The false light
Cornwall has plenty of wrecking stories.
Some of the wrecking and plunder were real. The more theatrical detail — false lights on the cliffs, sometimes supposedly tied to horses, luring ships onto the rocks — is much harder to substantiate.
Which is slightly disappointing, because it is an excellent story.
It is also an excellent metaphor for bad web design.
The internet is full of false lights: interfaces that appear to offer one thing while quietly steering you towards another.
The terminology has become more precise. The Competition and Markets Authority describes online choice architecture. Harmful versions are often called dark patterns or sludge.
Whatever label you prefer, the useful question is the same:
Is the interface helping somebody make a clear decision, or making the decision less clear because that produces a better number on a dashboard?
You have definitely seen one
A button says “Continue” when the meaningful consequence is “Start paid subscription”.
An optional extra is already selected for you.
A cancellation route exists, technically, but finding it begins to feel like a small administrative quest.
A countdown reaches zero and, in an extraordinary act of commercial resurrection, starts again.
The “accept everything” option is enormous. The alternative appears to have been typeset in fog.
The mechanics vary. So does the intent behind them.
The effect can still be similar: steering somebody towards a choice they might not otherwise make, or making another perfectly legitimate choice needlessly difficult.
I understand why businesses are tempted.
If one version produces more sign-ups than another, the dashboard says it is working.
The dashboard may even be right.
That is not the same as saying it is good.
A better number is not always a better result
This is the problem with optimising a single metric.
Imagine two checkout flows.
The first produces slightly fewer completed orders, but customers understand the price, know what they are buying and can leave easily if they change their mind.
The second produces more immediate conversions because an extra is pre-selected, a warning is exaggerated or backing out is made deliberately awkward.
Which one is better?
If you only care about the conversion rate today, perhaps the second one.
But the cost does not disappear simply because it has moved off the conversion dashboard.
You may end up dealing with more complaints, refunds, support requests, chargebacks, poor reviews or customers who simply do not trust you next time.
Not every deceptive interface causes every one of those outcomes. The point is simpler: short-term optimisation can push the cost somewhere else.
Research into dark patterns supports the broader concern. Common designs can materially change user decisions, and consumer research has also linked harmful choice architecture with reduced trust.
That should at least make us cautious about celebrating the first metric that goes up.
Would you say it out loud?
One of my favourite tests for interface copy is very simple.
Would you be comfortable explaining it to the customer face to face?
If the button says one thing but does another, would you describe that openly?
If the cancellation process is intentionally awkward, would you say:
We made this harder because we would prefer you kept paying us.
Probably not.
That is quite a useful warning.
Good UX should survive being translated into normal English.
Make leaving as clear as joining
There is a strange idea in growth culture that every person who leaves is a failure.
They are not.
Sometimes somebody does not want the product.
Sometimes the timing is wrong.
Sometimes they need to cancel.
Let them.
If somebody has to fight their way out of your service, you may keep the revenue for another month. You have also shown them how the business behaves when its interests and theirs stop aligning.
I would rather somebody leave thinking:
That was easy. I may come back.
Clear cancellation does not guarantee loyalty. It does remove unnecessary friction.
And in the UK, this is not merely a matter of taste. The CMA’s guidance on optional paid extras says customers must actively consent to those additional charges; pre-ticked boxes are not enough.
You do not need to turn the website into a legal essay.
You do need to avoid designing against the user.
Clarity is not anti-commercial
Ethical design is sometimes presented as the soft, worthy alternative to serious commercial optimisation.
I do not buy that.
Clear pricing helps people understand the offer.
Clear buttons help them understand what happens next.
Clear cancellation gives them a genuine exit.
Clear expectations reduce the number of surprises you have to explain later.
Good accessibility means more people can use the thing in the first place.
None of this is charitable.
It is competent.
The cleverest interface is often the one the user barely notices because it behaves exactly as expected.
Be careful what you optimise for
There is a wider lesson here than dark patterns.
Any system becomes strange if you optimise hard enough for the wrong thing.
Optimise only for sign-ups and you may create reluctant customers.
Optimise only for time-on-site and you may make information harder to find.
Optimise only for short-term retention and you may make cancellation unpleasant.
The metric improves.
The experience gets worse.
Pleasing spreadsheet. Displeasing business.
The light has to mean what people think it means
That is why the false-light metaphor still works, even if the most famous Cornish version belongs more securely to folklore than documented history.
Navigation depends on trust.
You see the light and act because you believe it represents reality.
A good website should work the same way.
The price should be the price.
The button should do what it says.
The choice should be a real choice.
The exit should exist.
If somebody genuinely wants what you sell, you do not need to trick them into it.
Help them make the decision clearly.
And if the answer is no, let no mean no.
That is not anti-growth.
It is decent design.
