Last reviewed by Robert Prime — May 2026
Introduction
You hire a cover designer. They send you 3 concepts. You like all of them. You pick the one your gut prefers. You publish.
The book doesn't sell as well as expected. You'll never know whether the cover was the problem.
A/B testing book covers solves this. For £30-£100, you can put 2-3 cover variants in front of 50-500 real target-genre readers and find out which converts highest before committing.
This guide covers the two tools most indies use, what to test, and the realistic impact on launch sales.
Why cover A/B testing matters
A book's cover is responsible for roughly 80% of whether a stranger clicks on it in Amazon search results. The other 20% is title, price, review count, and category placement.
Different covers can produce 30-150% differences in click-through rate. That difference compounds across every reader who sees your book.
For an established author with a list: cover matters less (your audience clicks regardless). For a debut indie discovering readers via Amazon search: cover is everything.
The two main testing tools
PickFu (pickfu.com)
Cost: £30-£80 per test depending on demographic targeting.
How it works:
- Upload 2-4 cover variants
- Define target demographic (age, gender, country, reading habits)
- 50-200 real respondents answer "Which would you click on?" + leave a comment explaining why
- Results in 30 minutes to 24 hours
Pros:
- Fast
- Cheap
- Qualitative feedback in comments often more valuable than the votes themselves
- Targeting filters let you reach genre fans specifically
- Statistical confidence on results
Cons:
- Self-reported preference, not actual purchase behaviour
- Some "professional respondents" optimise for paid responses, not authentic reactions
- Doesn't predict Amazon-specific click-through
Verdict: Best starting tool. Run 1-2 PickFu tests per cover decision.
Facebook / Instagram Ads cover test
Cost: £50-£150 (small ad spend over 5-7 days).
How it works:
- Create 2-3 ads with different cover images, identical headline, identical body copy
- Target the same audience (genre-matched: e.g., readers of similar authors)
- Run for 5-7 days with equal budget
- Measure click-through rate per ad
Pros:
- Real-world click behaviour, not self-reported preference
- Reveals actual interest in the book overall (not just cover)
- Test data informs your launch ads
Cons:
- Requires basic Facebook Ads knowledge
- Slower (5-7 days)
- Costs more
- Audience targeting on Facebook is increasingly weak post-iOS 14
Verdict: Worth doing for high-stakes books (large investment, hoped-for big launch). Skip for tight-budget tests.
Amazon AMS / Sponsored Products pre-test
Cost: £100-£300 in pre-launch ad spend.
How it works:
- Launch with one cover; run Sponsored Products ads for 14-30 days
- Note baseline click-through and conversion rates
- Update to alternate cover; run identical ads
- Compare metrics
Pros:
- Real Amazon platform data
- Tests covers in actual buying context
Cons:
- Slow (30+ days)
- Confounded by other variables (review count changes, keyword competition)
- Requires book to be live already
Verdict: Useful for established authors testing whether to re-cover an existing book. Not a pre-launch tool.
What to actually test
Test: 2-3 fundamentally different concepts from your designer.
- Not "this with red title vs this with blue title" — too narrow
- Not "this image vs same image with different filter" — too narrow
- Yes "atmospheric typographic" vs "character-focused photo" vs "object-symbol" — three real strategic alternatives
If your designer gave you 3 cover concepts, those are the 3 to test.
If your designer gave one concept, ask for 2 more strategic alternatives before paying for the final.
How to set up a PickFu test for fiction
- Title: "Which cover would you click on?" or "Which book would you pick up?"
- Body copy: Brief — genre, target audience, 1-line premise. "A small-town cosy mystery set in Yorkshire. Which cover catches your eye?"
- Demographics targeting: Set country (UK or US), age range matching your reader avatar, gender if relevant, reader-frequency filter ("reads 1+ books per month").
- Variants: 2-3 cover mockups at the size they'll appear on Amazon thumbnail (around 300px wide). PickFu lets you upload up to 8 but 3 is the sweet spot.
- Sample size: 50 is the cheap option; 200 is more statistically robust.
Time: 15 minutes to set up + 1-24 hours for results.
Reading PickFu results
You'll get:
- % preference per variant (e.g., A: 52%, B: 28%, C: 20%)
- Individual respondent comments explaining their choice
- Demographic breakdown
The percentages matter, but the comments matter more. Look for:
- Pattern of feedback — "the colour feels off-genre" vs "the typography looks amateur" vs "the woman's expression seems modern for the historical setting"
- Genre-mismatch flags — "this looks like a thriller, not a cosy"
- Confusion — readers saying they can't tell what genre the book is
A 60/40 winner with strong "this fits the genre" comments wins. A 55/45 split with confused comments means neither cover is working and you need a third concept.
