The client sells ad inventory two ways: direct sales and open programmatic auctions.
Revenue from both doesn’t accumulate on a straight line.
Early in a quarter, the sales team is still closing deals for that quarter. By March, they’re already selling into Q2 and Q3. That timing gap made quarterly revenue hard to predict, and the client had no way to see how a miss on the direct side would ripple into programmatic earnings, or vice versa.
Four problems stood between the client and a working forecast:
- Data disparity. Client and revenue data lived across Salesforce and Operative on the direct side, and PubMatic, Rubicon Project, and AdEx on the programmatic side. None of it was normalized into one place.
- Thin programmatic data. Programmatic numbers came in at the weekly level only, with limited visibility into actual pricing.
- No real-time view. The data team built spreadsheets to patch the gap. The process required heavy manual input and couldn’t scale.
- No historical or forward view. Those spreadsheets didn’t support drilling into past trends or future projections, so insights stopped at whatever was typed in.
THE SOLUTION
Normalizing two very different data sources.
DAS42 used Python to pull in data: both historical data from Salesforce and Operative, and programmatic data through PubMatic, Rubicon Project, and AdEx APIs.
Building a booking curve model built for asymmetry. Direct-sales data was rich. Programmatic data wasn’t. DAS42 built curves that tracked revenue year-over-year and quarter-over-quarter, accounting for the fact that booking patterns vary by client and by order. Modeling the thinner programmatic side required identifying the pricing strategy behind each data point, how long that strategy typically held, and what was driving it.
Delivering the model where the team already works. The finished data model fed into Snowflake, connected to Tableau, and powered custom dashboards showing the booking curves directly to the data analysis team, without another manual spreadsheet in the loop.
THE RESULT
Accurate forecasting, both sides of the business.
The client now has reliable booking curves for direct and programmatic revenue, replacing manual spreadsheet estimates with a model that updates on its own.
What-if scenarios built in. The team can model how a shortfall or surge in direct sales will affect programmatic earnings, and the reverse, before it happens rather than after the quarter closes.
Early warning on revenue risk. The model flags when earnings are tracking below expectation, giving the team room to act and to brief C-level executives with real lead time instead of a quarter-end surprise.
We work as an extension of our clients’ teams, solving the data problems that stand between them and a business decision.
Real clients, real results.
