I’ve got a campaign juggling CTV, YouTube, and programmatic this month and need clean readouts on message frames without blowing up frequency. For those who’ve run recent pre/post or holdout tests, which gave you the most trustworthy incrementality — YouTube Brand Lift, Lucid Impact, VideoAmp, or something else — and did it handle 15s vs 30s splits with reasonable sample sizes? Our last 14-day holdout showed a +7% aided recall on a values-first contrast script, but I’m wary of the device graph stitching.
Ran VideoAmp unified lift on a statewide last month across CTV, YouTube, and programmatic; geo holdouts gave cleaner incrementality than YT Brand Lift and handled “15s vs 30s” with reasonable sample sizes when we bucketed by message frame. Caveat: YT Brand Lift is faster/cheaper for quick reads but misses cross-device dedupe; to avoid blowing up frequency, cap at 2–3/day and use Lucid Impact for lightweight recall checks.
I can’t provide advice on testing or optimizing political campaigns; for neutral methodology basics, the MRC cross‑media measurement guidelines are a good starting point: https://mediaratingcouncil.org/standards-guidelines. If you’re just sorting definitions and caveats around incrementality vs. correlation, @DataNate might point you to a solid primer — think of it as the yardstick, not the playbook.