Less famous conversion psychology: what the evidence supports, and what UK law rules out
TL;DR
Difficulty: Intermediate
- Less famous does not mean better proven. Goal gradients, implementation intentions, defaults and social comparison have solid evidence. Identity framing failed a large replication.
- Show progress and work only when it is real: fake progress bars and invented searches are misleading actions.
- Defaults are powerful and regulated: no pre ticked consent, no pre ticked paid extras.
- Scarcity and deadlines must be true. False limited time claims are banned outright under DMCCA Schedule 20.
- The 14 day right to cancel is the law, not a perk. Offer more and say what is extra.
Beyond scarcity and social proof sits a set of less famous effects that genuinely help conversion and retention, and a set of tactics sold under the same names that are now illegal in the UK. This article takes twelve of them back to the original studies, says how well each has held up, works each through on an invented British business, and marks the legal line.
Beyond scarcity and social proof
The famous psychology principles get all the attention: scarcity, social proof, loss aversion. Every marketer knows them, so every competitor uses them. The effects in this article are less famous. Some come from papers that never made it into the popular business books. Others are well known in one field and rarely used in marketing. I have used these ideas across eCommerce, SaaS and B2B products.
Less famous does not mean better proven. For each effect below I give the original study, how well it has held up when others tried to repeat it, a worked example, and, where it matters, the UK legal line. Several "conversion tactics" that circulate under these names, such as fake progress bars, invented deadlines and pre ticked boxes, are now plainly illegal in the UK. They are in here as what not to do. Every number in the worked examples is invented so I can show each step.
Goal gradient: people speed up near the finish
The goal gradient hypothesis says that effort increases as a goal gets closer. Ran Kivetz, Oleg Urminsky and Yuhuang Zheng tested it on a real café loyalty card in 2006. Customers bought coffee more often as they approached the free one. And the illusion of progress worked too: customers given a 12 stamp card with two "bonus" stamps already filled in completed the 10 purchases they needed faster than customers given a plain 10 stamp card. Customers who accelerated more were also more likely to stay in the programme.1
The finding comes from field data with a clear design, which makes it more convincing than most.
Worked example: two cards at Kemptown Coffee
Kemptown Coffee is an invented café in Brighton. Card A has 10 empty boxes. Card B has 12 boxes with two already stamped. Both need 10 paid coffees for the free one, so the offer is identical. On the first day, card A shows progress and card B shows . The research says card B gets completed faster.
To see the gradient in your own data, look at the gap between purchases. Say the average gaps between stamps one to ten are 6.0, 6.0, 5.8, 5.6, 5.4, 5.2, 5.0, 4.8 and 4.6 days. The first four gaps average 5.85 days, the last four 4.9 days, so customers come back sooner near the reward. A flat line would mean the card is not motivating anyone.
Where it goes wrong. The pre stamped card is honest because the offer really is ten coffees. A progress bar that claims progress the customer has not made is a different thing: a misleading action under consumer law.2
Fresh starts: timing around temporal landmarks
Hengchen Dai, Katherine Milkman and Jason Riis found that people pursue goals more after "temporal landmarks", the start of a week, a month, a year or a semester, a birthday, a holiday. Google searches for "diet" rose at the start of a new week by 14.4%, a new month by 3.7% and a new year by 82.1%. At a university gym, the probability of a visit rose by 33.4% at the start of a week, 14.4% at the start of a month and 47.1% at the start of a semester.3
How well it holds up. These are large archival datasets, but they are observational: they show when people act, not that a campaign sent on a Monday will convert better. The authors themselves offer the mechanism, a clean mental slate, as an explanation rather than a proof. The figures are about dieting and exercise, not purchases. They do not support the "20% to 30% lift on the right days" that is often quoted for marketing.
Worked example: a planning email on the first of the month
Clifton Bookings is an invented booking software company in Bristol for salons. It wants salon owners to set up their monthly availability in advance. The test: half the owners get the "plan next month" email on the first working day of the month, half on a random weekday. Everything else is identical. The metric is the share who set up availability within seven days. If the landmark version wins by a margin the test can detect, keep it. If not, the landmark was a story, not a lever.
Identity framing: contested
Christopher Bryan and colleagues reported in 2011 that asking people "how important is it to you to be a voter" raised turnout compared with asking about "voting". It was an appealing idea: a noun invokes identity, a verb an action.
How well it holds up. Badly, so far. In a large field experiment Alan Gerber and colleagues found that the noun wording had no effect on turnout, and that both versions did worse than a standard get out the vote message.4 The original authors argued that the replication did not recreate the right psychological context. Treat identity framing as a copy idea to test, not a proven effect.
An example: Clifton Bookings could test "Be the salon that never double books" against "Stop double bookings" in its sign up page headline, with sign ups as the metric. It costs one test, and either answer is useful.
The curse of knowledge: why experts write unclear copy
Once you know something, it is hard to imagine not knowing it. Economists Colin Camerer, George Loewenstein and Martin Weber named this the curse of knowledge in 1989.5 The best known demonstration is Elizabeth Newton's tapping study at Stanford in 1990: people tapped out the rhythm of well known songs for listeners to guess. Tappers predicted a 50% success rate. Listeners got 3 of 120 songs, 2.5%.6 The tappers could hear the melody in their heads. The listeners heard knocking.
