Value-based bidding signal, governed monthly from the CRM

A value-based bidding setup decays. The values were set from a snapshot of closed business, the mix shifts, and by the third quarter the bidder is buying against a price list nobody has checked. This retainer exists to check it. Every month the realised value of each conversion action is recomputed from closed CRM outcomes and compared with the value that was uploaded when the conversion happened. Either it holds, and we say so in two lines, or it has moved, and we restate it before the next flight. The monthly cycle: Closed outcomes are pulled from the CRM and matched back to the conversion records uploaded three to nine months earlier, depending on the sales cycle. Realised value per conversion action is computed by segment — channel, campaign type, product line, region — and set against the assumed value. Where the gap is material we restate the model and push new values through the pipeline. Where it is not, we leave it alone. A value model that changes every month is noise, and Smart Bidding responds badly to a moving target. Guarding the loop: Feeding outcomes back to a bidder that selected the traffic creates a loop that can narrow itself: the model buys what converted, that becomes the only data, and the value model then confirms the model's own preference. We hold it open with periodic geo or campaign-level holdouts, and by watching whether the realised-value distribution is tightening for reasons the business cannot explain. This part of the work needs somebody willing to argue with a good-looking result. Operations: The pipeline runs through your warehouse and a reverse ETL layer where one exists — Hightouch, Census, or a scheduled job against the platform APIs where it does not. Alerts fire on match rate below its established floor, on upload failures, on CRM schema changes, and on a value distribution moving faster than the business plausibly did. When an administrator renames a stage picklist on a Thursday afternoon, you hear it from us before it reaches bidding. Reporting: One written note each month to whoever owns the number: what changed in the value model, what it will do to spend, what we chose not to change and why. Two pages, no dashboard tour. A quarterly session with finance and the channel lead to re-agree the value definitions themselves. What we do not do: We do not manage campaigns, set targets, write creative or sit between you and your media agency. We build and govern the signal; the people spending the money decide what to do with it, and keeping that line clean is why our figures stay usable in a disagreement. We do not run alongside a second team pushing values into the same accounts. And we will end a retainer where CRM stage definitions keep changing without notice, because the model cannot be governed on shifting ground and continuing to bill for it would be dishonest. Who this is not for: Paid programmes under roughly fifty thousand euro a month, or with fewer than a few hundred closed outcomes a quarter. Restatement needs enough closed business to be a measurement rather than an anecdote. Below that this is a quarterly consulting conversation instead of a monthly retainer, and you should ask us for that. It is also wrong where nobody internally can say what a customer is worth. That question has to be answerable before it can be governed.