Mix modelling for $1M/month spenders

Quick Answer

Marketing mix modelling (MMM) is statistical analysis that estimates the contribution of each marketing channel to total revenue, accounting for cross-channel effects, seasonality and external factors. At $1M+/month spend, single-channel attribution becomes too unreliable to drive decisions, and MMM becomes the most accurate (though imperfect) measurement option. It typically costs $30-100k for an annual setup and gives you channel-level lift estimates that incrementality tests alone can't.

The Framework

1. Why MMM matters above $1M/month

At this spend, decisions about channel allocation move millions of dollars. Attributed ROAS is too noisy to trust. MMM gives you statistical confidence intervals around channel contributions, even accounting for what other channels are doing simultaneously.

2. What MMM actually delivers

Channel-level lift estimates (Facebook contributed X% of revenue), saturation curves (your next $100k on Facebook will produce Y in revenue), and cross-channel effects (Facebook's brand effect lifts Google search). It does NOT give you ad-level optimisation — it's strategic, not tactical.

3. Choosing a vendor

Established MMM vendors: Recast, Lifesight, Northbeam, Triple Whale (with their MMM module), or custom-built with a data science team. Start with established tools — building MMM from scratch is a year-long project.

4. Data requirements

MMM needs at least 12 months of weekly data across all channels: spend, impressions, conversions, revenue. Plus external factors: weather, holidays, competitor activity. If you don't have this history, MMM can't run.

5. Refresh quarterly, not annually

MMM models drift as your channel mix and audience change. Quarterly model refreshes keep it accurate. Annual refreshes are common but produce stale insights for fast-moving accounts.

6. Combine MMM with incrementality tests

Use MMM for strategic channel allocation. Use incrementality tests for tactical campaign decisions. They answer different questions and aren't substitutes for each other.

Real Numbers from the Field

A fashion brand spending $1.4M/month commissioned MMM after their attribution numbers stopped matching reality. The MMM revealed Facebook was contributing roughly 22% of total revenue (vs 41% claimed by attribution) and TikTok was contributing 18% (vs 9% claimed). They reallocated $200k/month from Facebook to TikTok over 4 months and total revenue grew 14% — MMM had identified a real underinvestment in TikTok that single-channel attribution couldn't see.

Frequently Asked Questions

Is MMM worth it below $1M/month?

Usually no. Below that spend, the cost of MMM ($30-100k+ annual) outweighs the decision-quality improvement. Stick with incrementality tests.

How long does an MMM project take to complete?

First model: 6-10 weeks. After that, refreshes can run in 1-2 weeks. Don't expect insights in week one.

What's the biggest MMM risk?

Trusting the model output without sanity-checking it against reality. MMM is a statistical estimate, not a truth machine. Always validate with hold-out tests.

Can I build MMM in-house?

Possible if you have a data science team. Recast and similar vendors have made this less necessary by productising MMM. Most brands at $1-5M/month buy rather than build.

Should I tell my media buyers the MMM results?

Yes — but warn them it changes channel allocation, not campaign tactics. Buyers shouldn't optimise individual ads against MMM output.

Stop Guessing What Your Ads Look Like at Scale

When you're spending £5k+ a day, you can't manually screenshot every ad on every device, in every placement, in every locale. Things break silently. Creators ship the wrong aspect ratio. Localised text overflows. Safe zones get clipped by Reels UI.

Pix-Vu lets you preview every ad creative across Facebook, Instagram and Reels placements before launch — desktop, mobile, all aspect ratios, all crop variants. Spot the broken creative before you spend £20k discovering it the hard way.

Try it free at https://pix-vu.com — built for media buyers who can't afford a single wasted impression.