Fifteen analyses, and how to read each one.
What every analysis in tea is for, how its result is read, where it usually misleads, and two cases for each. Grouped by the question you arrive with, in the order a practitioner would run them.
Discovery Path
Something moved. What moved it? You have a number that went up or down and five theories about why. Four analyses, in the only order that does not let one driver take credit for another.
Trend Analysis
The structural slope of a series, with its confidence band, read before anything else gets credit for growth. Where a trend line helps, and where it averages two different businesses into one.
Seasonality & Event Impact
Separating the calendar from the campaigns: seasonal indices, moving holidays and one-off events, measured with everything else held constant.
Contribution & Driver Decomposition
Marketing mix modeling in practice: what each driver added, week by week, with the intervals that say which numbers to believe.
Base vs Incremental
The split a board asks for: the year you would have had anyway, and what marketing added on top of it.
Media Effectiveness Path
The media ran. Did it work, and where does the next euro go? From how long a burst keeps working to where the money should sit next quarter. This is the path that ends in a reallocation, not in a report.
Lag & Carryover
How long a euro of media keeps working after the week it ran. Adstock, half-life and the distribution of an effect over time, and the files where it cannot be identified.
Saturation Curves
Response curves per channel: where the returns flatten, what the next euro is worth, and how wide the interval on the bend really is.
Contribution & Driver Decomposition
Marketing mix modeling in practice: what each driver added, week by week, with the intervals that say which numbers to believe.
Budget Optimization
Moving the same budget to where the next euro returns most, under real constraints, and knowing when a recommendation is too thin to follow.
Geo / Segment MMM
The same model per region or segment, with partial pooling: where the national average hides two different markets.
Optional, when the file has regions or segments
Pricing & Promo Path
Are we leaving margin on the shelf? How demand answers a price move, what the discounts really bought, and whether the competitor is the reason. Ends on a number you can take into a pricing meeting.
Price Elasticity
How demand answers a price move, measured with its band: when a rise pays, when a cut does, and when the file holds no price variation to learn from.
Promo Elasticity
What a promotion really bought, net of the demand it pulled forward, and why a calendar full of discounts cannot be measured against itself.
Cross-Elasticity
Which competitor takes your volume when they move their price, and which one only sits next to you on the shelf.
Scenario Simulation
Running the plan you are considering against the measured effects before committing money, and knowing where a simulation stops being evidence.
Planning Path
What does next year look like if we do nothing? The business-as-usual number, built well enough to defend in a budget meeting, and then bent by the decisions you are actually considering.
Baseline Forecast
What the next weeks look like with nothing new in them: the business-as-usual number a plan should be measured against, and the cases where the past is not a baseline at all.
Time Series (ARIMA)
Forecasting with ARIMA models chosen on the data, read through their residuals and their intervals. How far ahead a forecast stays useful, and how to tell.
Seasonality & Event Impact
Separating the calendar from the campaigns: seasonal indices, moving holidays and one-off events, measured with everything else held constant.
Scenario Simulation
Running the plan you are considering against the measured effects before committing money, and knowing where a simulation stops being evidence.
Competitive Path
They moved. Do we? The one path that starts outside your own data. What their activity structurally costs you, and whether answering it would cost more than it recovers.
Competitive Pressure
Share of voice against share of market, excess share of voice, and what competitor activity structurally costs you.
Cross-Elasticity
Which competitor takes your volume when they move their price, and which one only sits next to you on the shelf.
Promo Elasticity
What a promotion really bought, net of the demand it pulled forward, and why a calendar full of discounts cannot be measured against itself.
tea, the product
Run them on your own file
A CSV of weekly data. Each analysis in minutes, with its intervals and diagnostics, and a plain sentence when the file cannot answer. Free while in beta, by invitation.
Request a place in the beta→tea, the product
Fifteen econometric analyses on your own CSV: Marketing mix modeling, saturation curves, elasticities, budget allocation. The diagnostics shown, and a plain sentence when the file cannot answer.
Free while in beta, by invitation.
Request a place in the beta →Now run it on your own data.
tea runs Marketing mix modeling and fourteen other analyses on any weekly CSV, in minutes. Free while in beta, by invitation.