and in-store purchases being the obvious examples. Call tracking numbers and point-of-sale data, matched via loyalty ID, phone number or email, let you feed those outcomes back in so they're not simply invisible to measurement.

Matching customers by hashed email, rather than relying on a single click ID, means one dataset can feed several ad platforms at once instead of building the same pipeline three separate times.

And then there's margin. Instead of sending raw revenue as the conversion value, you send actual margin or contribution profit per order. The mechanism is the same as everything above, just applied to profitability rather than lead quality. "Maximise conversion value" starts meaning maximise profit, not maximise whatever number happens to be biggest.

One more piece worth including: suppressing customers you already have. Building an audience from login activity or account behaviour and excluding it from prospecting campaigns stops you paying to "acquire" someone who's already yours.

Making sure the value you send is honest

Once the real outcome is flowing back, the number itself needs some care too.

Predictive scores tend to get overconfident about their best-looking prospects, so it's worth capping the value scale at a realistic ceiling rather than letting bidding chase the model's own optimism. People who decline tracking, or outcomes you haven't matched yet, still convert at a normal rate — valuing them at zero teaches the algorithm to avoid a real chunk of your customers, so a sensible average is safer. And sending an early estimated value soon after the event, followed by a confirmed value once things settle (a closed deal, a matured order), tends to work better than waiting for certainty before sending anything at all.

Where this leaves you

If your account is currently optimising for volume, revenue, or plain conversion count without much thought to what those conversions are actually worth, there's a reasonable chance you're leaving efficiency on the table. Not because the media team is doing anything wrong, because the number they're optimising toward was never quite the right one to begin with.

Most of what's needed to fix this is already sitting in a CRM, a POS system, or a finance spreadsheet somewhere. The work is connecting it properly and getting the value itself right.

Get in touch with Fresh Egg's data engineering team if you want a second opinion on what your ad platforms are currently optimising for, and what a more honest signal would look like for your business.

Get Fresh Thinking and Stay Ahead of Digital Marketing Trends

Subscribe to our Fresh Thinking Newsletter and get...

✅ Exclusive invites to our helpful knowledge-share events ✅ Curated news roundup of pertinent and topical industry news ✅ Early access to our industry resources and insights

Register for Fresh Thinking and keep up to date with our helpful digital marketing news insights.

Extraction complete. All navigation, ads, and non-content elements have been removed. The content is now ready for CMS import with proper markdown formatting, preserved links, and logical heading hierarchy.