Original research
The benchmarks
Quarterly original research into what $1M+ educators, coaches, and consultants are actually building with AI. Primary data, published method, no vendor sponsorship.
Issue 01 · In fieldwork
AI Implementation Benchmarks: What $1M+ Business Owners Are Automating in 2026
Every AI report on the market surveys enterprises. Nobody has asked the $1M to $5M knowledge business what it actually built, what it abandoned, and what it cost.
So the operators who most need a benchmark have nothing to measure against. They are comparing themselves to a Fortune 500 case study or to a vendor's landing page. Issue 01 fixes that with primary data from the segment itself.
- Fieldwork
- Open now
- Publishes
- Q4 2026
- Sample
- Target n=150
What it measures
What actually got built
Not what people plan to build. What is running in the business today, which function it sits in, and who uses it.
What got abandoned
The builds that were started and dropped, and the reason. This is the number nobody publishes, and it is the most useful one.
Where the founder is still the bottleneck
Which parts of delivery still route through one person, after the AI was introduced.
Real spend
Tooling, build cost, and the internal hours nobody counts. Reported as a range, not an average, because the spread is the story.
Adoption, not deployment
Whether the team uses the thing. A system that shipped and sits unopened is a failed build, and it gets counted as one.
Method
Published before the data, not after, so the method cannot be bent to fit the result.
Who is in the sample
Owner-operators of education, coaching, and consulting businesses turning over $1M or more, self-reported. Responses outside that band are collected but reported separately and never blended into the headline figures.
How the data is collected
Through the assessment on this site, which doubles as the survey instrument, plus structured interviews with a subset of respondents who agree to a follow-up.
What gets published
Aggregate figures only. No respondent, business, or client is identifiable in the report. Individual responses are never shared, sold, or passed to a third party.
What we will not do
No projections presented as findings. No percentage without the sample size next to it. If a cut of the data has too few responses to mean anything, it gets published with that stated, or not at all.
Raw data
The anonymised response set is available to any researcher or journalist who asks. A finding you cannot check is a claim, not research.
Take part in Issue 01
The assessment doubles as the survey instrument. It takes about six minutes, you get your own results immediately, and your responses feed the anonymised benchmark set.