Why “we’ve tried advertising” almost never means that advertising doesn’t work

It is the most frequently heard phrase in introductory meetings: we've tried advertising, it doesn't work for us. I understand that conclusion too, because their own experience tells them so.

However, in almost all of those cases I see the same thing: a simplified version has been implemented, the bare minimum has been done in the hope of maximum results, and the plug was pulled before anything could be measured. That is not a test of the channel, that is a test of that setup.

That also makes it a difficult conversation. As a party that sells advertising, it is hard to claim that it does work with a better setup, because that is precisely what a salesperson would say. That is why I would rather not make a promise, but instead give four control points that you can check yourself.

1. Was the conversion measurement correct?

Start here, because this is most often the cause. What we find with almost every new client: conversions are indeed taking place, but they are not being measured. Either they are attributed to an undefined channel, or everything ends up as direct traffic.

If that was the case, then you don't know whether the campaign yielded nothing. You only know that you didn't see it. And worse, the algorithm didn't receive any good feedback signals either, so it strongly optimised in the wrong direction.

Here is how you can check this afterwards: compare the number of enquiries you actually received in that period with what the advertising platform reported. If there is a large gap, it was down to measurement.

2. Was there enough budget to learn something?

Algorithms need conversions to recognise patterns. If the number of conversions per week remains too low, a campaign stays in the learning phase crashing and thus expensive and unpredictable.

We therefore work with a lower limit of approximately 1,250 euros per month. Not because nothing is possible below that, but because below that amount you are mainly paying for the learning phase without reaping the rewards of it.

Here is how you can check this afterwards: how many conversions per week did the campaign get? If it was just a handful or fewer, the system never had enough to learn anything.

3. Did the message differ per phase?

Someone who doesn't yet recognise their problem needs a different message than someone who is comparing three providers. The same advertisement for both works for neither.

In practice, I usually see a single message for everyone, and that is aimed at the small group that is already ready to buy. That is the most expensive group, because all competitors are bidding on them too.

Here is how you can check this afterwards: pull up the adverts. Were there any variants that explained something or made a problem recognisable, or did everything directly ask for a quote?

4. Has it spun for long enough?

Most purchases are not impulse buys. Anyone who advertises according to the customer journey must give people the time to complete that journey. Depending on the channel and your decision-making cycle, it takes two to six months before you can make a reliable judgment.

What I often see: trying for six to eight weeks, no immediate return, stopping. That is too soon to draw a conclusion, and at the same time long enough to feel like you gave it a fair go. That combination is precisely why the conviction afterwards is so persistent.

When it really just doesn't work

To be honest, there are situations where advertising won't work out in your case, even with a good setup.

• The calculation is wrong. If the customer lifetime value is too low relative to the cost per click in your market, making the numbers work is no mean feat. On LinkedIn, for example, a click typically costs between 10 and 20 euros; your offer has to be able to support that.
• The target audience cannot be targeted. If there is no metric you can steer by and nobody is actively looking for what you are doing, there is little to go on.
• The offer is not selling anywhere. Advertising magnifies what is already there. If there is no evidence of demand yet, it is too early.

The difference from the previous four points is that you can calculate this in advance. And if the outcome is that it doesn't add up, then you know why, and that is something different from a suspicion based on disappointment.