← Writing

Why did Meta Value Optimization stop working?

Meta confused me, and I am not afraid to admit it.

I thought running Value Optimization alongside a bid cap for a direct purchase app was clever. Early on, back in January, it seemed to work. tROAS was not spending, while VO and the bid cap were delivering solid ROAS.

Then February happened. March was worse.

Meta bid strategy ROAS, PROAS 365, week over week from 8 February to 5 April. Value optimization starts near 89 percent, peaks at 111 percent on 15 February, dips to about 93 percent on 22 February, recovers to about 105 percent on 1 March, then falls to 70 percent, 34 percent, 0 percent by 22 March and recovers only to 12 percent. tROAS tracks alongside at 105 percent on 1 March then settles between 34 and 56 percent. Bid cap runs between 20 and 29 percent through February then drops to zero from 8 March. Weekly PROAS 365 by bid strategy on one direct purchase app. Value optimization in purple, tROAS in green, bid cap dashed in red.

VO fell off a cliff. We scrambled and tried a lot of things. Eventually we found stability back in tROAS, which held between 34% and 56% through the same stretch where VO went to zero.

What I take from it

The bid cap was the first thing to go quiet, dropping to zero from the second week of March and never recovering. VO followed a few weeks later.

What made this hard to read in real time is that VO’s best weeks came first. When a strategy spends its first month mostly above 100%, you attribute it to the setup rather than to the conditions. The collapse then looks like something breaking, when the more likely reading is that the conditions that made it work were never permanent.

Will VO ever work for us again on this account? I genuinely do not know. We will test it. But it is not where the budget sits while tROAS is holding.

If you are running a similar setup, watch whether the strategy holds its level once the account changes around it, not just the level itself.