Milestone 19: Collect and Evaluate Publication Evidence
Gary WhittakerSelf-Publishing Path · Milestone 19 of 20
Collect and Evaluate Publication Evidence
Review what actually happened after publication and launch, separate evidence from interpretation and decide what the book needs next.
Evidence before judgement
A book with few early sales is not automatically a failed publication, and a burst of attention is not automatically a durable success. Milestone 19 asks whether the correct editions worked, the intended readers understood the offer and the available evidence supports a responsible next decision.
Collect evidence by source
Publication evidence
Availability, file defects, listing changes, returns, delivery incidents and platform status.
Reader evidence
Questions, reviews, direct comments, completion signals, companion-resource use and support requests.
Commercial evidence
Sales, downloads, page reads, royalties, refunds and actual author costs in CAD where known.
Launch evidence
Reach, clicks, replies, review-copy response, event attendance and channel-specific actions.
Classify every finding
| Classification | Meaning |
|---|---|
| Verified fact | Supported by a reliable record or direct observation. |
| Reader report | A real response that may not represent all readers. |
| Author interpretation | A reasoned reading of evidence, clearly labelled. |
| Hypothesis | A possible explanation requiring further evidence. |
| Unknown | Cannot currently be determined. |
Review the whole reader path
Look beyond sales: could readers find the listing, understand the book, choose the right edition, complete access, use companion resources and receive support? A failure anywhere in that path can distort performance.
Choose an evidence-based decision
Possible outcomes include maintain, correct a defect, update metadata, adjust price, improve the reader page, continue collecting evidence, revise an edition, add a format, pause promotion or retire an edition. Large content changes must return through the earlier editorial and production controls.
Responsible AI use
AI may organize supplied data, compare periods or identify patterns and missing fields. It must not fabricate analytics, infer reader motives as facts or treat correlation as proof.
Bounded prompt: Analyze only the supplied publication, reader, sales and campaign evidence. Separate verified facts, reader reports, interpretations, hypotheses and unknowns. Identify data gaps and decision options without declaring success or failure.
Milestone 19 completion standard
- The review window and evidence sources are documented.
- Publication, reader, commercial and launch evidence are collected.
- Actual CAD costs are recorded when known.
- Facts, reports, interpretations, hypotheses and unknowns are separated.
- Technical and reader-path failures are visible.
- Protected strengths are identified.
- A decision and evidence requirement for the next stage exist.
- A Publication Performance and Reader Response Review is complete.
Frequently asked questions
How soon should I review?
Choose a window that can produce meaningful evidence for the book and channels; there is no universal interval.
Do low sales mean the book failed?
No. Examine visibility, reader fit, access, pricing, timing and the amount of evidence available.
Are reviews reliable evidence?
They are real reader reports, but a small or self-selecting sample should not be treated as the whole audience.
Should I change the price immediately?
Only when the evidence and publishing constraints support a controlled test or decision.
Can AI interpret my results?
It may assist with supplied data, but interpretations must remain labelled and human-reviewed.
What is the main deliverable?
A Publication Performance and Reader Response Review.
