Context. Our outbound had grown campaign by campaign for months, and nobody had looked at the whole picture. Opinions about what worked were plentiful. Evidence wasn’t.
What I did. I pulled every campaign, over 100 of them, through the Lemlist API and analysed the lot: channel, sequence shape, audience, reply and acceptance rates, and who actually responded. Then I wrote up the findings with prioritised recommendations rather than a data dump.
What happened. Three findings mattered most. LinkedIn invites were accepted at 12 to 16% and out-replied cold email three to six times over, while cold email to enterprise converted at around 1%. Over 4,000 sends had gone to prospects outside the UK who could never buy our levy-funded product, so geography became a hard filter. And the replies that did come in clustered in a few sectors, which now shape the account list. The write-up went in as a prioritised list, high-confidence moves first. The top one, mine the warm database, became a running play: the re-engagement workflow in the next case study is that recommendation, operating.
What I’d reuse. Audit the system, never the anecdote. Individual campaign reviews had been happening all along and missed every one of these findings, because each looked fine in isolation. The waste only shows up in aggregate.