How to choose an outbound provider

Every outbound vendor's website promises the same things, and a website is easy to copy. The work behind it is not. Here is how to tell vendors apart before you pay, and how to test them after.

Updated 2026-10-30 · Markdown

Read the work, not the website

Start with the cold emails the vendor sent you. A vendor sells itself harder than it will ever sell anything else; if that is its best work, your product will sound the same. Then ask to read complete work: the research, the emails and the targeting for a real client. We publish three complete playbooks for exactly this reason, so you can put them next to anyone else's work in an evening.

The usual questions check surprisingly little. "Cases like ours" and "how much money did you make your clients" often can't be answered honestly: every company is different, and clients' revenue stays under NDA for years. A lack of references proves nothing either, because busy clients have no time for strangers' calls and a strong vendor won't burden them. "What is your conversion rate?" Into what, and counted how? It also depends on your own team. And the honest answer to "how many deals in three months?" is "nobody knows in advance", which sounds weaker than a confident invention.

Ask what you can check

Who does the work, and how. At an agency, talk to the people who will run your campaigns, not only to the person selling them.
How many inboxes run one client's campaigns, on which IP addresses, and how they rotate.
Where the contacts come from and how each address is verified.
How the vendor knows which folder an email landed in. "Sent" is not delivered, and delivered is not the inbox.
What the price consists of, line by line. If an offer is very cheap, the first question is what it leaves out.

Then put the offer next to what your last leads cost you. A buyer who doesn't care how a vendor gets results lets the vendor fool them.

Measure what matters

The right unit is the cost of a targeted reply from the right person, then of a conversation and a deal. The price of an email or a seat says nothing about results.
Opens don't count. Mail apps such as Apple Mail load messages in the background, so an "open" doesn't mean anyone read anything. Count events that really happened: sent (accepted by the receiving mail server), delivered, replied. Our emails carry no tracking pixels at all.
Not every reply is a lead. Auto-replies, refusals and polite "not now"s don't belong in the result. We count a targeted reply: the right person showed real interest in your offer.
Count conversion one way: from emails sent to targeted replies from the right people.

Traps in comparing numbers

Different denominators. Some vendors count replies per email sent, some per contact, some per opener, so the same market can show a fraction of a percent in an independent study and several percent in a sending tool's report on its own users. Compare only numbers counted the same way.
Different meanings of "lead". For one vendor a lead is an email address, for another any reply, for a third an agreed call. Agree on the word before comparing percentages.
A subscription is not the work. A tool's price and the price of complete work can't be compared until you add the people, data and infrastructure the tool needs.
Someone else's percentages. Another company's conversion doesn't predict yours: offers and markets differ, and so does the sender. A company with famous investors and a contract to boast about gets different replies from an unknown one.
Leftovers instead of the market. Companies often give an outside vendor the experiments nobody inside wants and keep the core market for their own team. Test any channel on your main market.

Beware promises nobody can keep

Nobody can honestly promise a number of deals: results depend on your offer, your market and how you handle the conversations. A vendor promising "50 guaranteed meetings a month" is usually selling meetings with anyone. Pay-per-meeting deals, often $200–700 per booked meeting according to Smartlead's 2026 pricing guide, reward exactly that: the calendar fills, your account executive takes call after call, and nothing closes. Cheap replies turn into expensive lost working days.

Revenue share sounds fair and isn't. It turns the vendor into the worst investor on the planet: it puts in money and work, gets no equity or control, and can hear "we'll take it from here" at any time. We had revenue shares in our agency contracts for years; by our count our leads produced more than a hundred deals, and the share was paid 13 times. The evidence sits with whoever pays: the conversations run from their inboxes, and most people forget the contract terms within months.

An honest vendor guarantees what it fully controls. For us that is deep research, strong emails, honest infrastructure with a real warmup, sending of every paid email without bursts or duplicates, every reply read, and statuses that tell the truth.

Run a fair test

If reading the work isn't enough, buy the smallest plans of two vendors and compare them in action:

One task. The same offer, the same markets, the same restrictions.
Full cost. Licenses, inboxes, domains, data, and your people's time for setup and management, not just the subscription.
The right unit. The cost of a targeted reply, a conversation and a deal.
About three months. New infrastructure needs about 21 days of warmup, then sending ramps up, and some people answer weeks later, after a follow-up. The first two weeks of sending say almost nothing about demand.
The content of conversations. Compare who replied and what you can talk about with them, not just how many replied.
No artificial equalizing. A vendor's interview, research and doctrine are part of what you pay for; stripping them out for "equal preparation" means testing a different product.

When you add up the result, credit each deal to the month the lead appeared: a deal closed in month seven belongs to the month the conversation started. Only then do you see what a lead really brings and what the channel costs.

Frequently asked questions

What should I ask a cold email agency before hiring it?

Who will do your work, and can you talk to them; how many inboxes will run your campaigns and on which IP addresses; where the contacts come from and how they are verified; how the agency knows which folder an email lands in; and what the price consists of. Also read the cold emails the agency sent you.

Are open rates a good way to compare vendors?

No. Mail apps load emails in the background, so opens are inflated and say nothing about reading. Compare targeted replies from the right people per email sent, then conversations and deals.

Is pay-per-meeting lead generation worth it?

It pays the vendor for booking meetings, not for meetings with people who can buy, so calendars fill with calls that close nothing. Pay for work you can check, and measure the cost of a useful conversation instead.

How long should a test of an outbound provider run?

About three months: roughly 21 days of warmup, then the sending ramp, then the slower replies that come after follow-ups. A verdict after the first week or two is noise.