Where the Twitter Budgets Went

Six months ago, brand safety teams across the industry started pausing Twitter spend, and the trade press covered every pause like a resignation letter. The more interesting story has gotten less coverage: where did the money go? Because paid budgets do not evaporate, they reallocate, and watching that reallocation across our own client mix this spring has been a live seminar in how media planning actually works when the slideware is stripped away.

What we saw in the accounts

Across the accounts we manage, Twitter was rarely a headline channel to begin with: for most brands it was a single-digit percentage of paid spend, useful for tech audiences, live moments, and a certain kind of high-intent B2B reach. When those dollars moved, they went, in rough order:

• To Meta, quietly. The same advertisers who spent 2022 complaining about post-ATT performance discovered that Advantage+ campaigns had meaningfully improved while they were complaining. The least fashionable answer was the most common one.
• To retail media and search, where the measurement story is cleanest. In a nervous economy, money flows toward provable.
• To LinkedIn for the B2B crowd, at CPMs that will make you wince, but with targeting nobody else can honestly claim.
• A slice to experiments: Reddit, podcasts, newsletters. The exodus gave teams political cover to test channels they had wanted to try for years, which may end up the healthiest side effect of the whole episode.

What mostly did not happen: a triumphant migration to any single "new Twitter." Audiences fragmented rather than relocated, and media plans followed them into fragmentation.

The uncomfortable lesson about incrementality

Here is the part clients did not expect. Several brands paused Twitter entirely, braced for impact, and measured no detectable loss. Not "acceptable loss." None. Which raises the impolite question: what was that budget line doing before? The honest answer is that many channel allocations exist because they existed last quarter. The exodus functioned as an accidental, industry-wide incrementality test, and every marketer should absorb the result: if turning a channel off changes nothing, the channel was measurement theater.

The generalizable practice is to run that test on purpose. Geo-based holdouts, deliberate pauses, matched-market experiments: unfashionable, occasionally scary, and the only way to know whether a channel earns its line. We now build one deliberate holdout into every quarterly plan. The channels that survive get defended with evidence instead of habit.

Brand safety is a spectrum, not a switch

The other lesson worth keeping: "brand safety" turned out to mean different things for different brands, and the ones who navigated this well had decided in advance what their actual thresholds were. A luxury client and a developer-tools client should not have the same adjacency tolerance, and neither should be improvising it during a news cycle. We now document a per-client safety posture: what adjacency risks matter, who decides, and what triggers a pause versus a monitoring note. Written calmly, in advance, it converts a panic into a procedure. This will not be the last platform to have a chaotic year.

What we are telling clients for H2

Twitter itself, or X, or whatever it is called by the time you read this, may stabilize; ad products are reportedly improving and the audience has not vanished. We keep a small test line open for clients whose audiences genuinely live there. But the portfolio lesson stands regardless of one platform's fate: channel concentration is risk, provable beats plausible, and every budget line must periodically re-earn its existence. The advertisers who came through this episode strongest were not the ones who predicted it. They were the ones whose plans never depended on any single channel behaving well forever.