AI models are becoming more sophisticated every day, and businesses and users can now easily, quickly, and almost freely access reliable knowledge (including scientific insights that were previously very expensive). In this situation, the demand for nonfiction books on Amazon Kindle is declining sharply and will continue to do so. Specifically about science-based books that are written by synthesizing reliable systematic reviews, how about such books?
Short answer: demand for nonfiction e-books is not collapsing, and “science-based syntheses of systematic reviews” can still win—if they deliver trusted, applied decision-support rather than generic summaries.
What the market data actually says
- In the United States, e-book revenue grew in 2024 to $2.1B (+1.5%); digital formats overall rose 11.4% year-over-year (driven mostly by audiobooks). That’s not a sharp decline. (publishers.org)
- The UK shows a similar picture: strong overall publishing revenues with digital book formats up; audiobooks surged +31% in 2024, while print still dominates. (Publishers Association)
- Kindle Unlimited continues paying out a large (and rising) global fund; per-page rates hover around $0.004–0.0045 in 2025—useful context for nonfiction authors considering KU as part of a mixed model. (Written Word Media)
Why evidence-synthesis books still have room
- Trust gap in AI outputs. New audits show high error and sourcing issues when assistants answer informational questions (including news and science). Readers—especially for health/science—still look for credible, citable, accountable sources. (Reuters)
- Science remains trusted. Surveys indicate sustained confidence in scientists and a public keen on reliable health/science guidance—even as people worry about misinformation. (Pew Research Center)
- There’s a formal ecosystem for translating reviews to the public. Cochrane’s Overviews (of reviews) and Plain-Language Summaries (PLS) exist precisely to turn SRs into usable guidance; readers do seek these, but the materials are dispersed and uneven in readability—creating a niche for high-quality, well-packaged books. (Cochrane)
The AI counter-pressure you must acknowledge
- LLMs are getting good at drafting lay summaries of Cochrane reviews (non-inferior to humans in a blinded trial)—which commoditizes basic paraphrase. (PMC)
- Yet hallucinations and omissions remain an intrinsic risk, especially in medicine; citation accuracy is a known weak spot. This reinforces paying for curated, accountable synthesis rather than raw AI output. (Nature)
So…will SR-based science books sell on Kindle?
Yes, if they compete on trust, utility, and packaging—not on summary alone. The opportunity shifts from “summarize the SRs” to “translate SR evidence into decisions for defined audiences.” Think: quantified effect sizes, GRADE certainty, benefits/harms tables, applicability to subgroups, and implementation checklists. Cochrane/academic PLS are a starting point; your book becomes the coherent, guided path.
What wins (actionable playbook)
- Evidence-to-practice design. For each claim: cite SR → report effect size & certainty (GRADE) → context (“for whom/when/how much”) → risks/costs → “what to do next” flowchart. (This is exactly what Overviews/PLS aim at, but your value is integration and actionability.) (Cochrane)
- Transparency signals. Declare search windows, inclusion criteria, and conflicts; register your protocol (e.g., OSF) and link your evidence tables. These are concrete differentiators in an AI-noisy marketplace. (Readers respond to visible rigor amid trust concerns.) (Reuters)
- Package for how people consume now. Release ebook + audiobook (audio is the fastest-growing digital format) and a compact “decision guide” PDF. (The Guardian)
- Living updates outside the file. Maintain a small website/appendix that posts quarterly deltas (new SRs, updated effect sizes). The Kindle book sells the framework; the site provides freshness. (Evidence ecosystems update frequently.) (Cochrane Library)
- Quant + narrative. Use small visuals: NNT/NNH, forest-plot-lite, and one-page “Should I try X?” decision aids per chapter. Readers pay for the decision layer, not just citations.
Business notes specific to Kindle
- Monetization mix. Consider KU enrollment for reach + à-la-carte sales + a short audio edition (even abridged “essentials”) to capture the growing audio segment. (Written Word Media)
- Positioning. Avoid generic titles (“The Science of…”) that LLMs can imitate; anchor on a tight use case (“Evidence-Based Sleep Apnea Self-Care: What SRs Actually Show, for Adults 40+”) and surface the GRADE certainty in marketing copy.
- Proof devices on the retail page. Add a screenshot of your evidence table, PRISMA-style diagram, and a link to your protocol/data; this directly addresses the AI-trust objection.
When an SR-based book won’t work
- If the topic is fast-moving and decisions hinge on very recent primary trials (pre-SR), a static book risks obsolescence; go newsletter/course first.
- If your manuscript is only a rephrasing of abstracts/PLS, LLMs + free PLS will undercut you on speed and price. (The randomized trial above shows parity on readability.) (PMC)
Bottom line
- The macro market does not support a “sharp decline” narrative for e-books; demand is stable to slightly growing, with audio booming. (publishers.org)
- SR-synthesis books can thrive—provided they are trust-maximizing, decision-oriented, and transparently engineered. AI narrows the gap on plain summaries, but the moat is rigorous curation, applicability, and ongoing updates grounded in the SR ecosystem. (Reuters)
If you’d like, I can sketch a table of contents and a chapter template (with effect sizes, GRADE labels, action checklists, and “update boxes”) tailored to your target audience.