The problem
Four account executives at Northwind SaaS were each keeping their own Sales Navigator export, and a fifth list lived in a CRM dump nobody fully trusted. Combined, the spreadsheets held roughly 1,900 rows - the same VP of Engineering appeared on three of them under three slightly different job titles, and nobody had a reliable way to tell.
Prioritisation was whatever felt urgent that morning. A senior title near the top of somebody's sheet got a message; a better-fit prospect sitting unseen in someone else's file did not. There was no shared definition of a good prospect, so four people were quietly running four different qualification bars.
What they configured
The four lists went in as a single bulk import. Duplicate detection matched on LinkedIn URL and on name plus company, folding the overlapping rows down to roughly 1,300 distinct prospects before anyone touched them by hand.
Northwind's Ideal Customer Profile put decision authority and technology relevance above the platform defaults - the signals that correlated with a fast yes in their own pipeline - while leaving the qualification threshold at the standard 70%. Round-robin distribution split the queue evenly across the four owners, with a daily limit of 8 prospects each, 32 in total, so no rep's list ever grew faster than they could actually work it.
What happened
68% of the de-duplicated list never cleared the threshold and never reached a rep's queue - which meant the 32 people who did were, on paper, the best-fit prospects Northwind had. Each queued prospect arrived with an AI-drafted opener built from Northwind's own case studies and the prospect's profile, ready for the owner to review and send rather than draft from a blank page.
Reply classification did the triage that used to happen in a shared inbox: a positive reply created a sales task for the owner immediately, a negative reply closed the prospect, and a neutral one returned to the pool for the next cycle instead of quietly going cold. Within six weeks the team was holding roughly 3.4 times as many qualified conversations a week as the four-spreadsheet process had produced.