.png)
The global rollout of Meta's Andromeda update created a frenzy. Ever since, a single question has dominated strategy meetings: what does the algorithm really want now? Since October last year, LinkedIn has been awash with typical hyperbole: that accounts need to be torn up and started again, and marketing teams have been inundated with requests for more volume.
We went into our latest tests with a similar assumption: that higher ad volume would naturally lead to better performance. We wanted to see just how much of a lift we’d get. The results surprised us, and they challenge that core "more is more" assumption.
We designed the tests to be as fair as possible. We used the same budget, audience, and optimization goals, controlling every variable except for the number of ads in the ad set. All creative used in the tests was diverse and high-quality. We let the ad sets fully exit the learning phase first, just to make sure the performance was stable and the comparison was fair.
The parameters were:
The data came back with a surprisingly clear and consistent story:
Across the board, even with a diverse range of creatives, simply increasing the volume of ads didn't just fail to improve performance; it actually hindered it.
Of course, both volume and content diversity are important on Meta. But these tests prove that similarity is the more important lever for performance.
We already know that content similarity is fundamental to Meta's algorithm since the Andromeda update. Our tests reinforce this from a new angle. The content within these tests was diverse, but simply scaling the volume had a limited and, in fact, negative impact on performance. This suggests that while diversity is crucial, the algorithm's ability to distinguish between creatives is the most critical factor, and this becomes less efficient as volume increases.
The insight here is that we should shift our focus from "How many ads can we make?" to "How many distinct, high-quality ideas can we test?"
The results show that focusing on creative quality and strategic diversity is far more important than simply hitting an arbitrary volume quota. Our recommendation is to empower brand and creative teams to spend their time developing three to five truly different conceptual routes for a test, rather than feeling the pressure to produce 50 variations.
It seems the path to more efficient marketing spend and better results isn't about feeding the machine endless content, but about giving it higher-quality ingredients to work with.
