The card’s claims had never been tested on anyone
A sales team at a media technology business sold against several competitors, and carried a battlecard for each of them. The claims on those cards had been assembled from a competitor’s website, a demo watched once, and the recollection of whoever lost the last deal. Sellers used a card for an opening line and put it down at the first pushback, because nothing behind the claim would hold.
A seller who leads with a claim and cannot support it has done worse than say nothing. The buyer now knows the seller does not understand the alternative. Rebuilding the cards meant asking people who had used both products: buyers who evaluated the pair, and staff who had worked on the other side of it.
We tested every claim before it reached a seller
We ran AI moderated calls at volume with buyers who had evaluated both products, customers who had switched away, and former staff from the competitors themselves. Participants were screened on direct exposure to both products rather than on how strong their opinion was, so every account came from somebody who had seen the two side by side.
The claims already on the cards went into those calls word for word, and came back tested. The output said which claims held, which needed narrowing, and which were better set aside. Volume is what made this a battlecard rather than an anecdote. A claim that holds across many separate conversations is one a seller can put in front of a buyer and stay standing behind.
Fewer claims, and sellers who would actually use them
The rebuilt card carried less than the original, and each claim arrived with the evidence beneath it. Sellers stopped putting it down at the first challenge, because there was somewhere to go next. Several claims came back narrowed rather than confirmed, which turned out to be the useful part.
A precise claim a seller can hold beats a broad one they have to retreat from. The team also kept the method, so the card can be refreshed when the competitors change. Competitive positioning goes stale quietly, and a repeatable process is worth more than a single rebuild.
