App Masters · Feb 2What is Marketing Mix Modeling (MMM)
CEO Roy Nam explains marketing mix modeling as an ATT friendly way to measure channels using aggregated daily spend and outcomes. Airbridge only models channels whose spend data marketers share, such as a podcast spending $200 to $400 a day.
Tips
- Use marketing mix modeling to regain visibility lost to ATT▶ 3:17Opt-in rate for the advertising ID has dropped to around 20 percent, so the matching rate between conversions and touch points has fallen dramatically.
- Feed the model daily spend per channel plus daily outcomes▶ 4:55Example: $1,000 to channel A, $2,000 to channel B and $3,000 to channel C on day one, with a holiday yes/no flag.
- Stop relying on last touch attribution alone▶ 0:43
- Include offline and brand channels like TV and podcasts in measurement▶ 9:54Steve cited running TV and podcast ads at Walmart.com, where tracking was hard because people just searched afterwards.
- Judge channels by their effect on the whole mix, not just cost per install▶ 8:02A developer told Steve he knew a channel looked unprofitable but believed it helped the overall marketing mix.
- Account for ad carryover when judging spend▶ 10:28
- Model saturation so you know when extra spend stops paying▶ 10:40
- Use MMM simulations as an advisor for next budget moves▶ 26:25
- Set a CPI ceiling from your budget and target user count▶ 11:56Example: 5,000 users for 5,000 USD requires CPI always below one USD.
- Check whether organic installs are inflated by lost attribution▶ 6:54
