Most in-house transitions fail the same way. The contract ends on the 31st, the new setup starts on the 1st, and pipeline falls off a cliff for two months while domains warm.
It is avoidable. The transition is not a date, it is an overlap, and the overlap needs to start roughly 90 days before the contract ends rather than after it.
Here is the sequence.
Day 90 to 75: find out what you own
Before planning anything, establish what transfers.
Domains. Whose registrar account? If yours, you are in good shape. If theirs, start the transfer conversation now, while you are still a paying client. Transfers take five to seven days once initiated and need an auth code from the losing registrar.
Mailboxes. Are they exportable over SMTP and IMAP, or locked inside a platform? If locked, the conversation history is at risk and you need to plan an export.
The list. Ask for the full lead database with enrichment, suppression list and do-not-contact list. The suppression list matters more than people realise; without it you will re-contact people who already asked not to be.
The learning. Sequences, subject line performance, which segments responded, which objections recurred. This is the most valuable thing and the most commonly lost. Ask for it in a document, not a dashboard screenshot.
If the answer on domains is that they belong to the agency and will not transfer, the plan below still works. You just budget a full warmup cycle rather than inheriting one.
Day 75 to 60: build the parallel estate
Register or claim the new domains and start warming them while the agency is still sending. This is the whole trick and everything else depends on it.
Work out how many you need first. Three mailboxes per domain is the sensible ratio, five the hard ceiling. A programme sending 30,000 a month needs roughly 20 to 30 mailboxes depending on per-mailbox volume, so eight to ten domains.
Configure SPF, DKIM, DMARC and MX on each, verify them, and start warmup immediately. Warmup takes three to four weeks and cannot be compressed, which is why this step sits at day 75 and not day 20.
If you are inheriting warmed domains from the agency, you still want a few new ones warming in parallel. Estates should grow rather than be replaced wholesale.
Day 60 to 45: decide who does the work
The uncomfortable question, and the one that decides whether this succeeds.
Outbound in-house needs somebody who owns it. Not somebody who will fit it around three other responsibilities. Roughly speaking, running a functioning programme is 15 to 25 hours a week: list building, sequence writing, deliverability monitoring and reply handling, with reply handling growing as volume does.
Three viable answers. A dedicated hire, if volume justifies it. An existing team member with their other work genuinely reduced, not nominally. Or a smaller programme run properly by someone with partial capacity, which beats a large programme run badly.
What does not work is assigning it to a founder who is already at capacity. It gets deprioritised in week three and the domains sit warm and unused.
Day 45 to 30: rebuild the sequences
Do not copy the agency’s sequences verbatim. Two reasons: they were written for their sending patterns and list, and you want to understand why each element is there rather than inherit it as cargo.
Take the performance data, identify what worked, and rewrite. Keep the structure that produced results and put it in your own voice. Your team will handle the replies, so the copy should sound like the people who answer.
Build one sequence properly rather than five roughly. You can expand once something is working.
Day 30 to 15: run in parallel
Start sending on the new estate at low volume while the agency continues at full volume. This is the point of the whole plan.
Begin at perhaps 20% of target volume on the new domains. Watch bounce rate, complaint rate and placement. Fix what breaks while the agency is still carrying your pipeline. Compare reply rates against theirs on comparable segments, which tells you whether the gap is infrastructure or messaging.
Expect the new estate to underperform at first. It is warming and your team is learning. That is precisely why the overlap exists.
Day 15 to 0: shift the volume
Ramp the new estate toward target as the agency winds down. Do not go from 20% to 100% in a day; follow a normal volume ramp, increasing roughly 30% every few days while watching metrics.
Take the final handover: list, suppression list, sequences, performance data, domain transfers completed, mailbox exports done. Check the suppression list is actually loaded into your platform before the first full-volume send.
The first 90 days after
Be realistic about the dip. Even done well, expect 60% to 80% of the agency’s output for the first two months, recovering through month three. Teams that plan for 100% on day one make panicked decisions in week two, usually by increasing volume on domains that are not ready.
What tends to go wrong:
Reply handling gets underestimated. Agencies absorb this invisibly. At a few hundred sends a day it is a real daily commitment, and it is the part that converts.
Deliverability stops being watched. The agency had someone monitoring it. In-house, nobody owns it until something breaks. Set up placement testing on a schedule from day one.
The list stops being refreshed. The agency was continuously sourcing. In-house, the initial list runs down around month two and nobody noticed it needed topping up.
Nobody iterates the copy. The sequence that worked in month one decays. Agencies iterate because it is their job; in-house it is the first thing to slip.
The short version
Start 90 days out. Establish what you own, warm the new estate in parallel while the agency still sends, decide honestly who owns the work, rebuild the sequences from their data rather than copying them, and run both for a fortnight before switching.
Plan for 60% to 80% output for two months. Assign deliverability monitoring, list refresh and copy iteration to a named person, because those three are what quietly degrade once nobody is billing you for them.
And get the suppression list. It is the least glamorous item on the handover and the one that will embarrass you if it is missing.
Next: agency versus in-house, priced, or volume ramp schedules.
Frequently asked questions
How long does it take to bring cold email in-house?
Plan 90 days, with the new domain estate warming in parallel from around day 75 while the agency still sends. Attempting it as a clean switchover on the contract end date produces a two-month pipeline gap.
What should I ask my agency for when leaving?
Domain transfers, mailbox exports over SMTP and IMAP, the full lead list with enrichment, the suppression and do-not-contact lists, the sequences, and segment-level performance data.
Will my results drop when I bring outbound in-house?
Usually yes, by 20% to 40% for the first two months, recovering through month three. The drop is smaller if you overlap the transition rather than switching cleanly.
How many hours a week does in-house outbound take?
Around 15 to 25 for a functioning programme, weighted toward reply handling as volume grows. It needs a named owner rather than being distributed across people with other priorities.