Bring outbound in-house when the agency's playbook is proven. That means three things: it has booked sales calls with the right buyers for several months in a row, some of those calls turned into signed deals or real proposals, and you can explain in two sentences who gets contacted and why they reply. A contract renewal date is a bad trigger. So is "we have been doing this for six months."
Once the motion is proven, move in this order. Collect the assets from the agency first: ICP and disqualification rules, sequences with their numbers, domains and mailboxes, the CRM with every reply, and the objection notes.
Then hire one person who can run an existing system. After that, run a 4-week overlap where the agency keeps sending while your hire takes over replies, then one segment, then everything.
This piece assumes you have already made the build or buy call and bought first. The question now is how to take it back without losing the pipeline.
Smirnov Consulting Group is a Prague-based B2B outbound lead generation agency that runs cold email and LinkedIn campaigns for founder-led B2B companies and books qualified sales calls. The agency you are leaving may well be one like mine, and the collection checklist further down is the list to bring to your exit conversation with it.
What does in-house cost compared with the agency you have now?
Most agencies do not publish prices, and the public ranges are wide. Clutch's buyer guide gives a lead generation agency a range of $1,000 to $25,000 monthly (page updated March 2025) and shows no method behind it. Your own invoice is the better number.
On the in-house side, Payscale lists an average US base salary of $43,709 a year for a sales development representative, with total pay between $36k and $77k once commission and bonus are in (324 salary profiles, updated September 2026).
Salary is the visible part. The data, the tools, the domains and your own management hours sit on top.
| Your agency today | Your in-house hire | |
|---|---|---|
| What you pay for | A monthly fee that covers people, tools and data | Salary and commission, plus tools, data, domains and your time |
| Published range | $1,000 to $25,000 monthly per Clutch, with no method stated | $43,709 average US base, $36k to $77k total pay (Payscale, September 2026) |
| Time to full output | Already running | About 3 months of ramp for a new SDR |
| Who holds the know-how | The agency, unless you collect it | You, for as long as the person stays |
The ramp figure is from a sales development consultancy that surveyed 351 B2B companies for its 2025 report and found an average SDR ramp of 3.0 months. I went through that survey and the turnover numbers around it in SDR ramp and tenure benchmarks.
Look at the last row. The whole handover is about moving that row from the left column to the right one before the agency leaves.
How do you know the playbook is ready to move?
The number of months since you started tells you nothing. I would check these five signals instead:
- The calls repeat. Sales calls with ICP decision-makers have come in month after month from the same one or two segments, not from one lucky burst.
- Some calls became money. At least a few signed deals, or proposals at your real price that are still moving. Booked calls alone only prove the top of the funnel.
- You know what works. The agency can show which sequence, which subject line and which angle produced the replies, with numbers per variant.
- You know the volume behind it. How many people were contacted per month to get those calls. Your hire needs this number as a target from day 1.
- You have heard the objections yourself. You have been on enough of the calls to know what prospects push back on and what answer moves them forward.
If one of these is missing, what you would be handing over is still an experiment, and a new hire is a bad owner for an experiment.
When your deals are simply slow
Some founders read a few months without a signed deal as failure. My short version: "People don't make decisions fast in every space. Some pipelines are just getting started."
If your sales cycle runs 6 months, 4 months of calls may not have closed a single deal yet, and the motion can still be working. In that case signal 2 means proposals that are alive, and the move waits until you see a few of them close.
The renewal date is the other false trigger. Moving because the contract is up forces a rushed handover. Know your notice period well ahead, so the date never decides for you. I compared typical terms and exit costs in this piece on agency notice periods and exit fees.
What should you collect from the agency before your hire starts?
Collect everything before the new person's first day. If the hire starts first, their first month goes into reconstructing what the agency already knows.
| What to collect | What it should contain | Why the hire needs it |
|---|---|---|
| ICP and disqualification rules | Titles, company size, industries, countries, plus who gets excluded and why | So the first list they build matches the one that worked |
| List sources and contact history | The filters and searches behind each list, every company and person already contacted, the opt-out list | So nobody gets contacted twice and nobody who said no hears from you again |
| Sequences with numbers | Every step and its spacing, the winning subject lines and angles, reply rate per variant | So they keep the winners and stop testing what was already tested |
| Sending infrastructure | Every domain and mailbox, logins, SPF, DKIM and DMARC status, warmup history, current daily volume per mailbox | So sending continues on warm accounts instead of new cold ones |
| CRM export with full reply history | Every reply, including "not now", referrals and objections, with dates and owners | So follow-ups and old conversations keep going |
| Objection and call notes | The objections that come up, the answers that worked, the outcome of each booked call | So the hire does not learn it all again on live prospects |
| Monthly baseline | People contacted, total and positive replies, calls booked and calls held, month by month | So you can compare the hire's numbers with the agency's fairly |
Two rows need extra care, because they are the hardest to rebuild later.
The sending infrastructure
My standard is 10-20 domains carrying 20-40 email accounts. Every domain gets SPF, DKIM and DMARC set up first, the authentication records that tell inboxes your mail is legitimate. Every account then warms up for 10-14 days at minimum, and only after that does the volume go up. I walked through it in the domains and inboxes post.
