Every persuasion technique that helps someone decide faster can also help them decide badly. That is not a reason to abandon persuasion — a world where nobody could be persuaded of anything would be a worse one — but it is a reason to be deliberate about which techniques you use, and to have a test you apply before you use them.
Here are the eight principles I rely on in client work, the evidence behind each, the test each one has to pass, and the four techniques I will not write no matter who is paying.
The line between persuasion and manipulation
The distinction is not about intensity or emotion. Plenty of manipulation is calm, and plenty of honest persuasion is emotional. The distinction is about information and agency.
Persuasion gives someone accurate information about a real thing, arranged so they can understand it, and leaves the decision with them. Manipulation exploits a predictable flaw in how people reason to produce a decision they would not make if they could see clearly.
The practical version: would this still work if the reader knew exactly what I was doing and why? A well-organised argument survives that question. A fake countdown timer does not.
The three-question ethics test
Before I use any technique, it has to pass all three:
- Is it true? Not technically defensible — true. Does the scarcity exist? Did the customer say that? Is the number measured?
- Would it survive disclosure? If the reader could see the mechanism, would they still find the argument reasonable, or would they feel had?
- Does it serve their decision or only mine? Does this help them judge whether the thing is right for them, or does it only make them act faster regardless?
Anything that fails one of these gets cut, even when it would work. Especially when it would work.
1. Reciprocity
What it is: People feel an obligation to return value they have received. It is one of the most consistently replicated findings in social psychology, and it operates well below conscious awareness.
The honest version: Give something genuinely useful before you ask for anything, and give it without conditions. A free audit checklist that would improve the reader’s site whether or not they ever hire you. A teardown that names the real problem even when the fix is not something you sell.
The dishonest version: A “free guide” that is a brochure, an audit that mysteriously always concludes you need the premium package, or value withheld until an email address is surrendered.
Test result: Passes easily when the free thing has standalone value. The tell is simple — would you be happy for someone to take the free thing and never come back?
2. Social proof
What it is: In conditions of uncertainty, people look to the behaviour of similar others to decide what is reasonable. The “similar” part carries most of the weight and is the part most often ignored.
The honest version: Specific, attributed, and matched to the reader. A testimonial from a sole trader persuades another sole trader; a logo from a multinational persuades them of nothing except that they are not the target. Numbers with context beat adjectives: “used by 41 clinics in the Waikato” beats “trusted by hundreds”.
The dishonest version: Stock-photo testimonials. Invented names. Cherry-picked results presented as typical. “Join thousands” when it is four hundred.
Test result: Passes when attributed and representative. Fails immediately when the result shown is an outlier presented as the norm — that is a claim about what the reader can expect, and it is false.
3. Loss aversion
What it is: Losses generally loom larger than equivalent gains. The precise magnitude is debated and the effect is weaker in some contexts than early work suggested, but the direction is robust enough to write with.
The honest version: Quantify the cost of the status quo the reader is already paying. “At your current conversion rate, roughly 380 of the 400 people who visit your pricing page each month leave without contacting you.” That is not a threat. It is a fact they own and had not calculated.
The dishonest version: Manufactured consequences. “Every day you wait, your competitors pull further ahead” — unless you can show that, it is a sentence designed to produce anxiety rather than understanding.
Test result: Passes when the loss is real, current and calculable. Fails when it is speculative and unfalsifiable.
4. Commitment and consistency
What it is: Once someone takes a small step, they are more likely to take a larger one consistent with it. Self-perception does much of the work: having acted, people update their view of who they are.
The honest version: Design a genuinely low-cost first step that is useful in its own right — a self-scoring quiz that returns a real diagnosis, a checklist they can run today, a 30-minute call with no pitch. The step earns its place by helping.
The dishonest version: Multi-step forms that hide the real ask until step four, after sunk-cost has set in. Free trials that require a card and bury cancellation.
Test result: Passes when every step is disclosed up front and each has independent value. Fails whenever the sequence relies on the reader not knowing where it ends.
5. Authority
What it is: People defer to signals of expertise, particularly under uncertainty. The signals are often cheaper to fake than the expertise, which is precisely the problem.
