When to fire your lead generation agency, and when to wait

Most agency relationships end badly for the same reason they started badly: nobody wrote down what month three was supposed to look like. So when month three arrives and the calendar is thin, both sides argue from feelings. The question worth answering first is narrower than "are they any good": is this a slow ramp or a broken engine, because those need opposite responses.

What does a normal outbound ramp look like month by month?

Outbound has a fixed physical schedule that no amount of urgency compresses. Knowing it stops you from firing a working engagement in week six.

• Month one, first half. Secondary domains registered, DNS and authentication configured, mailboxes created and warming, ICP defined against evidence, first list built, first sequence written and approved. Warm-up alone takes two to three weeks. Anyone sending full volume from a fresh domain in week one is not being fast, they are burning your reputation.
• Month one, second half. First sends at deliberately low volume. Not a performance test. A deliverability test.
• Month two. The first real numbers: delivered rate, bounce rate, reply rate, positive reply share. Enough replies to judge whether the message lands. First meetings if your sales cycle is short.
• Month three. The first fair judgement point. Delivery should be clean, replies should be arriving somewhere inside the 1 to 5 percent band typical of B2B cold email, positive replies should exist, and meetings should be on the calendar unless your deal size and buying committee genuinely justify longer.
• Months four to six. Compounding. The best-performing segment gets more volume, weak angles get cut, follow-up depth gets tuned. This is where a good engagement stops feeling like an experiment.

An agency that promised meetings in week two either sold you a number it cannot control or is sending from shared infrastructure you will never own. Both are reasons to worry in month one, not month three.

Slow ramp or broken engine: what is the difference?

A slow ramp is a working system producing weak output. A broken engine is a system that is not running at all, where every downstream number is meaningless. Telling them apart takes an hour of looking at the right data.

• Broken engine: emails are not landing. Bounce rate above 5 percent, missing authentication, spam placement, domains flagged. Nothing after delivery can be judged until this is fixed.
• Slow ramp: emails land, replies arrive, but the positive share is thin. This is a targeting and copy problem, and it is exactly what iteration exists to solve.
• Broken engine: the list has no defensible logic. Nobody can tell you why these companies, at this size, in this market.
• Slow ramp: the right people are being reached with the wrong angle. Replies say "wrong timing" or "we handle this internally" rather than "why are you writing to me".
• Broken engine: the reporting cannot be verified. You are shown opens and impressions instead of delivered, replies, positive replies and meetings.
• Slow ramp: meetings happen and do not convert. That is frequently a close problem, not a lead problem.

Measure against your own previous months rather than a published average. A campaign moving from 0.8 percent to 2.1 percent reply rate is working, even if it has produced two meetings so far.

The diagnostic questions to ask before you fire anyone

Send these in one email and give a deadline of two working days. The speed and shape of the answers tell you more than the answers themselves.

1. What are the delivered and bounce rates, per sending domain, for the last 30 days?
2. Show me the last 200 emails exactly as they were sent, with merge fields resolved.
3. What were the list criteria, and where did the data come from?
4. What changed between month one and month two, and what did that change do to the positive reply rate?
5. Of all replies, how many were positive, neutral, negative, and opt-outs?
6. How quickly was each positive reply answered, and by whom?
7. What specifically are you changing next, and what do you expect it to move?

An agency running a real campaign answers all seven inside a day, because the answers are its working file. An agency that needs a week and returns a slide of impressions is not measuring the campaign at all, which is a worse problem than an unlucky campaign. You cannot fix what nobody is watching.

When to fire your lead generation agency

These signals do not improve with patience. Each one describes something structural rather than a bad month.

• The sending domains and mailboxes are not in your name. Your reputation, your asset. If it sits in the agency's account, you are renting the one thing that compounds.
• They refuse to show the copy or the list. There is no legitimate reason to hide what is being sent under your brand to your market.
• Reporting is vanity only. Opens, impressions and connection counts instead of delivered, replies, positive replies, meetings booked and shows.
• Bounce rates stay high after you raise them. Verification is cheap and fast. Ignoring it after a warning is a competence signal.
• Two documented rounds of iteration changed nothing. Not two months of the same sequence. Two deliberate changes with a stated hypothesis and a measured result.
• Opt-outs are not honoured, or nobody can state the legal basis for the data. That is your regulatory exposure, not theirs.
• They guaranteed a meeting count and are now filling it with anyone. Guaranteed numbers create junk meetings by design, which is why we never promise a fixed count.
• Nobody can name the next change. "We are optimising" is not a plan.

