A Single Metric Can Hide More Than It Reveals

A healthy blended number is one of the easiest ways to miss a problem. Cost-per-click looks reasonable. Cost-per-lead sits comfortably inside target. Nothing on the dashboard is flashing red, so nothing gets a second look. The trouble is that a single blended figure is an average, and an average's whole job is to smooth over the differences underneath it. Sometimes those differences are exactly the thing worth knowing.

This isn't a client-side story. It's Halo's own agency, Alexander Twelve, catching it in its own operating discipline. Paid social and display campaigns were being evaluated the way most agencies evaluate them, on cost-per-click or cost-per-lead, taken as a single number per campaign. On the surface, that's a completely normal way to run media. It's also a number that treats every lead the same, regardless of what that lead was actually worth to the business it came from.

It wasn't true. Two campaigns could produce leads at what looked like an identical cost, and be worth entirely different amounts, because a lead from someone encountering the brand for the first time is not the same commercial event as a lead from someone re-engaging after already knowing the business. Blending them into one figure didn't just lose some nuance. It actively hid which spend was doing the most useful work and which wasn't.

The fix was structural, not cosmetic. Campaigns got restructured around three distinct journey stages, awareness, new-customer acquisition, and re-engagement, each measured and reported on separately rather than rolled into one blended figure. Once the three stages were split apart and tracked consistently across the client base, the range underneath the average told a very different story: against a blended cost-per-lead of £6.50 across £102,663.59 of spend and 15,797 leads, individual stage costs ranged from under £2 to nearly £20. A blended number that looked entirely acceptable was quietly averaging a stage performing exceptionally well with one performing quite poorly, and nobody could see either one while they stayed merged.

That range is the whole point. £6.50 is a defensible number to report and a genuinely useless number to make a decision from. It doesn't tell you which stage to invest more in, which to fix, or which to stop funding. Split into three, the same spend became a decision-making tool instead of a status update, which is exactly the distinction behind reporting only earning its place if it changes a decision. A blended metric can be perfectly accurate and still change nothing, because there's nothing actionable left once the variance underneath it has been averaged away.

This is also why "the number looks fine" is such a poor test of whether a marketing system is actually working. A number can look fine and still be hiding a stage that's failing, propped up by a different stage that's overperforming enough to drag the average back into an acceptable range. The overperforming stage isn't being recognised or invested in further. The underperforming one isn't being fixed. Both sit invisible underneath a figure that, taken at face value, gives leadership no reason to ask a single follow-up question.

A short way to check whether your own marketing metrics have the same problem:

• Pick your single most-trusted marketing number, the one that gets quoted in a leadership meeting without anyone asking what's underneath it, and ask what it's actually averaging together.
• Split it by a real distinction, not an arbitrary one. Journey stage, channel, campaign type, whatever variable would actually mean something different to how budget gets spent, not a split for its own sake.
• Look for the range, not just the average. A number sitting comfortably in target can still be blending a strong performer with a weak one, and the average will never tell you that on its own.
• Ask what decision the blended number currently supports. If the honest answer is "none, it's just reported," that's the same test every piece of reporting should pass, and a metric that fails it is worth restructuring, not just repeating.

A single blended metric can hide more than it reveals.

This is the same pattern behind why more traffic, more followers, and more leads aren't the same as growth: an output metric moving in the right direction doesn't confirm the outcome underneath it is actually healthy. Clicks, followers, and blended cost-per-lead are all the same category of problem wearing different clothes, numbers that are real, honestly reported, and still capable of concealing exactly the thing leadership most needs to know. It's also one more example of where revenue quietly leaks without ever showing up as a red flag on a report, because the report itself was built at the wrong resolution to catch it.

None of this means blended metrics are wrong to report at all. A single top-line figure has its place, in a board update, in a quick health check, in the four things Halo actually measures on every engagement. The point is narrower: a blended number should never be the only number a real spending decision gets made from, and if it currently is, that's worth treating as a finding, not a formality.

If your business is confident in a headline marketing number and hasn't recently checked what's blended underneath it, that's exactly the kind of question a Commercial Diagnostic is built to test, a structured 90-minute session to find out whether a number that looks healthy is actually telling you something useful. More on the thinking behind this on About, or get in touch directly.