What "winning" looks like
Strong wins:
- 60%+ preference for one cover
- Comments that converge on positive themes ("intriguing", "atmospheric", "I'd read that")
- Genre-fit confirmed
Weak wins:
- 50-55% split
- Mixed comments
- Some "neither" responses
If you have only a weak win, consider whether you've tested fundamentally different concepts or near-duplicates.
The two-test pattern
For high-stakes launches, a 2-test pattern produces the best results:
Test 1 (concepts): 3 fundamentally different cover concepts from designer. Find the winning direction.
Test 2 (refinements): 3 variants of the winning direction (different typography, slightly different colour palette, alternative image). Find the optimal execution.
Cost: ~£100-£160. Time: 1-2 days total.
Where most indies under-test
The cheap fix indies miss: re-testing the chosen cover at thumbnail size against your direct competitors.
After your designer's final cover is chosen:
- Create a mockup placing your cover at Amazon-thumbnail size next to your 3-5 closest comp titles
- Run a PickFu asking "Which of these books would you click on?"
- Include your cover + 3-4 real comp covers
If your cover loses to existing market leaders consistently — you have a problem to solve before launch.
What A/B testing doesn't fix
- Genre-wrong concept. Testing won't help if you're showing readers thriller covers and they're literary fiction readers.
- Below-genre quality. A great cover for £50 may still test worse than a competent cover for £600. Don't use testing to justify staying with low-quality designs.
- Title problems. If your title is the issue, no cover test reveals it. Test titles separately (PickFu also supports text-only tests).
UK-specific considerations
- PickFu has UK demographic filters — use them. Your UK reader avatar may prefer different cover styles than US-default respondents.
- UK readers index slightly higher on typographic + atmospheric covers than US (which leans photographic). Worth testing both.
- British-set fiction — when testing, include "interested in UK-set mysteries" as a filter for richer feedback.
- VAT on PickFu invoices: VAT-registered UK authors can claim back; below VAT threshold, treat as a deductible business expense.
Common mistakes
- Testing near-duplicate concepts. Test fundamentally different strategic concepts, not minor variations.
- Testing without a target demographic. Random respondents = noise.
- Ignoring the comments. Vote counts are noisy; comments are signal.
- Testing then ignoring the result. "My favourite lost — let me publish it anyway." Defeats the point.
- Testing only on PickFu. PickFu is one input. Real Amazon click data (Sponsored Products auto-campaign) tells you more once live.
- Skipping the thumbnail-size test. Cover that looks great at full size can fail at 200px thumbnail. Always test small.
When testing isn't worth it
- Standalone book with low marketing investment. If you'll spend £80 total on the cover, don't add £80 on testing.
- Repeat customer base (newsletter > 5k). Your audience clicks anything from you. Cover matters less.
- Genre with very strict conventions. Sweet romance has narrow cover conventions. Testing reveals little when the convention is rigid.
For most indies launching a debut commercial novel: testing is worth it. £30-£100 of PickFu is cheap insurance on a £500-£800 cover + thousands of pounds of marketing investment.
The bottom line
Run a PickFu test with 100-200 target-genre respondents before committing to a final cover. Use the comments more than the votes. Consider a Facebook Ads test for high-stakes launches. Don't test near-duplicates — test fundamentally different concepts.
A £50 cover test that prevents a £600 cover mistake is the easiest ROI in indie publishing.
Frequently asked questions
Can I A/B test on my own newsletter?
Yes — send a "help me pick the cover" email with poll. Your newsletter is biased (they already like you), but a strong divergence still signals something.
How does this differ from getting beta readers' opinions?
Beta readers know you and want to support you — biased. PickFu respondents don't know you and judge cold — what matters for Amazon discovery.
What if my favourite cover loses the test?
Trust the data unless you have specific reason to override (you know something the respondents don't). Most authors' personal favourites lose tests because their taste differs from their reader's taste.
Can I test cover + title together?
PickFu supports this but it muddies the signal. Test cover first, lock it, then test title-on-cover variants if needed.
Do PickFu results predict actual sales?
Imperfectly. Strong PickFu winners typically do convert better on Amazon, but the correlation isn't 1:1. Treat PickFu as one signal among several.