If those accounts transfer to you, you inherit the warmup and the sending history. If they stay with the agency, you start from zero.
Check who is the registrant of every domain and who holds admin rights on every mailbox, well before the overlap starts. I covered the registrar and admin side in detail in who owns the domains an agency buys.
Then check their health. Google's sender guidelines set a ceiling for anyone mailing Gmail users: a Postmaster Tools spam rate that stays below 0.3%, ideally below 0.1% (checked September 2026). Bulk senders, which Google defines by a 5,000-a-day threshold to personal Gmail accounts, also need SPF, DKIM and DMARC all in place.
A domain sitting near those limits is a liability, even if it is yours.
The reply history
Which CRM you pick matters less than what goes into it. A CRM is there so no lead gets lost, and at handover that means the agency's reply inbox ends up inside it.
A prospect who said "ask me in Q2" is a lead your hire should call in Q2. If that reply lives only in an agency mailbox, it is gone. More on what that CRM should hold in setting up a CRM before outbound.
How should the overlap month run?
Four weeks is enough to move ownership. It is too short to finish the ramp, so plan for months 2 and 3 to run below the agency's numbers.
| Week | The agency owns | Your hire owns | You own |
|---|---|---|---|
| Week 1 | All sending, all replies, all booking | Reading every reply thread, the ICP document and the sequence numbers; sitting in on the weekly review | Introductions, access, and the targets for month 1 |
| Week 2 | All sending; approving the hire's reply drafts | Drafting replies on every thread; logging each one in the CRM | Reading 10 of the hire's replies and correcting tone |
| Week 3 | Sending on every segment except one | One segment end to end: list, sending, replies, booking | Comparing that segment's numbers with the agency's segments |
| Week 4 | Final export, credential handover, removing its own access | All segments on the inherited mailboxes | Changing every password, ticking off the checklist table above |
Two rules hold the month together. Nobody sends from the same mailbox or LinkedIn account as someone else in the same week. And the agency's access comes off only after the export has been opened and checked by you.
If the pipeline cannot afford the dip, keep the agency on one slice after week 4, for example one country or one segment, until your hire is at full output. A hybrid like that works only while the border between the two sides stays clear.
Who should your first in-house hire be?
Someone who can run a proven system well. Hire for execution, because the building part is already done and paid for.
| Inheriting a working system | Starting from nothing | |
|---|---|---|
| Who to hire | An SDR who has run sequences and handled replies at volume before | A senior builder who has set up targeting, copy and infrastructure from zero |
| First 30 days | Run what exists, answer every reply fast, keep the CRM clean | Research the ICP, write the copy, set up the domains |
| Green flag in the interview | Asks for your numbers per sequence and per segment | Has a view on which market to test first, and why |
| Red flag in the interview | Plans to rewrite everything in week 1 | Has only ever run someone else's playbook |
A builder hired into a working system is tempted to rebuild it, because that is the job they know. An executor hired into nothing has nothing to execute. Match the hire to where you actually are.
When the team grows past one
Split the work so every lead has exactly one owner. In practice, give each rep a fixed set of segments or countries, give every mailbox and every LinkedIn account a single owner, and put an owner field on every company in the CRM. If the agency keeps a slice after the overlap, treat it as one more rep in that split.
What does inheriting a system look like week to week?
Mostly repetitive work, which is what you want. The hire refreshes lists inside the same ICP, keeps volume steady, answers replies the same day and logs everything.
Growth comes from width. Horizontal scaling means growing by adding accounts: more LinkedIn profiles and more mailboxes running a sequence that already works. When you want more calls, add mailboxes and sender accounts to the proven sequence, each warmed up before it carries volume, and keep the copy steady.
Changes to the playbook come later, once 2 or 3 months of in-house numbers match the agency baseline. Then test one thing at a time: one subject line, one angle, one new segment.
What can go wrong in the handover?
- The renewal date drives the move. The handover gets squeezed into the last 2 weeks of the contract, and the export arrives half done.
- The opt-out list gets lost. Your new sender contacts people who already asked the agency to stop, and the complaints land on your domains.
- The playbook freezes. My blunt version for hired SDRs and BDRs: they end up "running strategies that worked 3 years ago and deliver nothing." Both agencies and in-house reps can land there. Inherited copy still needs a monthly look at the numbers.
- The agency is cut off on day 1. The pipeline gap then equals the whole ramp.
Questions founders ask about taking outbound back
What if the agency will not hand over the domains?
Buy new ones at least 2 weeks before the overlap ends and set up SPF, DKIM and DMARC on day 1, so the new accounts finish warmup before they carry volume. Then push hard for the contact history, opt-out list and reply history. Those matter more than the domains.
Can the agency and my own SDR prospect in parallel?
It works when their territories never overlap. Give each side its own segments or countries, its own mailboxes and accounts, and keep one shared CRM with an owner on every company. Skip that and the same prospect hears from two senders, which makes both look careless.
How do I know the playbook works before I hire?
Run the five signals above. A quick test: try to write the playbook on one page from what the agency gives you. Who gets contacted, with which sequence, at what volume, and what they object to. If the page has gaps, keep collecting before you hire.