Using customer history in AI account targeting

Two sales teams using similar AI tools can find the same earnings calls, hiring announcements and company news. They may end up pursuing many of the same accounts for the same reasons. A team's own customer history could help it choose differently. It might know why a previous sale fell through or what stopped a customer from using the product. That information can help a seller decide whom to call and what to discuss, provided the team has kept it up to date.

Suppose an agent recommends the largest accounts, and the sales manager assigns the strongest sellers to them. If those accounts convert, the agent's choices may look good. But the extra attention could explain some of the result. Feeding those wins back into the agent without recording who worked the accounts and what they did could lead it to keep recommending the same kinds of customers.

Keeping the original recommendation alongside a record of the sales team's work would help an analyst investigate. Did highly ranked accounts get more experienced sellers or help from specialists? How much attention did other accounts receive? The records could show that most of the specialists' time went to the agent's preferred accounts. The team would then have a reason to test how much of the result came from that extra help.

Pursuing some accounts the agent didn't recommend could help the team find opportunities it would otherwise miss. Where practical, randomly assigning comparable accounts different amounts of sales attention would provide better evidence about what that attention changes. Enterprise sales makes this difficult. There may be few comparable deals, they can take months to close, and managers have quarterly targets to meet. A small trial could suggest changes to the account list without providing enough evidence to put a reliable number on the agent's contribution.

What the seller knows about the account

An earnings call might announce an expansion into Europe. Your team might also know that the customer's previous rollout stalled because regional administrators couldn't configure permissions, and that a new operations leader has funded a fix. A seller could use that history to start a discussion about implementation. Without it, they might send the same expansion-themed message as everyone else. An old CRM note could also be wrong, so the seller needs to see what the agent relied on and be able to correct it.

If AI reduces the time sellers spend on public research, some of that time could go into learning why customers bought, why they stopped using the product or why a deal failed. Customer success and product teams may know things that contradict the explanation in the CRM. Those findings need to reach the people deciding which accounts to pursue. If all the saved time goes into sending more messages, the team may do little to improve the information behind its recommendations.

A customer may look promising even though the implementation team has no room to take on the work this quarter. The agent should flag that constraint before recommending the account. A new segment may be worth pursuing despite a slower sales cycle because the company wants to learn how to serve it. While waiting for those deals to close, the sales manager could review why prospects declined and what they asked for, then use those findings to change which accounts the team pursues.