Return on ad spend fell. Nobody on the team had changed anything.

It started with a question.
Sam Okaforgrowth analyst at Simcorp Beautyin Veloccy
Why is ROAS sitting near 2.5, and should we pull the budget back before the festive build?
AI with access
Access
Experience
none
2.07 yesterday’s ROAS, against a usual 3.0 to 3.5

With access to the data, the budget rise did it.

Return on ad spend had been around 3 all of August. The largest campaign’s budget went up, and the number fell. Pulling the budget back before the festive season is the obvious move.

AI with experience
Access
Experience
Purchases settle by day fiveMeta ad account · change logand more
3.02 → 2.73 the part of the drop that is real

With experience of the account, yesterday’s number is not finished.

Veloccy knew what an experienced buyer knows by heart: Meta’s purchase counts read low for days and then settle. On settled days the real drop is smaller, and the campaign whose budget went up is the one holding its return.

How it happened in Veloccy.

A simulated beauty brand selling on its own site, with a Meta ad account and a creator programme. We built it and planted the cause, so we know what the right answer is. These are a few moments from the work.

Simcorp Beauty Arcs PM-1Meta ROAS is below its usual range
Running

The arc for this work, as the team sees it: what Veloccy read, what the work led to, and the thread it happened in.

01It notices first

Veloccy had opened the work before anyone asked.

It checks every number against its usual range, and it took this one to the person who owns the ad account. Its first line was a warning about the number itself.

Simcorp BeautyArcsPM-1
Opened by VeloccyAKAdam Keane
Meta ROAS2.07
usually 3.00 to 3.51

Before this is read as a drop: Meta’s purchase counts read about 18% low after one day and settle by day five. On settled days the move is 3.02 to 2.73. That part is real, and I am looking at why.

02It knows what changed

The cause was a change nobody on the team made.

It was in the account’s change log, at 03:17 one night, made by “system”. Veloccy also looked outside the company and found the trade press reporting the same change that week. Two days later, legal changed the creator brief, which Veloccy called a strong lead and not proof.

Simcorp BeautyBrainCalendar
Calendar

Placement exclusions removed

22 Aug · 03:17 · changed by system

Spend on those placements
0% → 21.2%
Return there
0.89, against 3.16
Share of the real drop
64.7%
Also moved by
Creator brief v2, 24 Aug
03It writes down what it expects

The bet keeps the budget and blocks the placements.

Pulling the budget would have cut the one campaign that was holding up. Veloccy wrote the change, what it expected, and the limits that would undo it, and took it to the person who owns the account. This bet has no result yet. When the read is due, Veloccy comes back by itself.

Simcorp BeautyChangesPM-1

Block Audience Network and the right column. Keep the budget.

PM-1
Predicted effect

Account ROAS 2.66 → 2.85

Limits

ROAS not below 2.59 · Messenger not above 2% of spend

Rollback

Lift the block if spend collapses in the affected campaigns

AKApproved by Adam KeaneApproved · read due after a settled week

Why access to the data was not enough.

01The first job was to distrust the number.

With access, yesterday’s 2.07 is a fact. With experience of this account, it is a number that has not finished arriving.

02The cause was in no dashboard.

It was in a change log, a legal brief and a news story. Knowing where to look is most of the answer.

03The obvious move was the expensive one.

Cutting the budget before the festive season would have cut the campaign that was working.

What stayed in the Brain.

Every one of these is experience the next question starts from, in every team. See the Brain

BSimcorp Beauty
  1. Known issuePurchases settle by day five

    How Meta’s purchase counts settle, so nobody chases yesterday’s number.

  2. CalendarPlacement exclusions removed · 22 Aug

    The night the placements changed, and the day the brief changed.

  3. ChangePM-1Block two placements, keep the budget

    The bet, its limits, and when its result is due.

Next use case · Growth and retention The data said a bigger first order makes a loyal customer. It was mango season.

Which bet is on your desk this month?

Show us, and we will show you what Veloccy does with it.

Talk to us

Become a design partner

We’re working with a small number of teams before a wider release. Tell us a little about yours and we’ll reply personally, usually the same day.