Insights/Quant method

Spot the Outlier: Why Averages Hide Where Revenue Is Lost.

Most marketing reports describe the average. Growth is rarely decided there. It is decided by a handful of segments performing far above or far below the line.

Bespoke Research7 min readNovember 20263 exhibits
Cover exhibit

One segment out of 46 returns more than four times the average.

Spend index (x) against return index (y).
0 50 100 150 200 0 25 50 75 100 The outlier 4.2x average return Average

Every monthly report has a headline number: blended return on spend, average cost per acquisition, overall conversion rate. These numbers are useful for tracking. They are almost useless for deciding where the next unit of budget should go.

Averages compress the information that matters. A blended return of 1.4x can hide one segment at 4x and three segments that destroy value. Spend more on the blend, and you buy more of both.

4.2xReturn of the outlier segment against the portfolio average
1 in 46Segments that sat far outside the average line
1Reallocation decision, made in one review cycle

Where outliers hide

Outliers rarely sit in the channel view. They appear when the same data is cut by a second dimension: region by channel, creator by product, audience by time of day. The question is not "which channel works" but "which combination works, and how much budget does it currently receive".

Exhibit 01Return on spend by channel
Creator commerce 4.2x Search 2.6x Paid social 2.1x Retail media 1.8x Display 1.1x Out-of-home 0.7x

Return on spend, multiple of investment, trailing 90 days.

The average tells you how the business is doing. The outlier tells you what to do next.

From outlier to budget

Finding an outlier is the easy part. The discipline is in proving it is real, then moving budget toward it in measured steps while tracking whether the return holds at higher spend.

Exhibit 02Campaigns spike and fade. Infrastructure compounds.
Infrastructure-ledCampaign-led
0 35 70 105 140 M0 M6 M12 M18 M24 Infrastructure Campaigns

Index of monthly revenue contribution, month 0 to month 24.

Key takeaways

  1. Treat blended averages as a tracking metric, never as an allocation signal.
  2. Cut performance by at least two dimensions before deciding where budget goes.
  3. Validate every outlier with a 30-day test before scaling it.
  4. Reallocate in steps and watch whether the return holds as spend rises.

How to apply it

  1. Consolidate spend and revenue data into one view, by segment, not by platform.
  2. Plot every segment on spend against return, and draw the average line.
  3. Investigate anything far above or below the line before the next budget cycle.

Notes

Charts in this article are modelled examples that explain the method. They do not represent any client.