The Curious Case of the Sunk CPA: Part III

Note: This is the third part of a series on the fastest account turnaround I’ve done. You should read Part I and Part II if you haven’t yet!

So, now I figured out one thing for sure:

The client had introduced a super-low price point on a new product. The spend on that scaled rapidly.

The shift in the pixel accounting for this new type of customer started to hurt the results of their normal, higher-priced product point. They cut spend on the campaigns optimizing to that normal product at the same time.

So they scaled down the normal one while scaling up the cheap one.

Disaster.

This obviously all happened in the background, the account manager at the time didn’t notice it, and the shiny agency that took over didn’t figure it out, either.

But was that all?

I wish it was.

Maybe Optimize a Little?

Pixel issues aside, the account was rife with laziness. There was a lot of things launched, that never seemed to be touched after, aside from some feeble attempts at new creative.

One of these patterns was performance by gender. Every ad set targeted both genders, but there were so many instances of one gender performing horribly and never being turned off, like so:

There were also clearly ages that didn’t perform, but do you think those were ever excluded? Of course not.

Something I had noticed in the initial setup of the Old Manager were some ad sets that specifically focused on one device type. There was plenty of data to suggest Shiny Agency should have done the same – here’s one such example from an ad group that continued to just run all devices, regardless of performance:

Lazy. Lazy. Lazy.

Hey, Let’s Spend Even When Results Suck

I looked in the account history as well to get a sense of the cadence Shiny Agency had, and was really confused to see budget increases. Things were performing horribly, but they kept increasing the spend. I could maybe see that when the account is first ramping up to get data faster, but it was done on stuff that was clearly not performing.

Check out the weekly spend increases here, and how the nowhere-near-the-goal-CPA continues to be awful:

 

The other thing that seemed really obvious to me and continued to be ignored was that the remarketing results were terrible. They went from a $20-range CPA to over $50.

This was also an indication there was an issue in the selling path…remember, there was another product in the mix now. So, you have people who visited the $20 product getting remarketing ads for an almost-$200 normal product.

We fixed the pixel issue, and I started rebuilding things. We were making progress! The CPA was going down, we were inching up volume, there was a huge light at the end of the tunnel.

Right as we were consistently getting some ad sets to work and scale…

Dun dun DUUUUUUNNNNNNNNNNNNN!

Facebook suspended the account.

For no verifiable reason, naturally. We were pleading with reps. In the midst of it, I scrambled to rebuild everything in a new account while trying not to lose my mind because we were losing all that history…

Or were we?

Pro tip: You can share your account pixel with other ad accounts. This saved. our. butt.

We opened a new ad account, shared the existing pixel within the suspended account, recreated our lookalike audiences, and marched on as fast as we could considering we had to rebuild so many things.

We actually managed to start to get THAT CPA under control within a few days…

And then….

The other account GOT TURNED BACK ON. With zero notice – just…POOF! Restart!

Wild. I try to be a silver lining person, so hey: we have a backup account if it ever happens again.

Would Shiny Agency Have Eventually Succeeded?

Without the pixel issues, Shiny Agency would have at least done better than they did. That said, they never would have caught up to Old Manager’s results.

They weren’t optimizing for the wins they WERE getting. They continued to launch and let things run in subpar channels and devices, to genders that weren’t performing, and to age groups that were clearly not interested in the offer.

This is also part of the reason I do not follow Facebook’s advice to let everything run everywhere because the algorithm will optimize it anyway. It frequently does not, and will continue to spend on things that have terrible results.

And Now?

I’m happy to say the results have become exponentially better since all of this was figured out, fixed, and we have now scaled to the point where we are holding on spend because their suppliers are in danger of falling behind on orders. CPA is within striking range of the old one, and we’ve been holding there since the increase of volume offsets the few bucks more a pop.

🙂 Yee haw!