How do I tell a real marketing performance change from a job-type mix shift?

Mix shifts are the most common false alarm in channel reporting

Revenue per lead is a weighted average across job types, and job type shares move for reasons that have nothing to do with campaign quality: weather, season, a competitor's promotion, a change in what people are searching for this month.

When the blended number drops, the first question should never be what changed about the campaign. It should be whether the campaign is still receiving the same kind of demand.

The reweighting calculation, step by step

Take last period as the baseline. For each job type, record its share of total leads. Then take the current period's revenue per lead within each job type. Multiply each current within-type figure by the baseline share, and sum across job types.

That sum is what this period would have produced at last period's mix. Compare it three ways: to last period's actual, to this period's actual, and to the difference between the two. The gap between this period's actual and the reweighted figure is the mix effect. The gap between the reweighted figure and last period's actual is the genuine performance effect.

Where mix shifts come from

• Seasonal demand rotation. The single largest driver in trade businesses, and the most predictable.
• Match type and audience expansion. Broadening keywords or geography reliably changes the job types that arrive, often before it changes volume noticeably.
• Landing page and form changes. A form that leads with maintenance produces a different mix than one that leads with emergency service.
• Intake scripting. How the phone is answered shapes which opportunities become bookings, a link between call handling and revenue mix that most reporting never draws.

What to do once you know

If it is mix and the mix was intentional, nothing is wrong and the report should say so. If it is mix and nobody chose it, find the change that caused it — that is usually a settings change with a date attached.

If the reweighted number moved, the problem is real and inside a specific job type, which is a much smaller haystack than the whole account. Running this decomposition automatically on every period is the sort of routine analysis that belongs in automated business analysis rather than in someone's spreadsheet at month end.