Spending a Hundred Dollars to Learn Something

What a small budget cannot do

It cannot tell you which of two adverts is better, which headline wins, or what your true cost per enquiry is.

Those need enough events to distinguish a real difference from chance, and a hundred dollars produces a handful of clicks and perhaps one or two enquiries.

One enquiry against zero is not evidence, and businesses regularly draw firm conclusions from exactly that.

Accepting the limit is what makes a small test useful, because it directs the money at a question the money can actually answer.

What it can do

Answer a yes-or-no question about whether anything happens at all.

Does anybody click this. Does anybody who clicks do anything. Is the cost per click roughly what I expected or ten times that. Does this offer produce any response.

Those are coarse questions with large differences between the answers, which is exactly what a small sample can detect.

If a hundred dollars produces four clicks and nothing else, you have learned something real. If it produces sixty clicks and two enquiries, you have learned something different and equally real.

Neither tells you what to optimise, and both tell you whether to continue.

How to structure it

• One message, not four.
• One audience, defined simply.
• One page, built for it.
• One action you can count.
• A fixed period, decided in advance.
• A written prediction of what you expect.

The last is the one that turns a spend into a test. Writing down what you expect before starting is what makes the result informative rather than something you interpret afterwards to suit yourself.

Concentrate rather than spread

The most common way small tests are wasted.

A hundred dollars split across three adverts, two audiences, and two platforms produces six fragments, none of which had enough spend to do anything.

The whole amount on one combination gives that combination a genuine chance and produces a result you can read.

The same applies to time: spending it over four weeks at a few dollars a day frequently means the system never leaves its learning phase and never delivers properly.

Concentrate it into a shorter period, typically a week, so delivery stabilises and the result reflects performance rather than start-up.

A worked example

A business considering a new service wanted to know whether there was demand before building anything.

They wrote a page describing the service and its price, and ran a hundred and twenty dollars over eight days to one audience, defined by location and nothing else.

Their written prediction was that they would get roughly forty clicks and would consider two enquiries encouraging.

They got fifty-one clicks and no enquiries.

That was a clear answer at a cost lower than a single day of anybody's time, and considerably lower than building the service and finding out afterwards.

They tested a different framing of the same service two months later, with a different problem in the first line, and got seven enquiries from a similar spend.

The demand had been there. The first description of it had been wrong.

Decide the threshold before you start

The discipline that prevents a test from becoming a rationalisation.

Write down what result would make you continue, what would make you stop, and what would make you test again differently.

Do it before the money is spent, because afterwards every outcome can be argued into encouragement.

One enquiry from a small spend is genuinely ambiguous, and the honest response is usually to test once more rather than to declare either success or failure.

Two consecutive tests producing nothing is a clearer signal than either one alone.

Zero is a useful result

Worth stating because businesses treat it as a failed test rather than as information.

No clicks means the message did not interest anybody in that audience, which is cheap to learn and points at the message or the audience.

Clicks with no action means people were interested until they saw the page or the price, which points somewhere specific and is arguably the most useful result available.

The only genuinely wasted test is one you cannot interpret, which is what happens when several things were changed at once or nothing was recorded.

Design it so every possible outcome tells you something, and then no result is a waste.

What to do with a positive result

Resist scaling immediately, which is the usual mistake.

A small test suggesting something works has established a possibility rather than a rate, and the cost per enquiry from a hundred dollars is not a number to plan with.

Increase gradually, watching whether performance holds, since results frequently degrade as spend rises and the audience widens.

Fix the obvious weaknesses first: the page, the response time to enquiries, and whether you can actually service the work.

A business that scales a working test before it can handle the volume converts a good result into an operational problem.

The counter-case

Small tests can mislead in ways worth knowing.

Timing matters more at small scale, so a week that happens to be a holiday, a heatwave, or a news event can produce a result that has nothing to do with your offer.

Some categories are also simply expensive, and a hundred dollars may buy so few clicks in a competitive market that no outcome is interpretable.

And for considered purchases with long decision periods, a week is too short for enquiries to appear at all, so absence proves nothing.

Check the likely cost per click for your category before deciding whether the budget is capable of answering anything.

Running one

1. Write the question as a yes or no.
2. Predict the result in writing first.
3. Set the thresholds for continuing or stopping.
4. Put the whole budget on one combination.
5. Run it over about a week, not a month.
6. Count the outcome on your own side.
7. Test again before concluding from an ambiguous result.

Step two is what separates a test from a spend, and it takes one sentence.

Why small budgets cannot answer everything is covered in a budget too small to learn anything.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.