The real example

One market, four situations.

Every company in this market was read against the same standard, then placed by what its own operating numbers are doing. Four groups came out, plus seventeen companies the evidence could not place. Nobody decided in advance how many there would be.

1,921 companies read 13 criteria 175 worth engaging read 18-21 September 2026

How the groups were made.

These are stated rules, applied to every company the same way. They are not clusters a model found and labeled afterward.

One standard, read across the whole market

Every company was scored against the same thirteen criteria before any grouping happened. The groups sit on top of that reading rather than replacing it.

The rule is written down first

Each group is a definition with thresholds in it, published below exactly as the research used it. A company meets a definition or it doesn’t, and anybody can check which.

Nothing is placed on a guess

Where public evidence could not place a company, it stayed unplaced. Seventeen companies are in that position and they are reported as their own line, because folding them into the nearest group would make this page tidier and wrong.

The four groups.

In the order the research uses, which runs from the companies in most operational trouble to the ones with the most practice at avoiding it. It is worth reading the fit column straight down.

Stage 0

Absorbing work with people

In plain terms

Work arrived that the bank is clearing by adding people. The cost of running the place is rising faster than what it is running.

The rule, as the research wrote it
Either limb places a bank here. The efficiency ratio deteriorated by three points or more in the quarters following a merger or branch acquisition on the FDIC structure record. Or the efficiency ratio sits above roughly 70% or is worsening over three years, while assets per employee is flat or falling and headcount holds or grows.
What this group has a reason to care about
  • What it costs to run two sets of operations past the conversion date
  • Where the work went after the deal closed, and who ended up absorbing it
  • Whether more headcount is the only way to clear an integration backlog
  • How comparable banks brought an efficiency ratio back down after a deal
Companies
244
In a live event
81
Worth engaging
65
Scoring high on fit
7%
Median employees
132
Mean fit
41
Mean timing
48
Stage 1

Holding steady

In plain terms

Neither losing ground nor gaining it. The largest group in the market by a wide margin, and the one with the least happening to it.

The rule, as the research wrote it
The residual stage. Any bank matching none of the others lands here. Typically the efficiency ratio sits in the 60 to 70% band and is broadly stable over three years, with assets per employee roughly flat.
What this group has a reason to care about
  • Why a stable efficiency ratio can still mean standing still
  • What separates a bank gaining leverage from one holding it
  • Where operating capacity goes when assets per employee does not move
  • What has to be true before an acquisition is survivable
Companies
872
In a live event
84
Worth engaging
60
Scoring high on fit
1%
Median employees
112
Mean fit
19
Mean timing
31
Stage 2

Gaining operating leverage

In plain terms

The best-run operations in the market on their own numbers. Getting more done per person, year after year.

The rule, as the research wrote it
Assets per employee up materially over three years, and the efficiency ratio improving on both the three-year and the twelve-month horizon. Where the two horizons disagree in direction the bank is not clearly gaining leverage and falls to the residual stage instead.
What this group has a reason to care about
  • What is actually producing the leverage, and whether it is durable
  • Whether a gain on both horizons survives the next acquisition
  • What happens to a well-run operation the quarter after a deal closes
Companies
695
In a live event
41
Worth engaging
28
Scoring high on fit
0%
Median employees
113
Mean fit
10
Mean timing
24
Stage 3

Absorbing acquisitions without a step change

In plain terms

They bought something and held the line. The largest institutions in the market by headcount, and the ones with the most practice at this.

The rule, as the research wrote it
A merger or branch acquisition on the FDIC structure record, and since then assets per employee has not fallen and the efficiency ratio has not deteriorated.
What this group has a reason to care about
  • Whether holding the line is the ceiling or the floor
  • Turning a clean absorption into leverage rather than parity
  • What made this one go well, and whether it repeats at the next size up
Companies
93
In a live event
47
Worth engaging
20
Scoring high on fit
5%
Median employees
164
Mean fit
30
Mean timing
40

The topics on each group are ours, read off that group’s own definition. Every figure above them came out of the run.

These are the figures from the run that finished on 21 September 2026. The market was read between 18 and 21 September, and it is read again every month, so they move.

17

The seventeen we could not place.

Public evidence was not enough to put these companies against any of the four rules. They are not a fifth group and they are not quietly added to the nearest one. In a market of this size, seventeen is under one company in a hundred, and the reason it appears at all is that the alternative is a page where every company happens to fit.

Public evidence could not place these seventeen against the ladder. They are not a stage, they are not folded into one, and no stage is guessed for them.

What the fit column shows.

The four groups run in the order the research uses, and the fit figures do not climb with it.

The companies gaining operating leverage, which on their own numbers are the best-run businesses in this market, score lowest of the four. The companies absorbing work by adding people score highest. A company that solves its own problem stops being a prospect, and a market read properly will say so rather than flattering the list.

The real divide here is not how far up the ladder a company sits. It is whether an acquisition has touched it at all. The two groups shaped by a deal hold 337 companies between them and two thirds of the ones scoring high on fit. The two groups moving under their own steam hold 1,567 and the remaining third.

One caution on reading that

Stage 0 is defined partly on the efficiency ratio, and an elevated efficiency ratio is also one of the scored traits. Some of that distance is two measures reading the same underlying fact, so the grouping marks a different kind of company rather than explaining the score.

Not every company in a live event is equal.

Among the companies currently inside an acquisition, the group they belong to still moves both scores. A company that was already under operational strain when the deal landed scores twenty-two points of fit above one that had been gaining leverage.

The companies currently inside an acquisition, broken out by the group they belong to, each scored separately on fit and on timing.
Group Companies Mean fit Mean timing
Stage 0 Absorbing work with people Companies 81 Mean fit 65 Mean timing 70
Stage 3 Absorbing acquisitions without a step change Companies 40 Mean fit 54 Mean timing 54
Stage 1 Holding steady Companies 82 Mean fit 52 Mean timing 68
Stage 2 Gaining operating leverage Companies 40 Mean fit 43 Mean timing 63

This comparison sits inside the live-event set on purpose. Comparing across it would be comparing against the standard’s own entry condition, which would prove nothing.

What we grouped on, and what we didn’t.

Four other ways of cutting this market were tested first. Each one is the way somebody would reasonably expect it to be done, and each one was dropped for a stated reason.

Company size
Two points of fit between the smallest and largest bands. It sorts a market without separating it.
Fit and timing themselves
Both are outputs of the scoring. Grouping a market on its own scores guarantees the groups look different and teaches nothing about why.
How well we could research them
A real split in the data, and an indefensible way to group somebody else’s market. It measures our source coverage, not their companies.
Whether a competitor is installed
22 of 1,921 have one. A group that holds one company in a hundred is not a group, and the rest of the market would be a single undifferentiated block.

Your market has its own groups.

They will not be these, because the groups come out of what your best customers have in common. The market plan is where that starts.

Get your market plan Explore the market yourself
Or see how we build this