The honest version: Demonstrate, do not assert. Show your working. Publish the method. Say what you got wrong last time and what changed as a result. Real authority shows up as specificity — experts can say precisely when their advice does not apply.
The dishonest version: Meaningless credential inflation, “as seen in” badges from a paid listing, borrowed authority through name-dropping clients you worked with for a week.
Test result: Passes when it is demonstrated. The most reliable authority signal in copywriting is the willingness to say “this is not for you, and here is who is”.
6. Scarcity
What it is: Limited availability increases perceived value and urgency. It is the most abused principle on this list by a considerable margin.
The honest version: Scarcity that actually exists. A solo consultant genuinely has a finite number of project slots. A cohort course genuinely starts on a date. Say the true constraint and say why it exists: “Three writing projects a month, because past that the research phase gets rushed.”
The dishonest version: Countdown timers that reset on refresh. “Only 2 left” on an unlimited digital product. Perpetual “closing soon” enrolments.
Test result: Passes only when the constraint is real and you can explain its cause. If you would be embarrassed to explain where the limit comes from, there is no limit.
7. Framing
What it is: The same information presented differently produces different judgements. Anchoring, reference points and unit framing all sit here.
The honest version: Frame against a reference the reader would recognise as fair. Comparing your NZ$690 audit to the cost of a month of unconverted traffic is legitimate if the traffic number is theirs. Presenting a monthly equivalent alongside the total, not instead of it, is legitimate.
The dishonest version: Anchoring against a “value” nobody ever paid. Hiding the annual commitment behind a monthly figure. Comparing your price to a category the reader was never considering.
Test result: Passes when both frames are visible. The rule I use: show the total somewhere on the same screen as the per-unit figure.
8. Processing fluency
What it is: Information that is easy to process is judged as more true, more likeable and less risky. Easy-to-read type, familiar words and clean rhythm all quietly increase believability.
The honest version: This is the one principle that is almost impossible to misuse, because it is just good writing. Short sentences. Familiar words. Clear hierarchy. Adequate contrast. Say the thing.
The dishonest version: Using fluency to make a weak claim feel solid — a beautifully typeset assertion with nothing behind it. Fluency makes claims feel true; it does not make them true, and a reader who checks later will remember that you knew the difference.
Test result: Passes. But pair it with substance, because fluency raises believability for false claims exactly as efficiently as for true ones.
In every case above, the honest version and the dishonest version use the same psychological mechanism. What separates them is not the technique. It is whether the underlying thing is true, and whether the reader would consent to the method if they could see it.
The four I won’t use
Confirmshaming. Opt-out links written as self-insult — “No thanks, I don’t want more customers”. It works slightly, and it costs you the good opinion of everyone who declines. That is a bad trade at any conversion rate.
Fake urgency. Timers that reset, stock counters that lie, deadlines that move. Beyond the ethics, it is a compounding liability: it works once per customer and it teaches everyone who catches it that nothing you say can be taken at face value.
Manufactured inadequacy. Copy that creates a problem in order to sell the fix — particularly around bodies, parenting, money and status. Naming an existing pain is service. Installing a new one is not.
Obscured commitment. Auto-renewals buried in terms, cancellation that requires a phone call, prices that appear only after data capture. If your business model needs people not to notice what they agreed to, the copy is not the part that needs fixing.
What ethical persuasion buys you
The commercial argument, since it is a fair thing to ask about. Manipulative techniques optimise for the immediate conversion and pay for it downstream: refunds, chargebacks, negative reviews, churn, and a slow erosion of the word-of-mouth that most service businesses actually live on.
Honest persuasion optimises for the decision being right. That shows up as customers who stay, who refer, who arrive at the sales call already half-convinced because someone told them you were straight with them. It is slower to build and considerably harder to lose.
If your copy only works on people who do not fully understand what you are doing, you do not have a persuasion strategy. You have an exposure risk with a good conversion rate.
Write the version that survives being explained. It is usually the better-performing version anyway, and it is always the one you can defend in three years when someone quotes it back to you.