every mistake I made year 1 of cold outbound
putting this down now while the scars are still fresh because i know in 6 months ill convince myself it wasnt that bad and make the same mistakes again at whatever company i end up at next
i joined as the first ops hire at a series A company about 3 years ago. background was marketing ops, mostly HubSpot and Marketo stuff, some Salesforce admin work. the CEO wanted to build an outbound motion from scratch and i figured cold email infrastructure couldnt be that different from marketing automation. that assumption cost us roughly 4 months and somewhere around $14k in wasted spend before things started actually working
the first thing that broke was the most embarrassing one. i set up 8 google workspace accounts, pointed them all at the same domain, loaded them into Smartlead, and started sending the next day. no warmup. i didnt even know warmup was a thing because in marketing ops you send from established domains with years of history and dedicated IPs and all that. within 11 days every single one of those inboxes was hitting spam on gmail. the SDR team was furious because they had been promised pipeline by end of month and i had basically torched our primary domain's reputation. we had to buy a new set of domains, wait 2 weeks for them to age, then do a proper 3 week warmup cycle. so thats 5 weeks of zero outbound because i didnt spend 20 minutes reading about email deliverability before jumping in. the SDR lead and i didnt talk much during that stretch lol
the bigger issue though was data flow. or more accurately the complete absence of any data flow. i had the SDRs pulling contacts manually from LinkedIn Sales Navigator, pasting them into a google sheet, then someone on my team would enrich the emails through Prospeo for enrichment and run verification through ZeroBounce, then paste the verified contacts back into a different sheet, then the SDRs would manually import those into Smartlead. this process took anywhere from 4 to 8 hours per batch depending on list size. the SDRs hated it. they kept skipping the verification step because it was too slow, which meant bounce rates were creeping up to 6-7% on some campaigns, which meant more deliverability problems. i was so focused on getting the tools working individually that i completely missed how the handoffs between them were where everything fell apart. took me about 2 months to realize the problem wasnt any single tool, it was that nothing talked to anything else. we eventually rebuilt the whole pipeline in Clay, pulling from Sales Navigator exports, enriching, verifying through ZeroBounce via API, scoring, and pushing directly into Smartlead through a webhook. that single change probably saved 15-20 hours a week across the team. but those first 2 months of manual copy paste workflows were brutal and the friction between ops and the SDR team got really bad because they saw us as a bottleneck (which... we were)
what really killed us though was the reporting gap. for the first 4 months i had no unified view of anything. Smartlead had send and reply data. HubSpot had deal data. the google sheets had list quality data. nothing connected. the VP of sales would ask me simple questions like "whats our cost per meeting from outbound" and id have to spend 2 hours manually pulling numbers from 3 different systems to give him a number that was probably wrong anyway. we were spending about $2,800/mo across all tools and inboxes (Smartlead at $94/mo, ZeroBounce maybe $40/mo, Clay at $149/mo, plus 12 inboxes through Maildoso at around $4 each, plus Sales Navigator licenses for 3 SDRs, plus ZoomInfo which was like $1,200/mo and honestly i still dont know if we got enough value from that contract). but i couldnt tell you if that $2,800 was generating $28k or $280k in pipeline because the data lived in 5 different places. we eventually built a reporting layer by pushing everything into HubSpot via Clay workflows and some custom API work, tagging each contact with campaign source and enrichment metadata so we could trace a meeting all the way back to which list and which sequence generated it. once we had that visibility we realized that 3 of our 7 active campaigns were producing basically nothing, like sub 0.3% reply rates, and we were able to kill them and reallocate those sends to the sequences that were actually converting at 2.8-3.4% reply rates
oh and one more thing i forgot to mention. inbox management. when we scaled from 8 inboxes to 20 i tried to manage the DNS records and warmup schedules manually in a spreadsheet. missed a DKIM record on 3 of the new domains. didnt catch it for almost 2 weeks. those 3 domains were basically sending unauthenticated email the whole time which tanked their sender scores before they even got through warmup. if youre managing more than like 5 inboxes you need something that handles the DNS setup for you or at minimum you need a checklist that you actually follow every time. i was too proud to use a checklist for something i thought was simple. it wasnt simple
the meta lesson from all of this is that cold email infrastructure is a system not a collection of tools. every tool decision is actually an integration decision. every integration decision is actually a data flow decision. and if you dont think about the data flow first youll end up rebuilding everything 3 months in which is exactly what happened to us
anyway thats the damage report. we're in a much better place now but getting here was way more expensive and painful than it needed to be