<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=331640213977849&amp;ev=PageView&amp;noscript=1">
Marketing mix modelling

Where should the next marketing budget go?

We model how each channel, campaign and price move drives sales, and check the model against experiments. You plan the budget on what returns, not on what is easiest to count.

See the decision tree
Two colleagues reviewing data on a laptop
Decisions
Next-gen marketing mix modelling

What the model shows

Signal 01

Return by channel

The model separates base sales from what each channel adds, and shows the return on every channel side by side.

Which channels earn more of the budget, and which earn less.

Signal 02

Privacy-friendly by design

The model runs on aggregate data, so it does not depend on cookies or tracking individual users.

How to keep measuring as tracking rules tighten.

Signal 03

Scenarios before you spend

Budget splits are tested in the model before the plan is booked, with expected sales for each.

Which plan to commit to, and how much to spend.

The decisions on the agenda

Attribution counts the clicks. It misses the rest of the mix.

Cookie-based attribution is fading, and it never saw TV, outdoor, price or the shelf. A marketing mix model reads all of it together. We build it, check it against experiments and turn it into a plan.

The calls clients bring us
  • Set next year's budget split

    Allocate across channels on modelled return, not on last year's plan.

  • Prove marketing's return to finance

    Show what each channel added to sales, and at what cost.

  • Find where spend has stopped working

    See the point where more weight in a channel adds little.

  • Test a plan before it is booked

    Run the scenario first, then commit the budget.

Decisions informed

One decision, three questions, what the model returns.

This is a modelling project. Each question ends in an output from the model, and each output in something you do.

Decision How should we split the budget to grow sales?
Research questions
Model outputs
Base sales and trendWhat you would sell with no marketing at all.Media contribution by channelSales each channel added on top of the base.Price and promotion effectSales moved by price changes and deals.Seasonality and outside factorsHolidays, weather and economy, separated from your effort.
Model outputs
Return on spend by channelSales back for every dollar in each channel.Diminishing-return curvesThe point where more spend stops paying.Carry-over effect of advertisingHow long an ad keeps selling after it runs.Calibration against experimentsModel results checked against real tests.
Model outputs
Budget scenariosWhat sales look like under each budget you are weighing.Optimised channel splitThe mix that gets more sales from the same budget.Forecast sales by scenarioExpected sales for each plan, before you commit.
What you receive

The model, the scenarios and the plan

A model of what drives your sales, checked against experiments. A scenario tool to test budget splits. A readout that ends in the split we would book.

Talk to our team
Mix modelVietnam · All channels · 104 weeksModel v2
Media share of sales18%
Highest returnOnline video
Past saturationTV

Response curves by channel, with current spend marked

TV

SaturationTV is past its response ceiling at current weight.Each extra unit of spend returns less than half the first.
Plan readoutRecommendation: move part of TV weight into online video.Same budget, higher modelled sales.
How we work

From brief to answer

We plan every step back from the day the budget is set.

  1. 01

    Brief

    We agree the decision, the budget it governs and the date the plan is due, and plan back from that date.

  2. 02

    Data

    We gather sales, media, price and outside data, and check it is complete before any modelling starts.

  3. 03

    Model

    We build the model, test it against experiments and agree the results with your team.

  4. 04

    Plan

    We show the return by channel, run the scenarios with you and say which split we would book.

The team

The people who run this study

The researchers who design the study, run it and read it out with you.

Brief us

Tell us the budget decision and the date the plan is due

We will tell you whether your data can support a model, or what to collect first.

Brief us your decision