Every few months someone declares cold email dead. Then someone else posts a screenshot of a campaign booking twenty meetings a month and declares it alive.
Both are right, which is how you know it is the wrong question.
Some markets meet them and some do not, and the ones that do not will burn six months and a pile of money finding out. We run this screen before we take on outbound work, and we kill a lot of good companies with it. Not bad businesses. Bad fits for this specific channel.
Here are the three tests, in the order we run them. A company has to pass all three. Failing any one is a stop.
Test one: can we build you a list?
That is the whole question. Not “is there demand.” Not “is the market big in dollars.” Can someone sit down and put together a few thousand companies that actually need what you sell.
There are two ways to answer yes, and most people only know about one of them.
The usual way: firmographics
Filter on industry, headcount, and geography. That gives you the list. Then you layer signals on top to work out who is warm.
This works fine when your buyer looks like a normal company profile. Vertical software is the easy case. Point of sale for breweries. Compliance software for dispensaries. Management software for summer camps. The buyer count is high, the filters are clean, and you can pull most of the market in an afternoon.
The trap is confusing a big market with a listable one. A company selling specialized equipment into forty refineries has a huge market by revenue and a terrible one for outbound. Forty accounts is a relationship business. You do not need a sequencer, you need a person on a plane.
Narrow industry with thousands of buyers is a bullseye. Broad industry with forty buyers is not.
The other way: let the signal build the list
Here is the part most people miss. Sometimes the thing that makes someone a buyer is visible from the outside. When that is true, you should not start with firmographics at all.
I spent last week on a build for a company that sells security patching for end-of-life open source software. Their buyer is defined by one thing: which specific library version is running in production.
“Financial services, 500 to 5,000 employees” cannot see that. Most of those companies run modern stacks. There is nothing to sell them.
So we flipped the order. We searched page source for the literal version string, and whoever came back became the account list. The signal generated the list instead of ranking one we already had. Firmographics came in last, as a multiplier on the score rather than a filter at the front.
If you go this route, verify the signal live before you send. Technographic databases are crawl snapshots. When we checked every top account by hand, one in eight had already changed. The highest-scoring account on the whole list had migrated off the old version months earlier. Stale data is worse than no data, because it looks like data.
So test one passes if either answer is yes. You can build the list from firmographics, or the buying trigger is something you can detect and the signal can build it for you.
Where it fails is when neither is true. No clean way to filter, and nothing observable to search for. That is a real stop.
Test two: can a generalist write credible copy?
This is the one that kills the most companies, and almost nobody screens for it.
Cold email only works if the first two lines prove you understand the reader’s problem. That means someone has to write copy that a specialist will not immediately dismiss. For most B2B products, a good writer can get there with a few hours of research.
For some, they cannot. If your buyer is a protein biologist, or a structural analysis engineer, or a bioanalytical chemist, the gap between “researched this for an afternoon” and “actually knows this field” is visible in the first sentence. Your prospect reads it and files you as noise. Not because the product is wrong, but because the message could not survive contact with an expert.
When we screened a few hundred companies against these tests recently, copy-writability was the single largest reason for a no. Ahead of company size, ahead of geography, ahead of everything. It is the test almost nobody runs and the one that kills the most deals.
This is not permanent. If you have a subject matter expert who will write with us, or a founder who will spend real time on messaging, the test flips to a pass. But it has to be someone’s actual job. “We will get you the technical details” is not the same as a person who will sit down and write.
Test three: can you attribute a return inside six months?
If your sales cycle runs eighteen months, outbound might still be the right investment. It will not be a provable one in a timeframe that survives contact with a budget review.
Month four arrives, there is no closed revenue, and someone reasonable asks what the spend has produced. Nobody defends a channel through two more quarters of silence, so the program gets cut right before the pipeline it built starts closing.
Capital equipment with a twelve month procurement cycle fails this. Enterprise deals that need three committees and a security review fail it. Both can absolutely be sold, just not through a channel that has to prove itself quarterly.
If this is you and you still want outbound, the fix is to agree on a leading indicator before you start. Qualified conversations, not closed revenue. Then hold the line on it when month four arrives.
What to do when you fail one
Failing a test is information, not a verdict on the business.
Fail test one and your motion is probably targeted account work, not volume. Fewer accounts, more research per account, a real person doing the outreach. That can work extremely well. It is just not the same machine.
Fail test two and the fix is a person, not a tool. Find the internal expert who will own messaging. If nobody will, believe that and spend the money somewhere else.
Fail test three and you need to fix your measurement before you fix your pipeline. Decide what a good month looks like when no deal has closed yet, write it down, and get whoever controls the budget to agree in advance.
The part that matters
The reason to run this screen is not to be selective. It is that outbound is expensive to start and cheap to keep running once it works. Domains, warmup, list building, and copy all land in the first six weeks. If the market cannot support the channel, you pay all of that and learn nothing you could not have known on day one.
Most of the failed outbound programs we see did not fail on execution. They failed on a market that was never going to answer, run by people who assumed the problem was their subject lines.
Run the tests first. They take an afternoon.
Outbound is expensive to start and cheap to keep running once it works. Nearly every failed program we see failed on a market that was never going to answer, not on execution.
