# £407 a Month: How B. Scott Hoadley automated his book ad profit analysis with Claude, Airtable and Windsor.ai

B. Scott Hoadley, indie author of the Old Blood Saga, used Claude in its Cowork desktop app to build a system in one working day. The system works out whether his Meta ads for the series make money. Windsor.ai pulls live data from his Meta ad account, and an Airtable base that Claude built stores ads, metrics, attribution and royalties. Claude scheduled tasks sync metrics every Monday and write a profit report on the 1st and 15th, and each report ends with a kill, cut or scale decision against a rule. The analysis found an organic baseline of £407 a month before ads, and he says, "I didn't write a line of code."

| | |
|---|---|
| **Who** | B. Scott Hoadley, indie author of the Old Blood Saga who also consults, writing on his blog Indie Author Notes ([indieauthornotes.com](https://www.indieauthornotes.com/)) |
| **Problem** | Six months of hand-assembling Meta, Amazon Attribution and KDP royalty CSVs without a straight answer to whether his ads made money |
| **Stack** | Claude (Cowork desktop app, scheduled tasks), Windsor.ai, Airtable, Meta ads, Amazon Attribution exports, KDP royalty reports, Word reports |
| **Confidence** | Medium (first-person, his own blog; results are modelled and self-reported; no follow-up results in source) |
| **Last verified** | 2026-11-02 |

## The problem
The Old Blood Saga is five books sold in UK and US markets, with roughly a 50/50 split between Kindle Unlimited page reads and straight sales. He says that mix makes ad measurement difficult. For six months he downloaded Meta exports, Amazon Attribution exports and KDP royalty reports, then spent hours assembling them in a spreadsheet. He kept asking "are my ads actually making money?" without getting a straight answer. He calls extracting it "a part-time analyst’s job" on top of writing books and consulting.

He lists three measurement problems. Amazon Attribution undercounts because KU borrows are invisible to it, read-through happens weeks later and conversions lag. He calls it "a floor." KDP royalties arrive without provenance, so they do not show which ad earned them or whether he would have earned them anyway. The data lives in three places on three different schedules.

## The blueprint (as he describes it)
1. **A live pipe to Meta.** Windsor.ai (£19/month) holds the live connection to his Meta ad account. He describes it as "just a pipe": when queried, it answers with fresh ad data.
2. **Airtable as memory.** The base has six tables: ads, daily metrics, attribution, royalties, books and report history. Claude built the whole base and seeded it with his last 90 days of ads. It pulled names, copy, budgets and launch dates straight from Meta. He says his old manual spreadsheet became obsolete "in about an hour."
3. **Claude scheduled tasks as the engine.** Every Monday the task syncs the week's Meta metrics into Airtable. On the 1st and 15th it runs a full profit analysis and writes him a Word report. He planned to use Make but found it unnecessary with Claude in the loop. He says he may add Make in a future iteration as he scales.
4. **Two manual touches a month.** His part is about fifteen minutes total. He exports Amazon Attribution fortnightly and a KDP royalty report monthly, then uploads them.
5. **The system checks itself.** Each run includes checking whether its own previous recommendations worked. He calls this "the part no dashboard ever does."
6. **Every report ends with a decision.** He wanted each report to end with "kill, cut, or scale" measured against a rule.
7. **Derive the rule from his own data.** Cross-referencing six months of royalties against ad history gave him an organic baseline. Read-through and KU gave him a lifetime value per converted reader. The two combine into the rule: "pay up to ~£4.50 per converted reader; never above £6".
8. **Convert everything to £-per-reader.** He compares cost-per-reader against reader lifetime value instead of ROAS, because ROAS varies with price points and KU mix. He also normalized KU reads by each book's length to map his series funnel.

## Stated results
- Organic baseline: "In January, before ads, the series earned £407/month on its own." He says his raw royalties-vs-spend comparison had been flattering the ads.
- A converted reader is worth "£3.54 lifetime, and £5.90 measured on recent months". Read-through multiplies attributed value by 1.6–2.6×.
- Best US ad: £3.26 per reader. He plans to scale it in ≤20% budget steps so Meta's learning phase does not reset.
- Main UK ad: £4.56/reader at £15/day, and £8.20/reader at its scaled £38/day budget. He cut it back to £15.
- Three other ads at £6.89, £9.86 and £25.62 per reader were killed.
- Post-cuts, the portfolio models out at a true ROAS of roughly 1.1–1.6 on about £35 a day, in his words. This is a model, not a measured result.
- Funnel: he loses 40–45% of readers between Books 1 and 2. From Book 2 onward, 80–95% continue at every step. He says lifting continuation from 60% to 70% raises the value of every reader by about 15%.
- No measured results after the changes are stated. August was to be the first full month under the new regime.

## Caveats
- This is a first-person post on his own blog, not a vendor's. The post does not sell a product, course or kit. The blog has a subscribe signup, and he mentions a consultancy.
- Results are self-reported by a single author. The ROAS figure is modelled. He says "I’ll report back," but the source contains no follow-up results.
- His stated costs are Claude £13/month, Windsor.ai £19/month and Airtable £20/month. He also uses Claude and Airtable for other work, so those costs are not only for this system.
- He says the three killed ads were "confirmed three different ways" but does not describe the three methods.
- The claim that true return is "likely 1.6–2.6× what Attribution shows" comes from his own series data. It may not hold for other authors.
- Handles: Bluesky @bscotthoadley.bsky.social (verified 2026-11-02: bio names him as Old Blood Saga author). Threads @bscotthoadley, Instagram @bscotthoadley, TikTok @bscotthoadley, Facebook bscotthoadleyauthor (verified 2026-11-02: linked from his blog footer). No X or LinkedIn profile is linked from his site.

## FAQ
### What tools does B. Scott Hoadley use for his book ad analysis?
He uses Windsor.ai for the live connection to his Meta ad account and an Airtable base with six tables. Claude scheduled tasks sync data and write Word reports. He considered Make but found it unnecessary for now.

### How much of his time does the system take?
He says his part is two touches a month, about fifteen minutes total. He exports Amazon Attribution fortnightly and a KDP royalty report monthly, then uploads them.

### Did the system increase his ad profit?
None stated. The post gives a modelled true ROAS of roughly 1.1–1.6 on ~£35/day after cuts. He says he will report back on August, but the source has no follow-up results.

### Did he need to code?
No. He writes, "I didn't write a line of code."

## Sources
- https://www.indieauthornotes.com/automating-my-book-ad-analysis-in/ (primary: first-person post by B. Scott Hoadley on his blog Indie Author Notes, dated 17 Jul 2026)

---
*Want agents like these working your distribution? [Get It Built](/get-it-built). Get new teardowns weekly: [subscribe](/subscribe) (double opt-in).*