SHIELD Perception · Worksheet
You have never seen the invoice for the people you lose.
Every organization has a group that leaves faster than the rest. Almost none have priced the gap. Six inputs. Five are in your HRIS. The sixth is an assumption, and this page tells you which one and what to do about it.
Get the full breakdown
The number above is the headline. The version you can take to a CFO needs more behind it.
- A sensitivity table showing the range across replacement-cost assumptions
- The sourcing behind every assumption, so the figure survives scrutiny
- A printable worksheet for pulling the four inputs from your HRIS
- The eighteen-month departure pattern, and where to find it in your own data
One email with the breakdown attached, then occasional writing on retention and system design. Unsubscribe any time.
Check your inbox in the next few minutes. If it hasn't arrived, look in promotions — it comes from peter@shieldperception.com.
Where to cut the data
The same headcount produces very different numbers depending on how you divide the population. Run it more than one way. The cut that produces the widest gap is usually where the problem lives.
By manager
Almost always the widest spread, and the most actionable. This is often the cut that produces the longest silence in the room.
Internal promotion vs external hire
Shows whether you are bringing capable people into a system that doesn't hold them.
By function, within comparable pay bands
Controls for the confounding that makes raw department comparisons meaningless. Engineering versus a call center tells you about jobs, not systems.
By hiring channel
Referral, inbound applicant, agency, recruiter. Referral hiring is a copy machine. This shows what it has been copying.
First year vs everyone else
Separates a hiring problem from a retention problem. These get confused constantly and have completely different fixes.
One practical caution. The five cuts above are proxies.
Few organizations can segment attrition on cognitive difference, because the field may not exist in the HRIS. What exists instead is voluntary self-disclosure, and self-disclosure is not a measurement. People decide whether to tell you based on whether they expect it to be used against them. Which means the group least protected by your systems is the group least likely to appear in your data.
A cut on that field measures disclosure confidence and reports it as prevalence. So a null result tells you nothing, and a positive result understates what it found by an amount you cannot calculate.
The five cuts avoid all of that. They run on fields you already hold, they ask no one to disclose anything, and they surface the same structural problem from a direction that carries no handling obligation. Start there.
If you want the direct cut later, run it to test a finding you already have. Not as the way you go looking for one.
How the number is built
Excess exits = group size × (elevated rate − baseline rate)
Annual cost = excess exits × salary × multiplier
The calculation asks one question: if your highest-turnover group left at the same rate as your lowest, what would you stop spending?
Everything above that floor is a cost your organization absorbs without having decided to.
It is deliberately conservative. It counts replacement cost only and does not include lost institutional knowledge, the productivity decline in the months before someone leaves, nor the effect on the people who stay and watch. Those are real and amplify the numbers here.
It also proves nothing about cause. A gap tells you where to look. It does not tell you what you will find. And how you cut the population changes what the gap means. That is the reason to run it several ways. Most organizations have never run it once.