Any of the first three is enough on its own. The rest are cumulative: two of them together is a conversation, four is an exit.

When to wait instead

Several failures that look like agency failure are not. Firing at week six resets the clock to zero, and you pay for domain warm-up a second time.

• Warm-up is not finished. Volume is still deliberately low. Judging reply counts here is judging arithmetic, not performance.
• You were slow to approve. Late ICP sign-off, copy sitting unread for ten days and no calendar access are the most common causes of a weak month one, and they are all on the client side.
• Your sales cycle is longer than the engagement. If deals normally take five months, a three month engagement cannot show closed revenue. It can only show meetings and pipeline.
• Replies are healthy but calls are handled badly. Outbound delivered a conversation and the conversation was lost. Different problem, different fix.
• Seasonality. August across most of Europe and late December are structurally quiet. A flat month there proves nothing.
• You changed the ICP mid-campaign. Every change restarts the learning cycle. If you did it twice, the campaign has no clean data yet.
• The market is genuinely small and you were told so. In the Baltics and other small markets the entire addressable list can be a few hundred accounts. Volume cannot rescue that, and an agency that warned you honestly is not the problem.

Is it the agency, the offer, or the market?

Outbound amplifies an offer. It cannot invent one. Before switching providers, read what the replies actually say, because the wording sorts the problem for you.

Replies saying "wrong person" or "we do not do that" point at targeting, which a competent agency fixes inside a month. Replies saying "interesting, but not now" point at timing and trigger selection, which is a list-signal fix. Replies saying "we already have this and it works" point at the offer, and no agency on the continent will fix that with a better subject line. Silence with clean delivery points at the message. Changing agencies only helps with the first and the last. If you switch while the real problem is the offer, you will buy the same disappointment at a new invoice number.

How to exit cleanly and keep your domains and data

Do this before you announce anything. Once notice is given, cooperation gets thinner and calendars get busier.

• Confirm domain ownership. The registrar account should be in your company name with you holding the login. If the agency registered the domains, initiate the transfer before the conversation, not after.
• Take DNS control. SPF, DKIM and DMARC records belong on domains you administer. Losing DNS means losing the authentication history that took weeks to build.
• Export the send log. Every contact, every send date, every step reached. Without it you cannot tell your next provider who has already been contacted.
• Export the suppression and opt-out list. This is the single most valuable file in the account. Losing it means re-contacting people who asked you to stop, which is a compliance failure rather than an inconvenience.
• Export reply threads and positive replies. Interested prospects who said "not this quarter" are pipeline. They should live in your CRM, not in an inbox you are about to lose access to.
• Get the list criteria and data sources in writing. Your next provider rebuilds in days instead of weeks with it.
• Get the sequence copy and test history. Even a campaign that underperformed tells the next team what has already been ruled out.
• Decide what happens to the mailboxes. If you are pausing outbound entirely, wind the domains down gradually rather than abandoning them. Warmed domains left idle for months lose most of their standing.
• Check the notice period and what runs during it. A campaign left on autopilot through a 60 day notice period can damage the domains you are trying to keep.

Contract length is the variable that decides how expensive a mistake can get. On a month to month arrangement your worst case is one wasted month. On a twelve month lock-in with a guaranteed meeting count, your worst case is a year of junk meetings you are contractually obliged to accept. We price flat and short for that reason: EUR 3,750 for the first month covering setup and launch, then EUR 2,850 per month, cancel anytime. The client who can leave easily is the client you have to keep earning.

Before you sign the next one

Write down what month three looks like before money moves. Agree the reporting fields, the iteration cadence, and who owns the domains from day one. Ask the questions in the selection checklist during the sales call rather than during the autopsy, and read the full agency guide for how the engagement models differ in what they reward. Most bad engagements were predictable at signature, and the predictable part was that nobody defined what working would look like.