Two partners, one difference
Take two partners of similar size, in similar markets, serving similar customers. One buys across six of your product lines. The other buys across two.
That difference is not a puzzle and it is not luck. It is a list — of specific things the second partner is either buying somewhere else or not selling at all. A partner's basket over time is the clearest statement they will ever make about their business, clearer than anything they will say in a QBR. Read it properly and whitespace stops being a guess about opportunity and becomes an inventory of it.
Why the campaign always underperforms
The same offer goes to the whole partner base, which makes it wrong for most of them at once: redundant for the ones already buying that line, unsupported for the ones with no capability to sell it. Response is poor, everybody concludes partners are not interested, and the exercise is quietly retired.
The offer was never the problem. Sending one offer to everybody was.
Reading the basket instead of blasting the base
Recommendations come from what comparable partners actually buy, and from what a given partner's own portfolio implies they should be selling. Each one carries its reasoning, so an account manager can decide whether it holds before repeating it to a customer.
It also reads how sophisticated a seller each partner is, because that changes the answer. Some partners move boxes and some sell solutions. The same recommendation to both is wasted on one of them.
The uncomfortable part
A recommendation is a hypothesis about somebody else's business, and some of them will be wrong. They arrive with their reasoning attached so your team can throw out the ones that do not survive contact with a partner who knows their market better than any model does.