How many ad creatives you actually need to test
Why the right number of creatives is set by your budget rather than by best practice. How to split concepts from variants before you commission anything.
The number everybody quotes is the wrong question
Somebody will tell you to launch with ten creatives. Somebody else will say three. Neither answer can be right in general because the useful number is set by how much you are spending and how often people convert.
A test works by giving each creative enough traffic for its result to be distinguishable from chance. Split a small budget ten ways and every creative gets a trickle. You end up with ten results that all look similar and a decision made on noise. The practical question is therefore how many creatives your budget can actually resolve.
What creative costs and what it does not include
Creative production excludes media spend the offer itself and any landing page work. A new image cannot rescue a proposition the market has already declined. It also excludes the losers. Testing means deliberately spending money on creatives that will fail and that spend is the price of the information rather than waste. Budget for it openly or you will kill tests early and learn nothing.
Production itself can be commissioned as a fixed scope. A set of performance creatives with a published scope and rate removes the open-ended quoting that makes businesses under-order. Service fees carry GST at 18%.
Who should test properly and who should barely test at all
Structured creative testing earns its keep when you have enough weekly conversions for differences to show up inside a month and when the audience is broad enough that creative is genuinely the variable. Consumer products local services and anything with daily volume sit here.
It does not suit a small budget. A business spending twenty thousand rupees a month should run two or three concepts and nothing more. Ten creatives on that spend is not a test and the fees spent producing them would have bought more media. It also does not suit low-conversion businesses. A firm closing a few dozen deals a year will never accumulate enough results per creative for a comparison to mean anything and should spend the money on better sales follow-up instead. And if you already have a creative that works and your audience is small then leave it running. Testing for the sake of activity has a real cost and no guaranteed upside.
Concepts first then variants
Separate the two before you brief anyone. A concept is a different argument for buying. Price against convenience against risk reduction against speed. A variant is the same argument expressed differently. A new photograph a different headline a changed colour.
Concepts are where the large differences live and they are what a modest budget should spend its testing on. Three concepts in two formats each gives six assets and a real chance of learning something. Variants only become worth testing once a concept has clearly won because you are then refining a known quantity rather than searching.
Run each set until every creative has accumulated a meaningful number of results rather than until a fixed date. Then keep the winner as a control and replace only the worst performer. Replacing the whole set each cycle destroys your reference point and you lose the ability to say whether anything improved.
The metric that matters and the one that flatters
Click-through rate flatters and so do video view metrics. A creative can be excellent at earning a cheap click and poor at earning an enquiry and the two are routinely inversely related. Curiosity is cheap to buy.
The metric that matters is cost per qualified result for each creative. Watch how long a winner holds its position alongside it. A creative that wins for six weeks is worth far more than one that wins for six days and the first is invisible if you only look at a single report.
The common mistake
The expensive error is calling a winner on a handful of results. Two creatives sitting at three conversions and one conversion look like a clear outcome and are statistically indistinguishable. A business then scales the apparent winner cuts the other and wonders why performance does not improve. Repeat that cycle monthly and you have paid for a year of testing with no accumulated knowledge.
The quieter version of the mistake is testing variants while believing you are testing concepts. Six images of the same product with six slightly different headlines is one concept in six costumes. Whatever wins tells you nothing about why anybody buys.
Creative decay is a running cost
No creative keeps working indefinitely on social platforms. Performance falls as the audience sees it repeatedly and the decline is a property of the medium rather than a fault in the work. This is why creative belongs in your monthly budget rather than in a launch budget. A business that commissions one excellent batch and nothing afterwards will see results fade over following months and will usually blame the platform or the agency. The correct reading is that the production line stopped.
This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.