Bayesian tools for competitive and market analysis.
Weigh what a rival's move could mean, compare options against criteria you chose, and estimate what you cannot observe directly.
Market intelligence reads a rival from the outside. You see an announcement, a few reviews and some serial numbers on shipped units, and you must say what they mean and how sure you are. These tools weigh competing readings, compare options on criteria you chose, and estimate what you cannot count directly, with the uncertainty in plain view.
What do you need to find out?
What could that announcement mean, and what would tell the readings apart?
How much each reading gains or loses from each clue, and which clue would separate them. Competing Hypotheses · How do you tell which explanation the evidence favours?
Which option wins on the criteria that matter, and does it survive a reweighting?
The winning option and how far the weights must move before it loses. Weighted Decision Matrix · Is the winner of a weighted decision matrix really the best?
What is a product's true average rating, given a handful of reviews?
The average a rating is heading for, with a credible interval that narrows as reviews arrive. Rating Estimator · How many ratings before an average means anything?
How large is the rival, from what we can see of it?
A range for the units produced, from serial numbers on the units you can see. Serial-Number Estimator · What is the German tank problem, and how do you solve it?
How much can a rival still produce from the stock it has?
Stock left by grade, and when output is likely to fall. Stockpile Pipeline Estimator
Worked examples (synthetic data)
Instruments
- Competing Hypotheses
- Weighted Decision Matrix
- Rating Estimator
- Serial-Number Estimator
- Stockpile Pipeline Estimator
- Usage Count Estimator
- Bayesian Innovation Options Simulator
- Customer Value Calculator
- Beta-Binomial Calculator
- Expert Prior Builder
- Collection Planner
- Key Player Removal
- DAG Sketchpad
Articles
- Is the winner of a weighted decision matrix really the best?
- How much is an experiment worth before you run it?
- How many ratings before an average means anything?
- How do you tell which explanation the evidence favours?
- What is the German tank problem, and how do you solve it?
- How do you estimate how many people a contractor keeps on a site?