Research on CRO, SEO, GEO and ASO
Each note sums up what a study or a standard says, where it stops, and what to check on your store. Pick a topic, or read down the list.
Showing every note, grouped by topic.
- 10–20%of Bing ideas succeeded A/B test findings2 min readMost A/B tests lose: what large testing programs reportBefore you trust a test on your store, look at what happened to the people who ran thousands of them.
- 5×how far false positives can rise if you keep peeking A/B test findings2 min readStop checking your A/B test every morningThe dashboard says “95% chance to win” on day three. Published research says that if you stop there, the number cannot be trusted.
- 18.7%more volume to offset a 5% price cut, McKinsey found AOV & upsells3 min readDiscount timing and depth: what McKinsey, Bain and Deloitte foundA sale is a price cut with a depth, a start and an end. Consulting-firm research on large retailers shows why all three are worth getting right.
- 20–50%of promotions showed no noticeable lift, in BCG’s work AOV & upsells3 min readHow many promotions pay back? What BCG, McKinsey and Bain foundConsulting-firm work finds that a large share of promotions deliver no lift or lose money, and that fewer, clearer promotions can do better.
- 12–20%of e-commerce revenue can go on fulfilment AOV & upsells3 min readBundles and basket size: what the research says about letting shoppers chooseFulfilment is a large share of e-commerce revenue, so bigger baskets matter. Peer-reviewed and consulting research on bundles gives a hopeful sign and a clear warning.
- 40%of US shoppers who abandoned a cart, excluding those just browsing, said “extra costs too high” AOV & upsells3 min readFree-shipping thresholds and order value: what Baymard, McKinsey and the journals foundThe usual advice is to set the threshold above the typical basket. Four sources give a more careful picture of orders, order value and returns.
- “X for $Y”did better than “X + N free” AOV & upsells3 min readMulti-unit offers: what the research says about “Buy 3 for $Y”Two peer-reviewed studies on multi-unit price promotions, and what BCG and McKinsey say a promotion can cost.
- 83%of sites don’t apply discounts automatically, Baymard found AOV & upsells3 min readPrice visible before checkout: what Baymard and the coupon studies sayBaymard’s checkout research on code fields and cart abandonment, plus a university study of shoppers asked for a code they did not have.
- 38studies put the decoy effect to the test AOV & upsells3 min readThe decoy price: why the famous trick is shakier than it soundsPricing advice often tells store owners to add a “decoy” option so shoppers pick the bigger bundle. The research behind that advice is more limited than the guides suggest.
- 62–90%a colour statistic with no study named Colour psychology3 min readColour psychology: what the research supports and what it doesn’tTwo claims show up in almost every article on colour and conversion. Neither stands on evidence we could verify.
- 30nations shared many colour-emotion pairings Colour psychology3 min readWhat colour research says about brand personalityThe best-supported use of colour in a store is saying who you are, not pushing a click. Here is what three published sources report, and where they stop.
- 4.5:1is the line for body text Contrast & accessibility2 min readContrast is a number: the WCAG ratios, and two common missesWhether text is readable enough is not a matter of taste. It is a ratio you can compute from two colour codes, and the web’s accessibility standard says where the line is.
- 24pxAA minimum for a tap target Contrast & accessibility3 min readTap targets: the size WCAG requires, and what thumb studies foundMost store traffic is a thumb on a small screen. The accessibility standard puts a number on how big a button must be, and one old study tells you why the number matters.
- 80characters is WCAG’s ceiling for a line Fonts & breathing space3 min readLine length and line height: what the standard says, and what the research addsTwo settings decide how a block of text feels before anyone reads a word: how wide the lines are and how far apart they sit. The accessibility standard gives a number for each. Here is where those numbers come from, and where the evidence is thinner than the numbers suggest.
- 0.12of the font size, as letter spacing Fonts & breathing space3 min readSpace between letters and words: can your page survive a reader’s settings?The text-spacing rule in WCAG does not tell you how to space your type. It asks something different: if a reader widens the spacing, does your page still work?
- 6pairs of phrasings typed into autocomplete SEO2 min readWrite the words buyers type, not the words you valueA product or collection title should carry the words a shopper types. We compared six pairs in Google’s autocomplete to see how far apart the two can be.
- 2addresses for one Shopify product SEO3 min readOne product, two URLs: what Shopify and Google say to doOn a Shopify store, one product can be reached at more than one address. Google has a plain answer for how it treats that, and Shopify says most of the work is already done for you.
- 4things you control GEO & LLM search3 min readWhat AI assistants can read about your store, and the four things you controlNobody can promise that ChatGPT, Perplexity or Claude will mention your store. You can control what they are allowed to read and how easy your facts are to find.
- 40%best visibility gain in one study GEO & LLM search3 min readWhat the research says about getting cited in AI answersThere is one peer-reviewed benchmark for “optimizing” a page for AI answers. It is useful, and it is also narrow. Here is what it found and where it stops.
- 30characters for an app name ASO2 min readThe four text fields Apple’s App Store search readsApple says App Store search weighs text relevance and user behavior. You can’t write downloads or reviews, but you do write the text.
- 3screenshots can show in a search result ASO3 min readWhat shows in App Store search results: screenshots, previews and ratingsSearch results show more than a name and a subtitle. Apple’s own pages say which images and ratings appear there, so the first screenshots are part of your listing’s search presence, not an afterthought.