Research
Koderead Research
We publish our method before our findings, and we do not publish a number we cannot substantiate.
Most published AI research in this market is vendor marketing with a sample size attached. The method is rarely disclosed, the sample is rarely described, and the conclusion is always that you should buy something. We would rather release the framework and let it be argued with.
Published methodology
The Koderead AI Value Framework™
Every candidate process is scored out of ten on eight factors. Six of them establish whether automation is worth doing; two establish what the resulting system is permitted to do unattended. Keeping those separate matters — treating risk as a veto rather than a boundary is why many organisations end up automating only trivia.
- Version
- 1.0
- Published
- 2026-09-01
- 01
Repetition
Value factorDoes each instance have the same shape?
Sameness is what makes automation reliable rather than merely possible.
0 = every case differs · 10 = near-identical every time
- 02
Human time
Value factorHow many hours a month does this consume?
The single largest input to the business case, and the one most often estimated rather than counted.
Measured from observation, not recall. Recorded as hours/month.
- 03
Cost
Value factorWhat does that time cost, fully loaded?
The same hours are worth very different amounts. Qualified people doing clerical work scores highest.
0 = junior admin rate · 10 = senior professional rate
- 04
Revenue impact
Value factorDoes doing this faster or better win more work?
Cost savings are finite. Revenue effects are not, which is why quoting and lead response outrank most back-office work.
0 = pure back office · 10 = directly decides whether you win the order
- 05
Data availability
Value factorCan a system reach the information it needs?
The factor most often skipped and the one that most often kills a project mid-build.
0 = lives in people's heads · 5 = present but no API · 10 = accessible via API
- 06
AI feasibility
Value factorIs this something the technology genuinely does well today?
Scored honestly, including where the honest answer is not yet. This is where vendor optimism usually enters.
0 = research problem · 10 = routine, well-proven pattern
- 07
Risk
Boundary factorWhat happens when it gets one wrong?
Sets what may run unattended versus what must be prepared for authorisation. Never a veto on its own.
0 = trivially reversible · 10 = irreversible financial, legal or safety impact
- 08
Implementation difficulty
Boundary factorWhat would it actually take to build and adopt?
Adoption effort belongs here too. A technically simple system nobody opens has still failed.
0 = configuration · 10 = new integration, new process, organisational change
Value — is it worth doing
- Repetition/10
- Human time/10
- Cost of that time/10
- Revenue impact/10
- Data availability/10
- AI feasibility/10
Boundary — what it may do
- Risk if wrong/10
- Implementation difficulty/10
Reading the result
| Value score | Risk | Treatment |
|---|---|---|
| High | Low | Fully automated |
| High | High | Prepare, approve |
| High | Judgement led | Assist a person |
| Low repetition or data | Any | Fix process first |
Six factors establish whether automation is worth doing. Two establish what the resulting system may do unattended. Keeping them separate is why risk sets a boundary rather than acting as a veto.
Reading the result
Fully automated
High value score · low risk · low difficulty.
Runs unattended with monitoring and an exception queue. Reserved for high-volume, cheaply reversible work.
AI-automated
High value score · higher risk.
The system does the work and a person authorises the output. Quotes, payment preparation and filings sit here.
AI-assisted
High value score · judgement dominates.
The system retrieves, drafts and surfaces. A person decides and owns the outcome.
Fix the process first
Low repetition or low data availability.
Automation would encode a problem rather than remove it. A legitimate and common finding.
AI Opportunity Score
The six value factors sum to a score out of sixty, normalised to 100. The two boundary factors do not add to it; they determine which of the four implementation bands the process falls into.
The free AI Opportunity Score on this site is a compressed, self-serve version: eleven questions estimating four of these factors across four areas of a business. A real assessment scores all eight against measured volumes, process by process.
The framework produces a ranked argument, not a decision. It is deliberately possible to disagree with a weighting — that disagreement is the most valuable conversation in an assessment, and it cannot happen while the reasoning stays in someone's head.
Free to use, adapt and cite with attribution.
Open study
The SME AI Implementation Study
Among small and mid-sized businesses that have attempted an AI implementation, what distinguishes the ones that reached production from the ones that stopped at a pilot?
Adoption surveys are plentiful. What is scarce is evidence about the gap between pilot and production, which is where most of the money is lost.
Method, published before fieldwork
- Structured interview of 30–45 minutes with someone who owned or sponsored the implementation
- Businesses between 10 and 500 employees, across the United States, United Kingdom and European Union
- Both outcomes recruited deliberately — reaching production is not a condition of participating, and stalled projects are the more informative half
- Questions published in full before fieldwork, so framing cannot be adjusted to fit an emerging narrative
- Participants anonymised; sector, size band and market reported
- Sample size, recruitment method and every limitation reported alongside the findings
- Raw aggregate data released so the analysis can be checked
On publication
Findings publish only once the sample is large enough to say something defensible. If it is not, we will say that instead of publishing anyway.
Take part
If your business has attempted an AI implementation in the last two years — successfully or not — we would like to interview you. Stalled projects are genuinely more useful to this study than successful ones.
Volunteer for an interviewNo findings are published on this page because none have been produced yet. We would rather have an empty results section than a fabricated one — a statistic that fails verification costs more credibility than it could ever buy.
You are building a view before involving anyone.
Apply the framework to your own business.
The assessment is a compressed version of it — eleven questions, scored in your browser, with the arithmetic shown.