Speaking to people
who were not looking for you.

We start by looking at whether your offering suits this channel. A product understood in three seconds and decided quickly works there. A complex, expensive service needing three meetings before signature rarely works there directly, and it is better to know that before committing a budget.

Then we build creatives, in the plural. Not one beautiful advertisement: several proposals telling different things, because nobody, here or elsewhere, can predict which will work. The only reliable method is to produce enough of them to let the market decide.

Finally we set up measurement that does not settle for what the platform claims, and we compare it against what your invoicing says.

For years, this channel’s sales argument rested on the precision of its targeting: reaching exactly the right people according to their declared interests. That time has passed, for two reasons.

The signal has shrunk

The tracking restrictions put in place by operating systems and browsers have taken away part of what the platform observed. Very fine audiences have become both less reliable and less populated.

The algorithm has become better than us

At the same time, automatic distribution has improved markedly. On a broad audience, the system finds the people who react on its own, provided it has material to distribute and correct measurement.

The practical consequence: spending hours slicing audiences today produces less than spending those hours producing one more creative. That is counter-intuitive for anyone who knew the previous era, and it is the adjustment we make most often on accounts we take over.

An advertisement that does not work is not a failed advertisement: it is information. The problem is never having a bad creative, it is having only one, because then you can compare nothing and learn nothing.

The first three seconds

That is where everything is decided. If the opening does not stop people, the rest does not exist, even if the rest is excellent. So we first look at how many people get past that threshold, before even looking at the sales.

Anything that looks like an advertisement is ignored

The feed is a place where people look at those close to them and at their own interests. An image obviously produced by an agency announces itself immediately and gets scrolled past. The formats that work best often resemble what surrounds them, which regularly upsets art-direction expectations.

Wear

A good creative does not stay good. As it is seen several times by the same people, its return falls. That is not a failure, it is how the channel normally works, and it calls for a rhythm of replacement rather than a one-off piece of work.

On this channel, the spend rises fast as soon as a creative works, so the question of who has an interest in it rising comes up straight away.

Three rules apply across the whole division. The media budget does not pass through us and we take no percentage of it: our fee is fixed. The account is created in your name, and the history as well as the audiences stay with you if we stop working together. And there is no minimum term on advertising management. The detail of those three points is on the division’s page.

One side effect deserves anticipating, because it happens regularly and catches teams out.

An advertisement that works attracts comments. Questions about the price, about lead times, about availability, sometimes public criticism. Those comments are visible to everyone who sees the advertisement afterwards, and they weigh more than the message itself: a question left unanswered for three days under a widely distributed advertisement costs sales.

So we build that workload into the framing, because it is not a matter of advertising management but of your organisation. Who answers, how quickly, with what latitude. A company that launches a campaign without having answered that question discovers the problem at the worst moment, which is when the campaign works.

A recurring disagreement deserves stating in advance, because it always comes as a surprise at the first report.

An advertising platform measures its own performance. When somebody sees your advertisement then buys, it claims the sale. When somebody sees your advertisement, forgets it, types your name into a search engine three weeks later and buys, it often claims that too. That is not dishonesty, it is what the measurement window allows it to see.

The result is that adding up what several channels claim regularly produces more sales than actually happened.

Our method is simple and unflattering: we compare what the platform says with what the company actually banks. When the two disagree, the accounts are right. Our reports therefore show lower figures than the interface, and that is deliberate.