An expert bench is current on the day it is built and decays from there. What decays is not the person but their distance from the operation — someone two years out of a role describes a market that has moved. Recency, not seniority, is the variable a bench has to track.
How fast does a bench decay?
Faster in fast-moving markets and faster than most practices assume in all of them. The relevant clock is time since the person was operational rather than time since they joined the bench.
Because the decay is invisible — a bench looks the same on paper as it ages — it is usually noticed when a project produces stale answers rather than in advance, by which point the re-screening cost has already been paid.
- 01
Record last operational date
Track when each expert was last in the role, not when they joined the bench.
- 02
Re-screen on a cycle
Screen at fixed intervals rather than when a project needs someone.
- 03
Retire, do not only add
Remove people whose market has moved, as a rule rather than a judgment.
- 04
Track usefulness
Record who produced usable findings, not only who was available.
What makes an expert current?
Recent operational exposure. On a call, a regional manager who left last quarter frequently knows more about how a market is trading than a former group executive who left three years ago, whatever the titles suggest.
That is uncomfortable for benches assembled on seniority, which is most of them.
Recency is also measurable in a way seniority is not, which is what makes it a workable screening criterion rather than a preference. A date of last operational role is a fact that can be recorded, checked and sorted on; judgments about how good someone is cannot. A bench organized on the first is maintainable and one organized on the second is not.
What happens between projects?
Usually nothing, which is the problem. Benches are built under project pressure and maintained under none, so the maintenance never happens and the next project rebuilds from scratch.
Screening on a cycle rather than on demand is the single change that fixes it, because it separates maintenance from delivery pressure — which is why it usually belongs with the research desk rather than the deal team.
| Maintained bench | Project-by-project sourcing | |
|---|---|---|
| Screened | On a cycle, for relevance | Under deadline, for availability |
| Average recency | Held roughly constant | Whatever the last search produced |
| Cost visibility | A named maintenance line | Buried in project fees |
How often should you refresh?
On a fixed cycle matched to how fast the sector moves, and always with retirement as well as addition. A bench that only grows accumulates people whose market has moved on.
Retirement is the part practices avoid because it feels like a judgment on the individual. Framing it as a recency rule rather than a quality one makes it routine.
Refresh cadence should also follow the sector rather than the calendar the practice happens to run on. A market where pricing and channel structure move every quarter needs a shorter cycle than one where the same five buyers have behaved the same way for a decade, and one interval across a practice's whole coverage guarantees that some of it is stale and some over-serviced.
How large should a bench be?
Small enough to know. Forty names nobody has spoken to in a year is a list, and a list is what the practice already had before it decided to build a bench rather than a board.
Depth in the segments the practice actually sells into beats coverage of the ones it might.
Size also interacts with retirement in a way that makes the problem self-correcting once the rule exists. A bench with a recency threshold has a natural ceiling, because names leave it at roughly the rate new ones are added, and the practice ends up maintaining the number it can actually keep current rather than the number it once collected.