Data Should Make You Braver, Not More Boring

We have more data than any generation of marketers in human history.

Beautiful.

So what are we doing with it?

Making everything look the fucking same.

We know the optimal video length.

The benchmark CTR.

The average CPM.

The highest-performing CTA.

The safest audience.

The best posting time.

The recommended number of characters in the headline.

We know people like faces.

We know short-form video performs.

We know the product should appear early.

We know vertical is good.

Fantastic.

We have successfully reverse-engineered:

An ad.

And now everyone is making it.

Data Was Supposed to Make Marketing Smarter

And it has.

Obviously.

We can see what people actually do.

Not just what they say they’ll do.

We can understand which creative holds attention.

Which hooks work.

Which creators perform.

Which messages drive action.

Which audiences convert.

Which formats disappear.

Which ads are quietly setting money on fire.

This is incredible.

The problem isn’t data.

The problem is what we do next.

Because there are two ways to use evidence.

One is:

“This gives us confidence to go further.”

The other is:

“This tells us exactly what we should repeat forever.”

Guess which one is easier to put in a deck.

“The Data Says…”

This might be one of the most powerful phrases in modern marketing.

“The data says shorter videos perform better.”

Okay.

“The data says customers respond to creators.”

Interesting.

“The data says humour performs.”

Great.

“The data says…”

Hang on.

Does the data actually say that?

Or did something happen and we've created a nice little story about why?

Because data is very good at telling you what happened.

Understanding why is where things get messier.

Maybe the shorter video performed because it had a better hook.

Maybe the creator performed because they were funny.

Maybe the funny post performed because it was about a topic people cared about.

Maybe the polished brand film tanked because the first three seconds were a drone shot of a fucking building.

Context matters.

If you strip the context away, insights become rules.

And rules become boring incredibly quickly.

This Is How We End Up With “Best Practice”

Something performs.

Excellent.

We analyse it.

We identify the pattern.

We turn the pattern into a recommendation.

We repeat it.

Competitors notice.

They repeat it.

Agencies put it into presentations.

Platforms publish it in best-practice guides.

Everyone repeats it.

Six months later, your entire category is doing the same thing.

“Three reasons why…”

“POV:”

“Things I wish I knew before…”

Person talking to camera.

Captions.

Quick cut.

Product.

CTA.

Technically optimised.

Completely indistinguishable.

Best practice has a funny habit of eventually becoming average practice.

Because if everyone knows the trick, it stops being much of a trick.

Your Dashboard Cannot Tell You What Nobody Has Tried Yet

This is the limitation.

Data is historical.

It tells you about things that happened.

Things you made.

Things customers saw.

Things they clicked.

Things they ignored.

Useful.

But your next brilliant idea may not exist in the dataset.

There is no benchmark for the thing nobody has done.

No historical CTR.

No average engagement rate.

No case study.

No comforting little green arrow.

Which means if you only make things your existing data can validate, you create a strange loop.

We make what worked before.

Then collect data showing that what we made worked before.

Then use that data to make what worked before again.

Eventually:

“Our data shows customers respond best to the type of content we exclusively make.”

Incredible.

Science.

Data Should Give You Permission

Here’s the more interesting version.

You look at 11 months of social content.

And you notice something.

Every time you open with a particular type of hook, retention improves.

Good.

What do we do?

Option one:

Make that exact video 47 more times.

Please don't.

Option two:

Ask what that hook tells us about the audience.

Maybe they respond to confrontation.

Curiosity.

Specificity.

A face.

A question.

A confession.

An unexpected visual.

Great.

Now use that understanding to make something bigger.

Could the learning inform the campaign?

Could it inform the TVC?

Could we test ten more versions?

Could we give the creator more freedom?

Could we make the idea stranger?

Could we put more money behind it?

That's exciting.

The data didn't give you the answer.

It gave you confidence.

Imagine Starting Your TVC With Your Best-Performing Social Hook

This is where data gets useful beyond the dashboard.

We’ve all had the brief:

“Hey, we made this TVC. Can you make it work on social?”

Of course.

Did anyone think about social when you made it?

“No.”

Great.

We'll just chop your beautiful 30-second film into vertical pieces and pray.

But what if the information flowed the other way?

What if before writing the TVC, you looked at the last year of social?

What openings actually held attention?

What talent performed?

What messages people responded to?

What jokes worked?

Which creators your customers already liked?

What if your TVC started with the strongest-performing hook from the last 11 months?

What if the creator who consistently earns attention on your social channels starred in the thing?

Now social isn't the place you dump the campaign afterwards.

It's the R&D department for the bigger creative.

That's a much more interesting use of data.

Your Audience Is Running Experiments for You Every Day

Every post gives you information.

People watch.

Or leave.

Click.

Or don't.

Share.

Save.

Comment.

Buy.

Ignore.

That doesn't mean every metric is equally useful.

And it definitely doesn't mean:

MOST LIKES WINS.

But over enough work, patterns emerge.

Your audience is constantly giving you behavioural feedback.

Not:

“Yes, I would theoretically purchase this product if it were available at an acceptable price.”

Actual behaviour.

They watched.

They clicked.

They bought.

Or they fucked off.

That is useful information.

And it should make creative teams more dangerous.

Not less.

The Average Is a Terrible Place to Build a Brand

Benchmarks are useful.

You should know them.

