How do you test ad copy before paying for a single click?

The expensive way to test ad copy is to write one line, put $200 behind it, and find out from thirty-six clicks whether strangers care. The cheap way is to find out most of that before any money moves, and the reason almost nobody does it is that the cheap methods feel unserious. They are not rigorous. They are, however, free, and they eliminate the lines that were never going to work — which is most of them. (Disclosure: published by BidSurvivor, which sells advertising and therefore benefits when you skip this step. The advice below is to not skip it.)

What a pre-test can and cannot tell you

Be clear about the limit before you start, because this is where people fool themselves.

A pre-test tells you whether a line is comprehensible and whether it is interesting. It cannot tell you whether it converts. Comprehension and interest are necessary and nowhere near sufficient: plenty of clear, intriguing lines send people to a page they bounce off in two seconds.

What that gets you is a filter, not a forecast. If you have eight candidate lines and pre-testing kills five of them, you have turned an unaffordable eight-way test into an affordable three-way one. That is the entire value, and it is a lot of value when thirty-six clicks is all $200 buys.

Four methods, cheapest first

The five-second test. Show one person the line for five seconds, take it away, and ask what the product does and who it is for. If they cannot answer, no amount of budget fixes it — you have written a line that requires a second reading, and an ad never gets one. Do this with five people and you will not need a sixth; the failures are unanimous and obvious.

This works because it tests the thing ads actually face. A search result, a card on a board, a line in a newsletter: all of them get a glance, not a read — the Nielsen Norman Group's eye-tracking work on how people read online found that they mostly do not, they scan. The one-line pitch patterns that survive this test are the ones that name a specific person and a specific problem in the same breath.

Reading it to somebody who owes you nothing. Not your co-founder, not a friend who wants you to succeed. The failure mode of friendly feedback is that it is about you rather than about the line. A stranger in a relevant community, a person on a support forum, someone in a Discord for the trade you are selling to — anyone whose reaction is not filtered through wanting you to feel good.

The tell to listen for is not "I like it". It is "wait, is this for—?" followed by a guess. If their guess is wrong, the line is wrong, and you have learned that for the price of a message.

Structured research with real participants. The step up from asking five people badly is asking twenty people well: a written study, a consistent set of questions, participants who match your buyer rather than whoever was around. This is what user research platforms exist to do, and the honest note is that it costs either money or a serious amount of your time — which puts it above the free tier but well below the cost of learning the same thing from paid traffic.

The version worth doing at a small scale is narrow: one question ("what does this promise you?"), one artefact (the line, on its own), twenty people who plausibly buy what you sell. Broad studies at this stage produce a document nobody acts on.

A free ad slot, which is where the pre-test stops being a pre-test. Ours: BidSurvivor runs two twelve-hour slots a day, and the first brand into an empty one pays nothing — no account, no card. Every visitor also starts with $100 of house credit, so contested slots cost nothing out of pocket at current prices either. What you get back is the number of people who opened your card and the number who clicked through to your site, published on the front page and on your brand's own page.

That is not a survey. It is cold strangers meeting your line with no context and no goodwill, which is the strictest reading it will get. The honest limitation is scale: the board is young, the numbers are small, and a handful of cold clicks cannot tell you whether a channel works. It can tell you whether a sentence does, which is the question this article is about. The full rules, including the no-refund rule when somebody outbids you, are in how BidSurvivor works.

What to test, specifically

Testing "the copy" is too vague to produce an answer. Test one of these at a time:

• The subject. Does the line name the thing you sell, or a category the reader has to translate? "Invoicing for electricians" beats "streamline your workflow" every time, and the second one is what most founders write first.
• The audience. Does a reader recognise themselves in it within the five seconds? Naming the buyer is usually worth more than naming the benefit.
• The specific number. Lines with a concrete figure survive glances better than lines with an adjective. "Two slots a day" beats "limited availability".
• The verb. What are you asking them to do, and is it small enough that a stranger would do it? "See today's price" is a smaller ask than "start your free trial", and it converts a colder audience.

Change one at a time or you will not know which change did anything — the same discipline that makes a small A/B test impossible to run properly makes an unstructured pre-test useless.

When to stop testing and start paying

There is a real failure mode on this side too, and it is worth naming because this article could otherwise be read as an argument for never buying anything.

Pre-testing has a floor. Once a line is clear, once strangers guess your audience correctly, once it survives a glance — further asking produces opinions, not information. Opinions about ads are famously unreliable, including the opinions of the people who would buy the product. The gap between "I would click that" and clicking is enormous and it is not closed by asking more people.

At that point the only remaining question is whether the line converts, and the only instrument that answers it is traffic. So: pre-test until you have two or three lines you cannot separate by asking, then spend the budget on separating them. Do not spend it on the eight-way test you started with.

The order that costs the least

1. Write eight lines. Deliberately more than you need. The eighth one is often the good one, because by then you have run out of the obvious framings.
2. Five-second test them with five people. Expect to lose half.
3. Read the survivors to strangers who owe you nothing. Listen for the wrong guess, not for praise.
4. Put the best one in a free slot and watch cold clicks land or not.
5. Then spend money, on two or three lines at most, on one channel, with enough budget for the result to mean something.

None of steps one through four costs anything but an evening. Step five costs whatever your conversion rate makes it cost — which is why it is worth arriving there with three lines instead of eight.