Persuade in the user's favor, or not at all.
Behavioral design works, which is exactly why it needs a limit. Six principles, each one drawn to move a person toward something they already want, and each one gated by a hard line: honest persuasion that serves the user, never a dark pattern that extracts against them. The test is simple. If a technique only works while the user does not notice it, it does not ship.
A technique earns its place only when it moves a person toward an outcome they already want. If it works only unnoticed, it is out.
Reversible, transparent, true. A default is a starting point the user can change. A number on screen matches reality.
Fake scarcity, false countdowns, confirmshaming, inflated anchors. Prohibited in this system, not merely discouraged.
Smart Defaults
A blank form hands every decision to someone who came to do one thing. A smart default pre-selects the sensible choice so the path of least resistance is also the good path, always visible and one tap to change.
Rule Pre-select the choice the data already points to, and style it so it plainly reads as changeable.
Why The path of least resistance becomes the good path while the user keeps a one-tap way out.
Pre-fill the field the way the data already points, and style the pre-filled state so it reads as changeable, never locked. See Patterns / Form Elements for the input primitives.
Detected from your connection. Change it any time.
Default to the option a typical user would pick anyway, labelled and one tap to change. The default saves effort on a choice the user still owns.
Pre-checking a paid add-on or a marketing opt-in the user never asked for, so inattention becomes revenue. Preselection is banned wherever it costs money or privacy.
Endowed Progress
Effort accelerates as the finish comes into view. Open a progress bar at a true, non-zero point, counting real work already behind the user, and the end feels reachable. The rule is not to inflate the number, it is to count only what actually happened.
Rule Open a progress indicator at a true, non-zero point that credits only work the user actually did.
Why A real head start makes the finish feel reachable and pulls the user forward.
Open a multi-step flow with the first step already credited when the user genuinely did something (arrived with an account, verified an email). The fill uses --color-primary on a --color-border track. See Patterns / Interactive.
Credit steps the user truly completed and show how near the finish is. The head start is real, and the destination is one the user chose.
A "90 percent complete" bar that only fills once payment is handed over, or invented steps that manufacture sunk cost. Progress that does not reflect real progress is a lie with a nice animation.
Reciprocity
People return a favor. Deliver something genuinely useful first, then make a fair ask, and it lands on willing ground rather than a cold wall. The value stays whether or not they say yes.
Rule Deliver a genuinely useful result first, then make a fair ask the user is free to decline.
Why Value given up front lands the request on willing ground and keeps its worth either way.
Let the product prove itself before it asks. Return a real result, then invite the user to save it. The request comes after the gift, never as a toll gate in front of it.
Yours to read now. Create a free account only if you want to save it.
Give a genuinely useful result up front, then ask to save or continue. The gift keeps its value even if the user declines.
A hollow "free" gift that is really a hook, or a preview that vanishes unless you pay. If the value evaporates the moment the user hesitates, it was bait.
The IKEA and Endowment Effect
We value what we build and own more than the identical thing untouched. Let a person name a project or arrange a layout and it becomes theirs in their mind before it is theirs in fact, so signing up feels like keeping something they already have.
Rule Let the user personalize real work before any gate, then carry that work intact through sign-up.
Why People value what they build, so keeping their own work feels like keeping something they own.
Let the user personalize before the gate, then carry that work intact through sign-up so nothing they made is lost. Investment they chose to make is legitimate; investment tricked out of them is not.
Offer real customization the user wants, and carry their work across the sign-up boundary intact. They keep exactly what they built.
Forcing busywork to manufacture sunk cost, then holding the user's own work hostage behind a paywall. Effort must never become a hostage.
Loss Aversion Contested
A loss tends to weigh heavier than an equivalent gain, so framing a choice around what a person stands to lose can sharpen attention. Use it only where a real loss exists: an unsaved draft, an earned credit about to expire. Never invent one.
Rule Frame a choice around loss only where a real, preventable loss exists, and pair the warning with the fix.
Why A true loss sharpens attention honestly, while an invented one is manufactured fear.
Name the real thing at risk, plainly, and pair the warning with an easy way to prevent the loss. No invented deadlines, no fake stock counters.
You have 3 unsaved sections. Create a free account to keep them, or they clear when this tab closes.
Remind the user of value that truly exists and truly could be lost, with a clear way to keep it. The loss is real and preventable.
