Enterprise model fine-tuning
Test your next move before you commit.
We fine-tune a model of your market on your data. Your teams compare the alternatives before the budget is spent.
Launch for at
Next quarter’s budget is committed before the market responds.
Your evidence trains the model.
We fine-tune on your company data and public market data. The model represents your current customers, the customers you could win and the alternatives they choose instead.
- Public dataThe market beyond your customers.
- Previous experimentsWhat you have learned.
- CRM dataWho you know.
- Purchase dataWhat people choose.
Example model: what 291 simulated buyers choose, scaled to a market of one million shoppers
| Metric | Source | Result | With the intervention |
|---|---|---|---|
| Simulated buyers | CRM data | 291 buyers | |
| Choose your jacket | Previous experiments | 72 buyers | 173 buyers |
| Share who choose it | Computed | 24.6% | 59.6% |
| Market shoppers | Public data | 1M shoppers | |
| Jackets sold | Computed | 246K jackets | 596K jackets |
| Margin | Purchase data | €179 a jacket | €176 a jacket |
| Profit | Computed | €44M | €105M |
Your turn
Pick an intervention and a price, or let the model optimize.
Each curve is one intervention: a bioplastic share and a certificate. Its price moves your jacket along the curve. Each optimize button finds the top of the top curve for its goal: the most revenue, the most profit, or the most market share without a loss.
- Your intervention: its price moves the jacket along it
- Efficient frontier: the best of all interventions
Simulated rerun of a published rain-jacket study (Klein et al., 2020): 291 simulated buyers choose this jacket, a €109 rival or neither. Profit uses example costs (€50 to make a jacket, €10 more for 50% bioplastic, €20 more for 100% and €3 for a certificate); the study did not measure costs. This run is not yet checked against human buyers. Across 43 published experiments we replicated, simulated results matched the human results with a rank correlation of 0.73 (the paper).
Bring the decision. We build the model around it.
Baseline
Before work starts, we agree with you on the decisions in scope and the baseline that savings are measured against.
Train
We connect your company and public market data and fine-tune the model on it.
Evaluate
We test it against evidence held out of training.
Use
Your teams get model access, experiment workflows and help interpreting results. We measure the savings against the baseline.
One model answers every team's question: marketing, sales, pricing and product. We confirm the best plan with a small live test before you commit the budget.
Launch your offer for your customers at your price.
Scoped to your decisions and data.
Performance-based engagement$1M
- Fee due
- After $10M in savings, measured against the agreed baseline
- Until then
- $0. The full fee is at risk.
We build the business case and the baseline with you before you commit.
Build your business caseWe are recruiting paying customers and will develop results through these engagements.