Questions to Ask a Lead Generation Agency — before you sign

If you're working out who to hire to run outbound for your SaaS — or which agency to use to book meetings — you've hit the awkward part: choosing an outbound agency is an unusually blind purchase. You can't test the product, the case studies are unverifiable, and the thing being sold — meetings with people who don't know you yet — depends on work you won't see happening. By the time you can judge it properly, you've spent a quarter and possibly damaged a sending domain.

So the shortlist call matters more than the proposal. Below are the questions worth asking, grouped by what they're really testing, with what a good answer sounds like. We're an outbound agency ourselves, so read this knowing we'd pass our own test — but every question here is one we'd want a client to ask us, and the reason is simple: the agencies that lose on these questions are the ones that make the whole category look bad.

First: what you're actually buying

Not a list, and not sending capacity. Both are cheap and neither is scarce. What you're buying is judgement applied repeatedly — who to contact, when, what to say, and what to change when it isn't working. Everything below is a way of finding out whether that judgement exists or whether you're buying volume with a nice dashboard.

Every question here has the same shape: can they answer in specifics about your business, or only in adjectives about theirs?

1 · Targeting and data

This is where most campaigns are won or lost, and where the weakest agencies are vaguest. If you only ask one group of questions, ask these.

• Where does the list come from, and which sources would you combine for our market? A single provider name is a poor answer — no one source covers a market properly. Look for a primary source, a fallback, and a reason.
• How do you verify, and where does verification sit in your process? They should have a clear position on whether it runs before or after enrichment, and why. "Our data provider handles it" means they haven't thought about it.
• What do you do with catch-all domains? A specific technical question with a specific right answer — separate them, send at low volume, watch before scaling. Blank looks here tell you a lot.
• How would you test coverage for our ICP before we commit? The good answer involves pulling a sample in your actual segment and checking it by hand.
• How big will the list be? If the number is enormous, the targeting is loose. Ask what they'd cut it down to and why.

The reasoning behind all five is in B2B data for outbound — worth skimming before the call so you can tell a real answer from a fluent one.

2 · The message

Most agencies will say the copy is human-written. Ask the follow-ups that make that checkable.

• Who actually writes it — can I see three first lines they'd send to my ICP? Ask for them before signing. It takes an hour and it's the single most revealing thing you can request.
• Is any of it generated, and where exactly? There's a real difference between using AI for research and using it to write the email. The first is sensible; the second is why so much outbound reads identically.
• How many message variants run at once, and how do you decide what to change? Look for a testing method, not "we optimise continuously".
• What's the ask in the first email? If the answer is "book a 30-minute call", they haven't thought about how cold buyers behave.

3 · Domains and sending infrastructure

The most expensive mistakes live here, and almost nobody asks about them until something breaks.

• Will you send from our primary domain? The answer should be no. Outbound belongs on separate lookalike domains so a problem can't touch your company mail.
• Who owns and pays for those domains and inboxes? Ask directly. If the agency owns them, you're renting your own sending reputation.
• What happens to them if we stop working together? This is the question that changes contracts. Warmed domains are an asset you paid to build.
• How long is warm-up, and how many emails per inbox per day? They should have real numbers to hand and be visibly cautious about them. See cold email deliverability for what sensible looks like.
• How do you handle unsubscribes and suppression across clients? A serious operation has a shared suppression list. A careless one will email someone who opted out of a different client's campaign.

4 · Who actually does the work

The pitch team is rarely the delivery team. That isn't automatically wrong, but you should know it before rather than after.

• Who will work on this account, by name, and how much of their week do we get? Vagueness here is the most common warning sign in the whole industry.
• Is any of it subcontracted, and to where? Not disqualifying. Undisclosed is.
• How many other clients does that person carry? There is a number above which nobody can do research properly. Ask what theirs is.
• What happens in the first 30 days, before any email goes out? Setup should be weeks, not days — research, list, domains, warm-up, messaging. An agency that can start sending on Monday is skipping something you're paying for.

5 · What happens when it doesn't work

It will, at some point. How they answer this predicts the whole relationship.

• What counts as a lead, in writing? The gap between "responded" and "booked a call with a qualified buyer" is where most disputes start. Get the definition in the contract.
• Month two produces nothing. What do you change first? A real operator answers immediately and in order — usually targeting before copy, copy before channel. Hesitation here means they've never had to.
• What have you stopped doing in the last year, and why? Our favourite question. Anyone who's run campaigns has abandoned something. A blank means either no experience or no honesty.
• Can we speak to two clients in a similar market — including one that didn't renew? The second half of that sentence is the real question.

6 · Contract, pricing and exit

• What's the minimum term, and what does notice look like? Three to six months is reasonable — outbound needs setup plus a couple of months of sending before data means anything. Twelve months with no exit is not.
• What's included and what's billed separately? Data, domains, inboxes and tooling are real costs. Find out whose budget they sit in.
• If we pay per lead, what exactly triggers the fee? Per-lead pricing aligns incentives towards volume and loose qualification. Not disqualifying, but read that definition twice.
• What do we keep when we leave? Lists, sequences, domains, reporting, the suppression list. Agree it while everyone's friendly.

Don't expect a public price list, and be suspicious of one — the real figure depends on your ICP and how hard your buyers are to reach. What you can do is compare the models properly, which is what how much B2B lead generation costs is for.

7 · Red flags

Any one of these deserves a hard follow-up. Two together usually means the model is volume rather than pipeline.

• A guaranteed number of meetings. Nobody can guarantee how strangers respond. What they can guarantee is activity, which is what's actually being sold.
• A price before any question about your ICP. The work can't be scoped before the target is known.
• No named clients. "We've worked with 200 companies" and no names is a course funnel, not a track record.
• Case studies with no numbers. Or numbers with no baseline — "3x more meetings" from what?
• Sending from your primary domain. Either they don't know or they don't care which is worse.
• Reluctance to name who does the work.
• Volume as the headline. "10,000 emails a month" describes cost, not outcome.

8 · What a good agency should be asking you

The most reliable signal isn't in their answers at all. It's whether they interrogate you. An agency that takes a brief without pushing back is planning to send whatever you asked for and invoice you.

Expect to be asked: who closed your last ten deals and what did they have in common; what's your average deal size and sales cycle; who handles replies, and how fast; what happens after a meeting is booked; what you've already tried and why it stopped. If nobody asks who handles the replies, walk — booked meetings that nobody follows up are the most common way outbound "fails" without the agency doing anything wrong.

The takeaway

You can't inspect outbound before you buy it, but you can inspect the thinking behind it — and that turns out to be a good proxy. The questions above all test the same thing: whether specifics exist under the pitch. Where does the data come from, who writes the words, whose domain carries the risk, which named person does the work, what changes when it fails, and what you keep when you leave.

Ask them all. It costs an hour and it's the cheapest diligence you'll ever do. And if an agency finds the list uncomfortable, you've learned what you needed to know before signing rather than in month three.