Skip to main content

Benchmarks

Benchmarks in the top bar shows medians across every engagement taking part. Only anonymised aggregates are used, and nothing is shown until at least five engagements contribute to a figure. Published figures are rounded (pounds to the nearest £500, percentages to the nearest whole point, spans and layers to the nearest 0.5), so no single engagement's figures can be worked back out.

The Benchmarks page with Compare an engagement set to No comparison and the Across engagements table for 5 engagements: medians with interquartile ranges such as 36% average automation potential, 74% average augmentation potential, £49k employer cost per head, 7.5 average span of control and 5.0 layers; and the start of By occupation group, with 5 groups published and 4 withheld as too thin.
The Benchmarks page. Every figure is a rounded median with its interquartile range.

Who contributes​

  • An engagement takes part when Contribute to cross-engagement benchmarks is ticked in its settings (the default).
  • Each participating engagement contributes the summary of its most recent finished staff list, so a large client counts once, like any other.
  • Summaries hold aggregates only: no rows, names or employee IDs.
  • Purged engagements still contribute their summaries unless the setting is turned off.

Reading the page​

Each figure shows the median in bold with the interquartile range (25th to 75th percentile) in brackets.

Across engagements (N engagements · median (interquartile range)):

  • Average automation potential
  • Average augmentation potential
  • Employer cost per head
  • Moderate capacity gain (% of cost)
  • Average span of control, Managers (% of people) and Layers, from engagements with reporting lines. These are withheld until five engagements with reporting lines take part.

By occupation group: for each published group, the number of contributing engagements, the Share of workforce, Augmentation and Cost per head. Groups seen in fewer than five engagements are withheld as too thin.

Comparing an engagement​

Compare an engagement positions an engagement's latest finished staff list against the benchmark. Its own figures are left out of the benchmark it's compared against, so it isn't compared with itself.

Each figure then gains a pill: Ours 41.3% · in range, · above or · below the interquartile range.

If the engagement has no finished staff list, the page says That engagement has no finished staff list yet. The threshold of five applies after the compared engagement is left out, so comparing needs at least five other contributing engagements; with fewer, the page shows Not enough other engagements yet.

When the page is empty​

Not enough engagements yet: N engagements contribute so far; benchmarks appear from 5. Each engagement contributes its most recent finished staff list. The threshold protects clients: with only a handful of contributors, medians and quartiles can be worked back to an individual client's figures. When we compare one of our own engagements, it's left out first, and the threshold applies to what remains.

Caveats​

  • Engagements differ in scope (a whole group, or one division), industry and filters at upload. Treat benchmarks as context for a conversation, not a target.
  • Contributing summaries may have been scored with different reference data versions.