90%
drop in cost per install on FESTAS, our own product, once budget and creative followed consolidated app attribution (AppsFlyer) instead of platform last-click.
Services
On-demand marketing mix modeling06 / 07
Budget allocation driven by incrementality, not last-click
We model each channel's true incremental impact with Google Meridian or Meta Robyn and turn the curves into practical budget reallocation scenarios. Project-based, no annual software license.
Max M · live operation production
signal qualitylive
When it makes sense
When MMM makes sense
- 01
You spend across three or more channels and every platform claims credit for the same conversion.
- 02
Last-click is the official budget yardstick, but nobody in the room actually trusts it.
- 03
iOS 14+, cookie deprecation and modeled conversions have left your digital attribution full of holes.
- 04
Upper-funnel channels (YouTube, TV, OOH, influencers) look expensive because last-click can't see their effect.
- 05
Questions like 'what happens if we cut this channel by 20%?' get answered with gut feeling.
- 06
You've priced enterprise MMM platforms and the annual license doesn't make sense at your stage.
How it works
How it works
- 01
Data audit
We gather 18 to 24 months of history: spend per channel, outcomes, seasonality, promotions and pricing. Before modeling anything, we tell you straight whether your data supports a reliable model.
- 02
Modeling and calibration
We build the model in Google Meridian or Meta Robyn, the two open-source reference frameworks, with statistical validation, robustness checks and calibration against experiments where available.
- 03
Reallocation scenarios
We turn response curves and saturation points into decisions: where to cut, where to scale, and the mix that maximizes return on the same total budget.
- 04
Quarterly recalibration
Every quarter we re-run the model on fresh data, compare forecast to actuals and adjust the plan. The model keeps learning from your operation instead of becoming a forgotten slide.
Deliverables
What you get
- A calibrated MMM in Google Meridian or Meta Robyn, with code, data and documentation you own outright01
- Response curves and saturation points per channel02
- An incrementality read: what each channel truly contributes, including the baseline you'd get anyway03
- Simulated budget reallocation scenarios: cuts, increases and the optimal mix04
- An executive deck with prioritized recommendations, ready for your board05
- Quarterly recalibration with forecast versus actuals06
Results
Results
10x
more installs for the same spend after reallocating budget to the formats and channels with the highest incremental response.
80%
drop even under the most conservative read, the platform's own attribution: when two independent sources tell the same story, the decision becomes defensible.
Every platform claims the conversion is theirs. Marketing mix modeling answers the question digital attribution can't: how much each channel actually adds, and what happens to results when the budget moves.
Last-click is lying to you
Click-based attribution sees the end of the journey and ignores the rest. Upper-funnel channels look expensive, retargeting looks cheap, and budget drifts toward whoever harvests the conversion rather than whoever created it. iOS 14+, cookie loss and modeled conversions made the gap worse.
MMM attacks the problem from a different angle: instead of tracking people, it statistically models the relationship between spend and outcomes over time, controlling for seasonality, pricing and promotions. It's the technique large advertisers have used for decades, now accessible through open-source frameworks: Google Meridian and Meta Robyn.
Our difference is the format: an on-demand project, not an annual license. You get the calibrated model, the scenarios and the recommendation. And you keep the code.
Questions the model answers
- What is the incremental contribution of each channel, including the ones last-click ignores?
- Where does each channel saturate and start burning budget?
- What happens to results if a channel is cut by 20%? And if it doubles?
- How much of your outcome would happen anyway, with zero media?
- Which mix maximizes return on the same total budget?
From raw data to a reallocation plan
- 01
Audit: 18 to 24 months of spend and outcome history, with an honest viability read.
- 02
Modeling in Google Meridian or Meta Robyn, with statistical validation and robustness checks.
- 03
Simulated reallocation scenarios: cuts, increases and the optimal mix for the same budget.
- 04
Quarterly recalibration: forecast versus actuals, model updated, plan adjusted.
90%
drop in cost per install once our own product's media followed consolidated attribution data instead of last-click
10x
more installs for the same spend after data-driven reallocation
0
annual licenses: you pay for the project and keep the model
FAQ
Frequently asked questions
How long does an MMM project take?
Four to eight weeks from kickoff to scenario presentation, depending on data quality. Quarterly recalibrations are much faster: the model already exists and is simply re-estimated on fresh data.
How much historical data do I need?
Ideally 18 to 24 months of spend and outcomes at weekly granularity. With less than that, the initial audit tells you whether a model holds up or whether it's worth collecting a few more months first.
How does pricing work?
Fixed-scope project: audit, modeling and scenarios, priced upfront before kickoff. Quarterly recalibration is an optional recurring package. No software license, no mandatory retainer, no percentage of media spend.
Who owns the model and the data?
You do. Code, data, documentation and assumptions live in your environment. If you later want your data team to run it in-house, there is zero agency lock-in.
How do you work with teams in other time zones?
Async-first: written updates, recorded walkthroughs and live calls scheduled in your time zone. Deliverables in English, Portuguese or Spanish.
Stop allocating budget on gut feeling
Tell us which channels you run and what you measure today. The initial audit tells you whether an MMM holds up on your data and what it would unlock.
Talk to Max MOther capabilities
The operation does not end at this deliverable.
Paid media management
More conversions from the same ad budget
Campaign management across Meta, TikTok and Google with structured accounts, data-driven creative and CPI, CPA and ROAS optimization. Our most recent engagement cut cost per install by over 90%.
Explore serviceDigital analytics and CRM
Analytics you can trust, CRM that converts
We implement your full data collection layer, from in-app event to warehouse, and activate behavior-based CRM across push, in-app, email and WhatsApp. Governed taxonomy, unified identity, funnels measured end to end.
Explore serviceAI creative and content automation
Creative and content at scale, without losing your brand voice
AI pipelines that turn one real product screenshot into 180+ ad variations, plus agents that research, write, review and publish content. For FESTAS, cost per install dropped by more than 90%.
Explore service