# CPG: Know what grows each brand.

Brand investment, distribution, media and portfolio decisions, connected. Understand what contributes to sales over time and what to test next quarter.

## What matters

Consumer brands sell through partners they don't control. Investment is decided by brand and category, while sales arrive through distribution, promotions and the market.

- Investment by brand and channel
- Distribution, sell-in and sell-out where available
- Portfolio and category roles
- Penetration and incrementality

## Questions you can ask

- Which media channels are driving incremental sales?
- How much of growth is associated with media versus distribution?
- Which audiences are underpenetrated?
- Which creative territory works best for each brand?
- How should we allocate next quarter’s investment?

## Data that becomes useful

- Media investment and activity by channel
- Sales by brand, category, geography and period
- Promotions and distribution
- Market context that is available

Sell-out coverage varies by retailer and market. Where it is missing, the analysis says so.

## What the agents can do

1. **Understand (Nexus):** Analyze contribution by channel and the hypotheses behind a change in sales.
2. **Define (Aura):** Prioritize the brand opportunity and the audience that is underpenetrated.
3. **Create (Prism):** Develop concepts for each brand from its territory and approved claims.
4. **Activate (Radian):** Prepare the campaign and the test that will measure it.
5. **Learn (Shared system):** Compare results with the model’s assumptions and update the next investment decision.

### Marketing mix modeling with Google’s Meridian framework

Marketing mix modeling relates sales to marketing activity over time, taking other factors and stated assumptions into account, to inform investment decisions. Random Truffle runs MMM workflows with Meridian, Google’s open-source framework.

- Contribution by channel
- Return estimates by channel
- Response curves
- Budget scenarios

Outputs depend on the data and the fitted model. Geo-level analysis and calibration with experiments are possible when the data supports them; they are not automatic. A model estimate is not proof of a single campaign’s effect.

## How learning closes the loop

The model informs where to invest; experiments test what changed. MMM does not create an individual audience on its own, so the loop carries its findings into Define as an opportunity to test.

- Incremental sales, estimated or measured according to the method
- Contribution by channel
- Creative response by brand
- Investment decisions and their review

## An example conversation

> **You:** How much of SOLA’s growth this year is associated with media?
>
> **Nexus:** In this constructed model, baseline demand explains most of the volume and paid search is the largest media component. Distribution also widened in the same months, so I’d treat the media share as an estimate and test the search budget before moving it.

*Illustrative conversation with a fictional brand. Not a transcript or a client result.*

## Start with the investment decision.

Explore how contribution, credit and incremental effect answer different questions.

---

Source: https://randomtruffle.com/solutions/cpg