Fake "only 2 left" counters, invented countdowns, or confirmshaming ("No thanks, I like overpaying"). Manufactured fear is the sharpest dark pattern of all, and it is banned.
Contrast Effect
A number means little in isolation. Beside a true reference, a real list price, a genuine alternative, or the value it returns, the same cost reads as proportionate. Anchor only on a reference the user could verify.
Rule Show every price beside a reference the user could independently verify.
Why A number reads as proportionate next to a true comparison, and every figure on screen stays checkable.
Never show a price bare. Set it beside a genuine reference on a --color-surface-secondary card, with the offer on a --color-primary bordered card. Let the comparison inform, not corner.
Three separate tools you already pay for, billed individually.
All three, in one. A real, verifiable total.
Contextualize a price against a true reference: a real alternative cost, a genuine higher tier, or the value returned. Every number on screen is one the user could verify.
A struck-through "original" that never sold, or a fake premium tier built only to make the real one look cheap. A reference the user cannot verify is a lie.
The Line: Persuasion or Manipulation
The mechanics are identical on both sides of the line. What separates honest persuasion from a dark pattern is whose goal it serves and what it hides. Brignull coined "dark patterns" in 2010 for interfaces built to trick people into what they did not mean to do. This system treats that catalogue as a list of prohibitions.
Honest persuasion advances a goal the user already holds. Manipulation advances the business at the user's expense.
If it only works because the user does not notice, it is manipulation. Honest nudges survive being pointed out.
A default they can change, a step they can skip, a choice they can undo. Traps that are hard to escape are out.
Every scarcity claim, countdown, progress figure, and reference price is literally true. No inventions.
Before You Ship a Persuasive Element
Notes and Sources
- 1 · Smart defaults. The mechanism is effort-avoidance and status-quo bias, not "decision fatigue": the ego-depletion account behind that phrase failed a large multi-lab replication (Hagger et al., 2016). A default harms when it is not the choice an attentive user would make, so where there is no majority choice or the stakes are personal, leave it unset. Johnson & Goldstein, "Do Defaults Save Lives?" Science 302 (2003); Thaler & Sunstein, Nudge (2008); NN/g form design.
- 2 · Endowed progress. The lift depends on an honest reason for the head start; steps the user never took read as a trick once noticed. Count only real progress: a confirmed email, a saved draft, a completed step. Hull (1932); Kivetz, Urminsky & Zheng, JMR 43 (2006); Nunes & Dreze, "The Endowed Progress Effect," J. Consumer Research 32 (2006).
- 3 · Reciprocity. Regan's effect is well established for offline, interpersonal compliance; "value-first content raises product sign-ups" is a reasonable but untested extrapolation, not something Regan or Cialdini measured. SPA Cialdini, Influence (1984); Regan, J. Experimental Social Psychology 7 (1971).
- 4 · IKEA and endowment. The effect needs successful completion: unfinished or trivially easy building does not raise valuation. Whether self-customization specifically lowers sign-up abandonment is an inferential step beyond these studies. SPA Norton, Mochon & Ariely, J. Consumer Psychology 22 (2012); Kahneman, Knetsch & Thaler, J. Political Economy 98 (1990).
- 5 · Loss aversion (contested). Treat it as real but bounded, not a universal law. Gal & Rucker (2018) argue the asymmetry is weaker and context-dependent; Simonson (2018) defends a contingent version. Use loss framing only where a genuine loss exists, and never assume fear out-pulls a clearly stated gain. Kahneman & Tversky, "Prospect Theory," Econometrica 47 (1979); Gal & Rucker, J. Consumer Psychology 28 (2018); Simonson (2018).
- 6 · Contrast and anchoring. Anchoring is potent, which is why the reference must be true: a comparison price never charged crosses from context into deception and, in many markets, unlawful pricing. Tversky & Kahneman, Science 185 (1974); Ariely, Loewenstein & Prelec, "Coherent Arbitrariness," Q. J. Economics 118 (2003).
- Dark patterns. Brignull originated the term (deceptive.design); Gray et al. formalized the taxonomy ("The Dark (Patterns) Side of UX Design," CHI 2018); NN/g tracks it as deceptive patterns. The strict truth-and-reversibility gate around each technique is this system's house doctrine, not a claim from the cited authors. SPA