Measurement
Platform reports disagree and attribution overstates the last click. Inflexion combines media mix modelling, incrementality testing and attribution into one view of what drives sales, then turns it into budget decisions.
The problem
Each platform credits itself for the same sales. Added together, the reports claim far more than the business actually sold.
Platforms claim 1.7 times the sales that actually happened. Budgets set on these numbers favour whichever channel sits closest to the purchase.
Illustrative example. Figures are for demonstration only.
Our methods
No single method is reliable on its own. Each answers a different question, and each checks the others.
Statistical models of how every channel, price change and external factor affects sales over time.
Controlled experiments, such as geo and audience holdouts, that show what a channel actually caused.
Campaign level signals used for daily optimisation, adjusted using the model and test results.
From measurement to decisions
Response curves show where each channel starts to saturate. The plan moves spend from channels past that point to those with room to grow.
| Channel | Current | Recommended |
|---|---|---|
| Search | 38% | 27% |
| CTV | 12% | 22% |
| Retail media | 30% | 33% |
| Social | 20% | 18% |
Illustrative example. Figures are for demonstration only.
What sets the approach apart
Media models rarely include AI search or the retail shelf, yet both now shape sales. Inflexion measures them alongside media.
Share of AI answers against competitors, tracked monthly as a leading indicator of demand.
AI search →Share of shelf, availability and listing quality across Amazon and other retailers, linked to sales.
Media and retail media →Branded search and direct traffic, used to capture the longer term effect of brand media.
Models are configured for each business rather than run on a platform's default settings. Open source frameworks such as Meridian are used only where they fit.
Clients own their data, models and results. Methods and assumptions are documented so finance teams can review them.
Every output ends in a recommendation on budget, channel or creative, reviewed with the people who make those decisions.
Our approach
Four stages, each with a defined output. Most clients start with the audit.
Review data sources, current reporting and the questions measurement needs to answer.
Build the media mix model and design the first incrementality tests.
Model budget scenarios and agree the reallocation with marketing and finance.
Refresh the model monthly, run tests continuously and report in one dashboard.
Questions
Media mix models work best with around two years of weekly sales and spend data. Where history is limited, incrementality tests provide early answers while the data builds.
No. Attribution stays useful for daily optimisation. The model and tests correct its bias towards the last click, so it can be used with more confidence.
Methods, assumptions and results are documented and shared in full, and clients own the models. That makes the measurement open to review by finance or a third party.
Yes. Retailer data, including Amazon Marketing Cloud, feeds the model and supports incrementality tests within retail media networks.
A clear view of how reliable current reporting is, and what it would take to measure with confidence.
hello@inflexionmedia.com