Marketing Psychology: What the Evidence Supports

Four buying decisions, sorted by what actually survived replication.

Marketing psychology is the worst-sourced subject in our industry.

Invented percentages. Studies misattributed to the wrong author. Effects that failed replication a decade ago, still taught in slide decks as settled fact.

So this page is organised by the four buying decisions you are trying to shape, not by a list of biases — and for each named effect, what the evidence currently supports. Where one is contested, we say so. Where we could not establish its replication status, we say that too.

It sits under our conversion rate optimisation guide, which covers the testing process itself. This page is only about the psychology you can defend in a room.

Why most marketing-psychology advice fails

Three failure modes account for nearly all of it.

• The lab-to-landing-page jump. An effect found in a numeric questionnaire gets presented as a rule about product pages. Those are different conditions.
• The orphan statistic. A figure circulates with no author and no year. If you cannot trace it, you cannot use it.
• The frozen citation. A 1999 or 2000 study keeps being quoted while the meta-analyses and replications published since are ignored.

The practical rule: treat every principle below as a hypothesis for your own audience, and test it. Your audience definition belongs in your buyer persona work; the test design belongs in A/B testing.

Decision 1: making a choice easier

What the principle says. People choose more readily from a smaller, better-organised set, and most people accept whatever option is pre-selected.

The honest evidence. The famous demonstration is the jam booth: Iyengar and Lepper (2000, Journal of Personality and Social Psychology) set out 6 jams or 24, and about 30% of shoppers who stopped at the six-jam display bought a jar against about 3% at the 24-jam display.

That result did not generalise. Scheibehenne, Greifeneder and Todd (2010, Journal of Consumer Research) pooled 63 conditions from 50 experiments and 5,036 participants and found a mean choice-overload effect of virtually zero, with wide variance between studies. They identified possible preconditions but no sufficient ones.

So "fewer options always converts better" is not supported. What is defensible is narrower: when options are hard to compare, comparison is the cost, and reducing or structuring them removes it.

Applied on a page. Three plans with the differences named in the same words on every row. One pre-selected plan, labelled as the recommended one and freely changeable. A guided filter rather than a 40-SKU grid. The structural side of this is covered in pricing page design.

Where it crosses the line. Pre-ticking a paid add-on in the cart. That is not a default, it is a charge the buyer did not choose.

Decision 2: making a price feel reasonable

What the principle says. A price is judged against whatever comparison the buyer has to hand. Change the comparison set and the same number reads differently.

The honest evidence. Anchoring is one of the sturdier findings here. In the Many Labs replication project (Klein et al., 2014, Social Psychology), all four anchoring tasks replicated in every sample, with effect sizes above the original estimates. In the same project, flag priming and currency priming did not replicate at all — a useful reminder that "it's psychology" is not a quality signal.

The decoy is weaker than its reputation. Frederick, Lee and Baskin (2014, Journal of Marketing Research) ran 19 studies with realistic options — products people tasted, photographs of quality levels — and found no attraction effect; it reappeared only when the same options were shown as numbers. Their conclusion: such effects do not typically obtain when consumers experience the product.

Applied on a page. Publish a real range instead of "request a quote". Put the price next to what it replaces — a month of agency retainer against a month of in-house salary. In India, content and SEO retainers commonly run ₹25,000–₹1,50,000 a month and paid-media management ₹25,000–₹1,00,000, fees excluding ad spend.

Where it crosses the line. A permanent "was ₹9,999" that nobody was ever charged. That is a pricing claim you cannot defend to a reviewer, a journalist or a regulator.

Decision 3: making a stranger trust you

What the principle says. People take behavioural cues from others, and the cue works best when those others are recognisably like them.

The honest evidence. This is the strongest-sourced section on the page. Goldstein, Cialdini and Griskevicius (2008, Journal of Consumer Research) ran hotel field studies comparing signs that described what most other guests did against the conventional environmental appeal. The descriptive-norm signs performed better, and the strongest version described guests who had stayed in that same room — what the authors call a provincial norm.

