Every ad platform wants credit for the same sale.

I do trust platform ROAS when the tracking is set up properly. I just don’t trust the idea that one platform owns the whole customer journey.

Here’s a normal ecommerce journey.

Someone sees a Meta ad.

Three days later, they search the brand and click a Google ad.

They add to basket, leave, then come back through a Klaviyo email and buy.

Meta can argue it created the demand.

Google can argue it delivered the click.

Klaviyo can argue it recovered the sale.

All three can be technically correct.

Who actually deserves the sale?

That’s where attribution gets messy.

Did Meta trigger the thought?

Did Google genuinely influence the purchase, or did it just bid on somebody who would have clicked the organic result anyway?

Did Klaviyo recover an abandoned basket, or did it email somebody who was always going to wait a few hours before ordering?

You can keep adding models, windows and rules, but there is no perfect answer.

The customer doesn’t care which platform gets the credit.

The business should care whether the combined journey was profitable.

ROAS is still useful

I don’t subscribe to the idea that platform-reported ROAS is useless.

If the tracking is configured properly, it is useful.

The platform needs that data to optimise. I need it to understand what the campaign thinks it is doing.

I just wouldn’t use it as the only proof that the advertising worked.

For most commercial decisions, I still like three levels of proof:

The platform. What does Meta, Google, ChatGPT or whichever channel say happened?

Independent analytics. Does another measurement system tell a broadly similar story?

The bank account. Did the business actually make more money?

The expensive metrics take longer

Some of the most important answers don’t arrive on the day of the sale.

Customer lifetime value takes time.

You need to understand the reorder cycle. Churn. Repeat purchase rate. Margin on later orders.

And acquisition cost is only part of the story.

There is also a cost to retaining a customer.

Discount codes. Email. SMS. Loyalty offers. Retargeting. Anything you spend to generate the next order belongs somewhere in the economics of that customer.

That means a channel can look expensive on the first order and turn out to be excellent.

Or look brilliant on day one and bring in customers you have to keep paying to reactivate.

Attribution will never be perfectly neat

Maybe somebody eventually builds perfect cross-platform attribution.

I’m not waiting for it.

Marketing channels interact with each other too much.

A Meta impression can influence a Google click. A Google search can make a Klaviyo email more effective. Organic, paid, email and direct traffic can all touch the same purchase.

Trying to force one channel to own the sale can create false precision.

I’d rather accept some uncertainty and look for a consistent commercial signal across the whole business.

Would I test ChatGPT Ads today?

Not yet.

The measurement is improving, but at the moment I don’t want to take meaningful budget away from a proven channel to fund a test that needs enough spend to teach me something useful.

There’s also a wider platform question.

Google versus Bing.

ChatGPT versus Claude.

Where is the audience actually spending its time?

New channels are always interesting.

That doesn’t automatically make them the best place for the next pound of budget.