How to Improve Advertising Creative: The Mathematics of Creative Probability

Dear clients and executors (traffic managers, performance marketers, SMM folks), let’s try to stop this medieval bloodbath around “the creative is trash, the author is trash”.

We’ll talk, specifically, about paid traffic.

It’s common that the traffic manager makes the creative: the text, the image, or both. That is, the “creative”, exactly as I said.

And then the client starts chewing on it — pushing the person either toward a nervous breakdown, white-hot rage, or a depressive-anxious episode at the very least.

— Your creative is… not very good. Honestly, it’s crap.

This is an approximate, but painfully close retelling of what one of my clients once said. And here’s the nuance: this kind of nonsense is usually produced by people who are extremely inexperienced with traffic management — those who’ve ordered these services very rarely, or never at all. Experienced clients behave much softer, even if sometimes they still look at you with that suspicious squint.

I will prove mathematically why “creative criticism” and endless pre-launch edits are wildly inefficient — and then I’ll give a practical recipe that works for both sides.

About the author:

Bogdan, creator of The Quite Orbit — a marketing and growth agency.

— Personally in internet marketing since August 2020;

— Before that: 10 years in IT (tech support, sysadmin, 1C analytics); (For readers outside Russia: 1C is a very common Russian ERP/accounting software ecosystem, similar in “role” to local business systems like QuickBooks + ERP pieces, depending on the company.)

— 6 years in organising concerts, music events, and theatre shows. Unfinished degree in philosophy.

So far I’ve worked with 94 projects:

• 75 in B2C
• 11 in B2B

The root of the problem

There’s always a clear number of clicks a creative brought — and whether there were leads, subscribers, or any other “quality” reaction after the ad. Not just quantity, but people doing something meaningful.

But all those nuances collapse into a binary state:

The creative worked = there is something to analyse afterwards.

You have impact: clicks close to the forecast, plus subscribers, plus leads.

The creative did NOT work = none of that exists, or it exists in a volume/quality too small to analyse.

So it gets packaged as worked / didn’t work, which looks like a 50% chance of a creative “taking off”.

Now watch what happens when a client insists on replacing the creative BEFORE it even hits traffic.

Replacing the creative because the client criticised it

• The probability that the new creative will work stays at 0.5 (50%), because the baseline for one creative is still: it either works or it doesn’t.
• The probability that the new creative will not work also stays at 0.5 (50%).

Result: the chance of success or failure doesn’t change with replacement, because you simply swap one single hypothesis for another single hypothesis.

“But what if the client is right?” — fair. They also care about the outcome. Should we ignore them? Or should we only listen to them — since, mathematically, “it’s the same”?

Let’s do something smarter: not replace, but add the client’s version alongside.

Adding the client’s creative

Now we have two creatives (the executor’s and the client’s), and we’re looking at the probability that at least one of them works.

Possible outcomes:

1. The original creative works, and the client’s creative doesn’t.
   Probability: 0.5×0.5=0.25 (25%).
2. The client’s creative works, and the original doesn’t.
   Probability: 0.5×0.5=0.25 (25%).
3. Both creatives work.
   Probability: 0.5×0.5=0.25 (25%).
4. Both creatives don’t work.
   Probability: 0.5×0.5=0.25 (25%).

Sum of probabilities:

Probability that at least one creative works:
0.25+0.25+0.25=0.75 (75%).

Conclusion:

• If you replace one creative with another, the success probability stays the same — 50%.
• If you add a creative, the probability of success increases to 75%, because you’re increasing the number of hypotheses — and your odds that at least one works. (And yes — this is a subtle hint toward a third creative.)

Adding the client’s idea (instead of fully replacing) is more profitable in terms of probability of success.

It turns out that in real mathematics — in probability theory — there’s a classic type of problem for this: “union of probabilities”.

“Union of probabilities” interpreted for advertising

Essence:
The “union of probabilities” problem answers: what is the probability that at least one of several independent events will happen?

This type of task appears everywhere: from estimating the odds of completing tasks successfully to assessing risk.

Union probability formula:

To find the probability of at least one success, you first find the probability that none of them works — then subtract it from 1 (from 100%).

Formula:

P(at least one event)=1−P(no event happened)

Where the probability that no event happened is the product of probabilities that each event fails:

P(no event)=(1−P(A1))⋅(1−P(A2))⋅…⋅(1−P(An))

Example calculation:
Let’s say we launch two creatives:

• The first creative has a success probability P(A1)=0.5 (50%).
• The second creative has the same probability P(A2)=0.5 (50%).
• Step 1: Probability that both fail:
  P(both fail)=(1−0.5)⋅(1−0.5)=0.5⋅0.5=0.25
• Step 2: Probability that at least one works:
  P(at least one works)=1−P(both fail)=1−0.25=0.75

So the probability that at least one creative succeeds is 75%.

Conclusion of the task

1. Replacing one creative with another:
   If you simply replace it, the probability of success stays the same as for one creative — 50%.
2. Adding a second creative:
   If you add a second variant, the overall success probability grows. In our example, it rises to 75%.
3. Did you catch the hint about adding a third creative?

Global conclusions:

Now the recipe to end the argument. Let’s say it out loud so all hints are crystal clear:

1. If you’re the client and you don’t like the executor’s creative, and you want to add your corrections — don’t demand a replacement. Propose making your version based on your corrections alongside the executor’s original.
2. If you’re the executor and the client wants to edit your creative — produce a version based on those edits that does not replace yours, but goes next to your original. For persuasion, use the simple probability breakdown from the “Possible outcomes” section in this article.