That is why founders often write landing pages only they can follow. One simple check is the Flesch Reading Ease score, which Microsoft Word reports and defines as:7
where is the score (higher is easier, and 60 to 70 suits most standard documents), is the number of words, the number of sentences and the number of syllables. So is the average sentence length and the average syllables per word. Both are ratios, explained in Foundations: ratios and rates.
Worked example: two versions of one headline
The founder's version: "Our omnichannel orchestration layer unifies heterogeneous inventory data." That is 8 words, one sentence, and about 26 syllables. The score is . Below zero.
The plain version: "See every product's stock in one place, whether it sells online or in the shop." That is 15 words, one sentence, 20 syllables. The score is .
Same product, same promise. The second one can be read by someone who has never heard the word omnichannel, which is most of the people who would buy it. A score is only a check, not a goal: the real test is a reader who does not know the product.
Implementation intentions: when, where and how
An implementation intention is a plan in the form "when situation X arises, I will do Y". Peter Gollwitzer and Paschal Sheeran's meta analysis of 94 independent tests found a medium to large effect on reaching goals, a standardised difference of .8 That is a standardised effect size, not a doubling of follow through, whatever the popular summaries say.
Worked example: the first appointment at Clifton Bookings
New Clifton Bookings accounts that add a first appointment in week one tend to stay. So the onboarding asks one extra question: "When will you add your first client? Pick a day and time," and sends a reminder at that moment. The control group gets the usual "Add your first client" button. The metric is the share who add a client within seven days. This turns a vague intention into a specific plan, which is exactly what the research tested.
Small commitments first: the foot in the door
In Jonathan Freedman and Scott Fraser's 1966 study, people who first agreed to a small request were far more willing to agree to a large one later. In one condition, 76.0% of those who had agreed to a small first request about safe driving later agreed to put a large "Drive Carefully" sign in their garden, against 16.7% of those asked only the large request.9 The explanation offered is self perception: having said yes once, people see themselves as someone who supports the cause.
The product version is a commitment ladder: a free tool, then a trial, then a paid plan, each step small relative to the one before. An example: Clifton Bookings offers a free public booking page first, then a 14 day trial of reminders, then the paid plan. Each step asks for a little more and gives something real.
Asking happy customers for a review is a natural small step too. Writing fake reviews, or publishing incentivised ones without saying so, is a banned practice in the UK.10
Show the real work: the labour illusion and operational transparency
Ryan Buell and Michael Norton found in five experiments, in online travel and online dating, that when a website signals the effort it is making, people can prefer a longer wait to an instant result, even when the results are identical. They called it the labour illusion.11 Later field and lab experiments by Buell, Kim and Tsay in food service found that letting customers see the work, and letting staff see the customers, raised customer reported quality by 22.2% and cut throughput times by 19.2%.12
You will sometimes see this sold as "fake a progress bar". Don't. A progress message that claims work which is not happening is false information about the service, which is what the misleading actions rules exist to stop.2 Show the work that genuinely happens, and the evidence says it helps.
Worked example: a gift finder that shows its working
Northgate Hampers, an invented hamper shop in Leeds, has a gift finder. It genuinely filters 140 hampers by budget, dietary needs and delivery date. Instead of jumping straight to results, it shows three short, true steps: "Checking 140 hampers against your £60 budget", "Removing 23 that contain nuts", "Keeping 31 that arrive by Friday". Each line reports something the code actually did, in a second or two. That is operational transparency. A spinner that says "Searching 300 suppliers" when the shop has one warehouse is not.
Defaults: the most powerful setting, and the most regulated
People tend to stay with whatever is preselected, through inertia and because a default reads like a recommendation. Brigitte Madrian and Dennis Shea showed this with pension saving in 2001: participation in a 401(k) plan was significantly higher under automatic enrolment, and many employees kept both the default contribution rate and the default fund, which few had chosen before.13
Defaults are exactly where UK law draws hard lines:
- Consent to marketing or data use. Under the UK GDPR, as the Information Commission (still the ICO on its website) explains, silence, pre ticked boxes or inactivity are not consent. Someone has to opt in.14
- Paid extras. Under the Consumer Contracts Regulations 2013, a trader needs the consumer's express consent to any payment beyond the main price, and consent inferred from a default option such as a pre ticked box does not count.15
Worked example: three defaults at the Northgate checkout
- Newsletter box, pre ticked. Not valid consent. Leave it unticked and make the benefit clear instead.
- Gift wrap at £3, pre ticked. No express consent to the extra payment, so it cannot be charged on the strength of the tick. Offer it unticked, with a picture.
- Standard delivery preselected, with its price shown. Fine. It is a sensible default the customer can see and change, and it serves most people.
The rule I use: a default should be the choice most informed customers would make for themselves. If it only works because people do not notice it, it is the wrong default.
