State of Operations 2027

State of Operations 2027 is in progress, and this page contains no results. It sets out what the volume will measure and how: how many tools owner-led companies pay for, how many hours a week go on retyping information between them, how much of the owner’s week goes on administration and where work stalls between people. The data will come from anonymized Verticulate diagnostics, and the same definitions will be used every year so companies can compare themselves with peers and with the year before.

The studies owners read describe much larger companies

Owner-led companies with 20 to 250 staff are numerous. In US professional, scientific and technical services alone, companies with 20 to 499 staff outnumber those with 500 or more by about 14 to 1.¹

1. 20 to 499 staff53,523
2. 500 or more staff3,776

Professional-services companies with 20 to 499 staff outnumber large ones by about 14 to 1

Firm size	Firms
20 to 499 staff	53523
500 or more staff	3776

Source: US Census Bureau, Statistics of U.S. Businesses 2022¹

Note: 20 to 499 staff combines two Census bands: 44,391 firms with 20 to 99 staff and 9,132 with 100 to 499.

The research these companies read is mostly about the large ones. Bain’s widely cited estimate of where software development time goes, for example, draws on surveys of developer teams of about 2,000 to 20,000 people.² A single one of those developer teams is larger than an entire company with 20 to 250 staff.

Size changes the answers. In Capterra’s survey of US software buyers, 51% of buyers at companies with 5 to 249 staff regretted a recent purchase, against 66% to 67% at larger companies.³ Owner-led companies have fewer layers of management, and more of the operation is held in the owner’s head. With little data on how they actually work, owners fall back on vendor surveys and anecdotes.

Volume 1 of this series, What business owners ask before buying operations help, described what owners worry about before they buy. Volume 2 will measure what their operations look like underneath.

Six measures, each defined so a company can count it

• Tools per company. The number of paid software tools, and the share used by fewer than half of their paid seats.
• Hours of re-entry per week. Time spent typing information that already exists in another tool.
• Owner time on administration. The share of the owner’s working week spent moving, checking or chasing work, instead of selling, leading or doing client work.
• Handoffs per job. The number of times a typical job changes hands, and the share of those handoffs with no named owner.
• Days from sale to start. The elapsed time from a signed agreement to the first piece of delivered work.
• Conflicting systems of record. The kinds of information, such as job status, that more than one tool claims to hold.

Four of the six can be counted directly from a map, an inventory or a pair of dates. The other two rest on estimates by the owner and the team, and the volume will report them as ranges instead of false precision.

Measure	Taken from	Reported as
Tools per company	Tool inventory	Count
Hours of re-entry a week	Staff estimates	Range
Owner time on admin	Owner's estimate	Range
Handoffs per job	Operating map	Count
Days from sale to start	Signing and delivery dates	Count
Conflicting records	Tool inventory and map	Count

Source: Verticulate research plan for Volume 2; no results yet

The data will come from diagnostics clients agree to share

Every Verticulate diagnostic produces a process map, notes from the mapping calls and an inventory of the tools in use. With each client’s permission, we will extract the six measures, remove anything that identifies the company and pool the results.

The plan for the sample and the period is:

• Who is in it. Owner-led companies that complete a Verticulate diagnostic and agree to share their anonymized measures. Companies that chose to work with an operations consultancy are not a random sample of companies, and the volume will say so on its first page.
• When the data is from. The volume will state the window of diagnostics it covers, so that later volumes can be compared with it.
• How large it must be. The volume will state its sample size on the first page. Breakdowns by industry or size band will appear only where a group is large enough to protect anonymity and support a conclusion.

Data collection is under way. None of the six measures has been reported, and this page shows no partial results.

Medians stop one unusual company from setting the number

Small samples are easy to distort. A single company with an unusual tool count can move an average a long way while the typical company has not changed. The volume will therefore report medians and ranges, not averages.

• Average
• Median

05101520

11.111.5

16.612

8 typical companiesAdd 1 company with 60 tools

One company with 60 tools lifts the average by about 5 tools and the median by half a tool

Companies included	Average	Median
8 typical companies	11.1	11.5
Add 1 company with 60 tools	16.6	12

Source: Verticulate illustration

Note: Tool counts for the eight typical companies: 8, 9, 10, 11, 12, 12, 13 and 14.

The definitions will stay fixed from year to year. If a measure proves unreliable in practice, we will drop it and explain why, instead of quietly changing what it counts.

The volume will take no sponsorship and publish nothing it cannot trace

The volume will carry no vendor sponsorship. It will not publish a figure that cannot be traced to its source, and it will not present a self-reported estimate as a measurement.

You can count two of these measures yourself this month

Two of the six measures need no outside help. To count your tools and the share that go unused, run the keep, connect or replace review, which takes about two hours. To estimate hours of re-entry, ask your team to tally a week of retyping, as described in issue 2 of our newsletter.

The volume publishes only when the sample can support it

We will publish when the sample is large enough to report honestly, and the first page will state the sample size, the period and the limits. Until then, this page describes a plan, and nothing on it is a finding.