Users remember and rate as better the first and last items in any sequence. Position shapes perceived quality, trustworthiness, and which choice feels like the obvious one — even when the content is identical.
Hermann Ebbinghaus documented the serial position effect in 1913, but researchers have since discovered something that goes beyond memory: position shapes perception itself. In wine tastings where identical wine is poured in different orders, the first and last glasses are rated significantly higher — not because the wine is better, but because the brain assigns quality to items at the edges of a sequence. The same effect operates in every list, every set of options, and every sequence of information a user encounters in a product.
This means position is not neutral. When you order a list of testimonials or AI-generated options, you are making a decision about which items users will remember as better, trust as more credible, and be drawn to select — even when the actual content is identical. Position is a design tool. Used deliberately, it steers perception and decision toward the outcomes you want. Used carelessly, it steers users toward the wrong things by accident.
“The same wine rated at position one or five will be scored higher than at positions two, three, or four — by the same person, on the same day.”
— Mantonakis et al., Psychological Science, 2009
When a potential customer reads a set of testimonials, they don't average them. They remember the first and the last, and those two reviews anchor their overall impression of the product. The middle reviews are processed and forgotten — their content doesn't meaningfully influence the final judgment.
Both sets of testimonials below are identical — same four reviews, same star ratings, same authors. The only difference is order. Notice how differently you feel about the product after reading each set.
First impression: “decent, interface could be better.” Last impression: “great product, would recommend” — too vague to stick. The two strongest reviews are buried in the forgettable middle.
Same four reviews. First impression: “+23% conversion rate.” Last impression: “I actually look forward to using it.” The lukewarm review sits at position 3 where it provides balance without anchoring memory.
The lukewarm review isn't removed from the good version — removing it would feel dishonest and reduce credibility. Instead it's moved to position 3, the slot where users are least likely to assign it undue weight. Balance is maintained; perception is shaped.
When an AI presents a list of options — names, subject lines, copy variants, design directions — users treat position one as the implicit recommendation. Research on choice from AI-generated lists shows selection rates for option one are often two to three times higher than for identical options placed in the middle.
The bad example puts a safe, forgettable option first. Users who pick “option 1” by habit or by primacy get the weakest choice. The good example leads with the strongest suggestion, making the default selection behaviour align with the best outcome.
“Flowstate” — the best name — buried at position 3. Users who instinctively pick option 1 get the weakest, most generic choice.
“Flowstate” leads at position 1. Users who pick by primacy get the best choice. The AI's confidence is signalled by position.
The “Recommended” label in the good example is not the mechanism — the position is. Even without the label, option one would be selected at a higher rate simply because of serial position. The label makes the implicit signal explicit, which is the honest version of using this effect deliberately.
Every ordered list in your product is a decision about perception. Testimonials, feature lists, AI options, button groups, navigation items — each one has a position one, a position last, and a forgettable middle. The question is whether those positions are filled deliberately or by accident.
The practical audit: go through every list in your product and ask — what do I want users to remember from this set, and what do I want them to choose? Put the answer to the first question at position one. Put the answer to the second question at the last position before the action. Everything that needs to be present but shouldn't dominate goes in the middle.
Ebbinghaus, H. (1913). Memory: A contribution to experimental psychology. · Mantonakis, A., Rodero, P., Lesschaeve, I., & Hastie, R. (2009). Order in choice: Effects of serial position on preferences. Psychological Science, 20(11), 1309–1312. · Carney, D. R., & Banaji, M. R. (2012). First is best. PLOS ONE, 7(6).