The lesson is about specificity, not volume. Proof from people the reader recognises beats a bigger number from strangers.

Reciprocity is shakier than the marketing books suggest. Many Labs did replicate a reciprocity-norm effect, but it was a survey-response effect about fairness judgements. Whether a free download on a landing page triggers the same obligation is a separate question, and we have not seen it answered.

Applied on a page. Named reviewers with a date, a role and a company, grouped by segment so a manufacturing buyer reads manufacturing reviews. Numbers you can audit. Case studies that name the constraint, not only the outcome. Placement sits with landing page design, and the words with conversion copywriting.

Where it crosses the line. Written reviews. Stock-photo faces. A "12 people are viewing this" counter produced by a random number generator.

Decision 4: making someone act now

What the principle says. A real constraint on time or supply pulls a decision forward that would otherwise drift.

The honest evidence — and this is where it gets uncomfortable. Loss aversion, the claim that losses loom larger than equivalent gains, is actively contested. Gal and Rucker (2018, Journal of Consumer Psychology) reviewed the evidence and concluded that it does not support losses being, on balance, more impactful than gains. They treat its persistence as a question about scientific belief rather than about data.

The deadline literature took a harder hit. In Data Colada post 138 (Simonsohn, Simmons and Nelson, 31 August 2026), the authors set out evidence that the Study 2 data in Ariely and Wertenbroch's widely cited 2002 precommitment-deadline paper were, in their words, severely tampered with or fabricated. It also reports a published failure to replicate that study by Hyndman and Bisin: demand for precommitment survived, the demonstration that imposed deadlines improve performance did not.

Scarcity we will not cite. The effect is plausible and widely taught, but we could not establish a replication status for it this run, so treat it as unsettled rather than proven.

Applied on a page. Deadlines that exist for a reason you can state out loud: a cohort that starts on a date, stock that is genuinely finite, a quarter-end that closes a price. Enforce them. A deadline you extend twice has taught your list to ignore the next one. If you want a timed prompt at exit, exit-intent popups covers the honest construction.

Where it crosses the line. A countdown that resets when the page reloads.

The four principles, honestly and dishonestly applied

The ethics line, stated plainly

Everything in the third column above works in the short run. That is exactly why it is a problem — it pays before it costs.

Digital Hangover does not build any of it — no resetting countdowns, no fabricated viewer counts, no written reviews, no confirmshaming on an opt-out. Not as a positioning line: the lift is temporary and the trust cost is not. The audience learns, and every honest claim you make afterwards has to work harder.

Regulators have taken an interest. In India, the Central Consumer Protection Authority notified its Guidelines for Prevention and Regulation of Dark Patterns on 30 November 2023, naming specific practices including false urgency, confirm shaming, drip pricing and subscription traps. We are not lawyers and this is not legal advice — if your site uses any of these, that is a conversation for your counsel, not for your growth team.

What we see most often in CRO work

First-hand observation from Digital Hangover's CRO engagements. This carries no measured uplift figure, and we are not going to attach one to it.

The change we suggest most often is purely psychological, not technical: rewrite the line above the submit button and the button itself so that together they state exactly what happens next. Who gets in touch, how soon, through which channel, and whether it costs anything. No new field, no new script, no redesign — only the removal of an unknown.

The change clients resist most is removing options. Cutting six plans to three, or four form fields to two, reads to stakeholders as giving up reach or data. It is the hardest recommendation to get signed off, and the one we most often end up proving through a test rather than a conversation.

Where to go from here

Pick the decision your page is actually asking the visitor to make, then apply only the principle attached to it.

1. Name the decision. Choice, price, trust or timing. One page usually carries one.
2. Apply the version the evidence supports, not the version in the slide deck.
3. Write down what you expect to happen before you ship it, so a null result is readable.
4. Test it on your own audience. Nothing above transfers automatically.
5. Delete anything you would not explain to the buyer if they asked how it worked.

If you want this applied across live campaigns and landing pages rather than one page at a time, that is the work our performance marketing team does. Organic and CRO gains typically take three to six months to show meaningfully; paid moves faster and stops when the spend stops.