You should understand what normal looks like.

Then maybe don't spend your entire career trying to become normal.

If the average completion rate is X, useful.

If the average CPM is Y, useful.

If the average engagement rate is Z, useful.

But the goal of a distinctive brand cannot simply be:

Perform approximately like everyone else, slightly more efficiently.

The category average is literally where everyone else is.

And yet we treat benchmarks like destinations.

“We're above benchmark.”

Excellent.

Are we memorable?

“Above benchmark.”

Does anyone care?

“4.7% above benchmark.”

Wonderful.

Put it on the gravestone.

Data Can Become a Shield Against Taste

This is where things get uncomfortable.

Eventually someone has to decide whether an idea is good.

Terrifying.

Because if you make a judgement, you can be wrong.

Much safer to say:

“The data suggests…”

Now the spreadsheet approved it.

Not me.

But marketing still requires taste.

Data can inform taste.

Challenge taste.

Correct taste.

Expose bad taste.

But it cannot completely replace it.

Two ideas can follow exactly the same best practices.

Same duration.

Same hook structure.

Same creator format.

Same CTA.

One is brilliant.

One is fucking awful.

Where's that cell in Excel?

More Data Does Not Automatically Mean More Certainty

This is especially relevant now.

We have behavioural data.

Social listening.

Search data.

Brand tracking.

Creative testing.

Synthetic audiences.

AI models capable of processing enormous amounts of information and generating predictions faster than a traditional research process could dream of.

Great.

Use it.

But don't confuse the ability to produce an answer quickly with the ability to remove uncertainty.

Marketing is still dealing with humans.

Annoying little creatures.

Context changes.

Culture changes.

Competitors react.

Platforms move.

Something weird happens on a Tuesday and suddenly everyone is talking about a cucumber.

The goal isn't perfect certainty.

It's better bets.

AI Should Make Experimentation Fucking Wild

This is the bit that excites us.

The cost of generating possibilities is collapsing.

We can create more hypotheses.

More hooks.

More scripts.

More variations.

More audience scenarios.

We can use synthetic data to pressure-test ideas quickly.

We can analyse mountains of social behaviour.

We can identify patterns humans might miss.

We can explore 100 directions before lunch.

So what are we going to do with this incredible new capability?

Hopefully not:

Generate the safest possible ad 100 times faster.

What a depressing use of technology.

If AI makes testing cheaper, the logical outcome should be more experimentation.

More strange ideas.

More informed risks.

More things entering the market.

Faster learning.

Bigger creative swings backed by better evidence.

Not an industrial-scale beige factory.

Stop Using Data to Win Internal Arguments

Another little habit.

We decide what we want.

Then go looking for numbers that prove it.

Creative likes Concept A.

Find the metric supporting Concept A.

Management likes Concept B.

Find the research supporting Concept B.

Media wants more spend.

Here's the reach curve.

Everyone arrives armed with charts.

This isn't data-led marketing.

It's corporate litigation.

Good data should occasionally tell you something you don't want to hear.

The creator management hated is outperforming.

The beautifully polished content nobody questioned is underperforming.

The weird joke worked.

The product-first opening didn't.

Great.

Now the difficult bit:

Do you believe the evidence when it disagrees with the hierarchy?

Sometimes the Data Should Make the Meeting More Uncomfortable

Imagine saying:

“Our last 11 months of data tells us the thing management is asking us to change is actually one of the reasons this works.”

Interesting meeting.

Or:

“The creator you don't personally like is consistently our highest-performing talent with the audience we're trying to reach.”

Awkward.

Or:

“The less branded executions are holding attention longer and producing better outcomes.”

Oh dear.

That's where data becomes useful.

Not because it eliminates judgement.

Because it gives good marketers something to stand on when they make an uncomfortable decision.

Evidence can give courage a backbone.

Don't Optimise the Life Out of It

Optimisation is good.

Obviously.

Find what works.

Improve it.

Remove waste.

Make the media work harder.

Make the creative work harder.

But every optimisation has a direction.

You need to know what you're optimising for.

Clicks?

Views?

Sales?

Memory?

Attention?

Lowest possible CPM?

Internal comfort?

Because if you optimise one metric hard enough, you can accidentally destroy everything else.

A thumbnail becomes clickbait.

A hook becomes annoying.

A brand becomes indistinguishable.

A campaign becomes 47 tiny performance tricks wearing a trench coat.

The numbers might improve.

The marketing might not.

Use Data Like a Trampoline, Not a Cage

That's probably the whole argument.

Data should tell you where the ground is.

Then help you jump higher.

It should tell you:

People respond to this.

Push it.

This creator works.

Back them.

This hook holds attention.

Take the learning somewhere bigger.

This format isn't working.

Try something else.

This thing surprised us.

Find out why.

Not:

This worked.

Repeat until death.

Because the competitive advantage isn't having data anymore.

Everyone has data.

Everyone has dashboards.

Everyone has AI.

Everyone has access to roughly the same best-practice documents telling them to put captions on vertical videos.

The advantage is what you do with the information.

Use it to retreat towards the average?

Or use it to make a smarter leap away from it?

Because if all your beautiful data, research, dashboards, testing tools and AI models ultimately produce marketing that looks exactly like everyone else's…

We may have slightly missed the fucking point.

Data should make you smarter.

And being smarter should make you braver.