sharing this because its more useful than me posting another "heres our reply rates" screenshot that makes it look like we know what were doing. we didnt. for about 3 months we were actively destroying our own infrastructure and i had to sit in a room with our VP of sales and explain why 14 of our sending domains were basically dead.
some numbers on the damage before i get into what happened:
12 google workspace domains burned. 2 microsoft domains burned. total cost of those domains plus workspace/365 subscriptions was around $2,800 over the life of them. bounce rate across our campaigns hit 11.4% in the worst week. reply rates dropped from 2.8% down to 0.6%. we had to pause ALL outbound for 9 days while we unfucked everything which meant zero pipeline contribution for almost two weeks, and our CRO was already skeptical that cold email was worth the headcount (thats a whole other story). basically i handed him ammunition to argue we should just dump the budget into linkedin ads.
the first thing that broke was volume. we went from doing around 8k emails a month across the team, which was working fine, to trying to push 50k in a single month because leadership wanted to "test scale" before board prep. my two SDRs and i were sending from 14 domains with about 3 inboxes each, so 42 inboxes total. the math seemed fine on paper, roughly 40 emails per inbox per day. but we ramped from 25/day to 40/day in like 4 days instead of gradually increasing over 2-3 weeks. i knew better than this honestly, id read enough posts in here about warmup schedules, but the pressure from above was real and i convinced myself it would be fine because 40/day isnt even that aggressive. turns out the ramp speed matters almost as much as the ceiling. google started throttling us within a week and by day 12 we had inboxes landing in spam at probably 70%+ rates based on what we could see in Instantly analytics.
the bigger issue though was list quality and this is where i really messed up. to hit 50k we couldnt be picky about our lists. normally we build pretty targeted lists, HR directors and VPs of people ops at companies between 200-2000 employees, and we enrich through Prospeo which has been solid for us, email accuracy around 82-85% and then verify with NeverBounce before anything goes into a sequence. that workflow gets our bounce rate down to like 1.3-1.8% consistently. but for the 50k push we basically grabbed every HR-adjacent title we could find on LinkedIn Sales Nav, dumped them into a spreadsheet, ran a portion through Prospeo and a portion through Hunter because we were trying to go fast and figured two tools in parallel would speed things up. and then here's the part that makes me cringe... we skipped verification on about 15k of those contacts because NeverBounce was taking too long on the bulk job and we had a deadline. just sent them raw. the bounce rate on that unverified segment was 9.2%. on one domain it hit 14.7% which is basically a death sentence.
what really killed us was that we didnt catch it fast enough. we were monitoring reply rates but not watching bounces in real time. Instantly shows you bounce data but we werent checking it daily during the push, we were just looking at the reply numbers trying to hit meeting targets. by the time i pulled the reports and saw what was happening we'd been sending from burned domains for 6+ days. those domains were cooked. no amount of warmup was bringing them back, i tried for 3 weeks with Instantly warmup cranked up and the deliverability never recovered past maybe 40% inbox placement.
so what changed. first we killed all 14 domains and started fresh. bought new ones through Maildoso which was easier than setting up workspace manually for each one. took about 3 weeks of warmup before we started sending again, during which time i had to justify to our CRO why we were producing zero meetings. that was fun. he literally asked in a pipeline review if we should "just hire another AE and give them the cold email budget" and i had to make the case that the channel works when you dont blow it up, which is hard to argue when you just blew it up.
once we were back online we went back to our old volume, around 8k/month, and i set hard rules. no inbox sends more than 30/day. ramp takes minimum 14 days from 5/day to 30/day. every single list goes through Prospeo for enrichment then NeverBounce for verification, no exceptions, i dont care if it takes an extra day. if a list comes back with more than 3% catch-all or unknown from verification we either scrub those out or run them through Scrubby which is decent at resolving catch-alls over time.
results since the rebuild (been about 4 months now): bounce rate averaging 1.4%. reply rate back up to 2.6% which is close to where we were before. booking around 18-22 meetings a month from cold email which at our ACV of $53k is meaningful pipeline. and honestly the constraint on volume forced us to be way more targeted with our lists which i think is actually why our reply rate recovered so fast.
the political part is still ongoing. our CRO brings up the "domain incident" basically every time cold email comes up in leadership meetings. its like a scar that wont heal. but the numbers are there now and pipeline doesnt lie so we keep going.
anyway i mostly wrote this because i see people in here talking about scaling to 50k 100k emails and i just want to say that the infrastructure has to come first. the domains, the warmup, the verification pipeline, all of it. if you skip steps to hit a number you will pay for it and it takes months to recover not days