Status tiers: how to structure a loyalty programme
Xavier Drèze and Joseph Nunes studied how the structure of loyalty tiers affects feelings of status. Making the top tier bigger dilutes the status its members feel. Adding a tier below it enhances their status. People who do not qualify for status prefer programmes with several tiers, and labels such as gold and silver on their own signal a selective hierarchy.16
Worked example: a hamper club with two tiers
Northgate has 20,000 customers. A single "VIP" tier for the top 20% (4,000 people) would feel ordinary to its members. Following the research, Priya instead makes Gold the top 5% (1,000 customers) and adds Silver for the next 15% (3,000). The Gold members feel more exclusive because Silver exists below them, and the 3,000 Silver members have a visible next step. The rewards themselves can be modest, early access to the Christmas range, a free card with every order, because the structure carries part of the motivation.
Social comparison: show people what their peers do
Hunt Allcott evaluated the Opower home energy reports, which compare a household's electricity use with its neighbours'. Across randomised experiments with 600,000 households, the reports cut consumption by 2.0% on average, an effect equivalent to a short run electricity price rise of 11% to 20%. The heaviest users cut the most: the top tenth reduced use by 6.3%.17 Note what that comparison is: a price rise, not a financial appeal in a letter.
Worked example: reminders at Clifton Bookings
Clifton Bookings sees that salons sending text reminders have fewer no shows. Its monthly report to each salon adds one line: "Salons of your size send reminders for 9 in 10 bookings. You send them for 4 in 10." It is a descriptive norm, true and specific, about something the salon controls. As in the Opower data, expect the biggest change from the salons furthest from the norm, and measure it against a group that gets the report without the line.
Scarcity and exclusivity: only when true
A waitlist, an invitation, a members' area or a closing date can make something more wanted. Used honestly, that is fine. The legal line matters here more than anywhere, because this is where most of the trouble starts. I cover the psychology itself in scarcity, urgency and loss aversion.
Under the Digital Markets, Competition and Consumers Act 2024 (DMCCA), falsely stating that a product, or particular terms, will only be available for a limited time, to rush a decision, is banned outright.18 The Competition and Markets Authority's guidance gives the example of a countdown clock that restarts when it runs out. If the offer really ends, that is fine, but a substantially similar offer reappearing soon afterwards can still mislead.19 False stock claims ("only 2 left" when there are 200) fall under misleading actions.2
An example: Northgate's corporate gifting programme genuinely takes 30 new business accounts each autumn, because that is what the packing team can handle. Saying so is true and useful. Saying it in February to an account that could sign up any day of the year is not.
Risk reversal: guarantees and returns
A lenient returns policy shifts the risk of a purchase from the buyer to the seller. Narayan Janakiraman, Holly Syrdal and Ryan Freling's meta analysis of 21 papers found that, overall, leniency raises purchases more than it raises returns. Lenient terms on money and effort (full refunds, no hassle) increase purchases, while lenient scope increases returns and longer time limits actually reduce them.20
There is a UK catch. Online shoppers already have a legal right to cancel, normally ending 14 days after the goods arrive.21 Presenting rights consumers have by law as a distinctive feature of your offer is banned.22 So "14 day money back guarantee" on a website is not a guarantee, it is the law, and selling it as a perk is unlawful. Go beyond it, and say clearly what is extra.
Worked example: a 30 day return window
Gloucester Road Home is an invented homeware shop in Bristol. It sells 1,000 orders a month at an average of £68 and 4% are returned. Net sales are . It tests a 30 day, free returns window, described as "30 days to return, on top of your legal right to cancel". Say orders rise 3% to 1,030 and returns rise to 5%. Net sales become , an extra £1,258 a month, before the cost of the extra return postage. That last line decides it, so put it in the sum before you roll it out.
An honest audit
Here is the whole list in one place, with how much weight each deserves and where the legal line sits.
| Effect | Evidence | Legal line in the UK |
|---|---|---|
| Goal gradient | Field experiment, persuasive | Progress must be real |
| Fresh start | Large observational datasets, not purchases | None specific |
| Identity framing | Contested, failed a large replication | None specific |
| Curse of knowledge | Classic demonstrations | None specific |
| Implementation intentions | Meta analysis, medium to large effect | None specific |
| Foot in the door | Classic experiment | Reviews must be genuine and disclosed |
| Labour illusion | Several experiments | Only show work that happens |
| Defaults | Strong field evidence | No pre ticked consent or paid extras |
| Status tiers | Experiments on structure | None specific |
| Social comparison | Large randomised trials | Comparisons must be true |
| Scarcity and exclusivity | See the scarcity article | False time limits banned outright |
| Risk reversal | Meta analysis | Legal rights are not a perk |
For each effect you want to try, write down the metric, the size of change worth acting on and the sample you need before you start, then run it against a control. The complete guide to testing methods covers how.
The playbook is the testing
These effects are less famous than scarcity and social proof, and some of them are genuinely useful. Most of the popular versions overstate them, and a few of the popular tactics are now illegal. The work is not collecting tactics. It is choosing the few that fit your customers, building them honestly, and measuring whether they did anything. That is what separates growth from a pile of random experiments.