Most people who conclude that cold outreach doesn't work reached that conclusion after about twenty messages. Twenty is not enough to tell you anything — but the belief sticks, the activity stops, and the pipeline goes back to being referrals and hope.
If a realistic reply rate is somewhere around one in ten, then twenty messages might produce two replies or might produce none, purely by chance. Zero out of twenty feels like proof. It isn't; it's noise.
You need somewhere around a hundred messages before the number means much, and they need to be comparable — same kind of prospect, same approach. A hundred sounds like a lot until you realise it's two weeks of sending ten a day, which is about forty minutes of work daily.
Six numbers in a spreadsheet. The value isn't the totals, it's seeing which step loses the most people — because that's the only part worth changing.
These vary enormously by niche and by how targeted the list is, so treat them as orientation rather than targets. For well-targeted, individually written outreach to local businesses, something in this region is normal:
Generic mass sending behaves completely differently and much worse — reply rates well under one per cent are common, which is why volume approaches need such enormous volume.
The key idea: don't judge outreach by whether you got work. Judge it by which step in the funnel is leaking, because every leak has a different cause and a different fix.
Almost no replies at all. The problem is upstream of the writing. Either the list is wrong — you're contacting businesses with nothing to fix — or the opening sentence isn't specific enough to be distinguishable from spam. Fix the list first; it's the more common cause and the easier one to verify.
Replies, but no conversations. You're losing the handoff. Usually slow responses, or asking for too much too soon: a calendar link and a 45-minute "discovery call" when a ten-minute chat would have happened.
Conversations, but no proposals. You aren't qualifying. Either they were never going to buy and you didn't find out, or you failed to establish budget and scope well enough to write anything.
Proposals, but no wins. Price appeared as a surprise, scope didn't match what they asked for, or you were writing to someone who couldn't decide.
The instinct when results are poor is to rewrite everything — new subject line, new opener, new niche, new offer. Then something improves and you've no idea which change did it, so you can't do it again.
Run one variable across a block of fifty, keep everything else identical, and only act on differences large enough to be real. Small differences across small samples are usually chance wearing a convincing costume.
Track how long outreach takes you. Forty messages a week at fifteen minutes each is ten hours — if that produces one £2,500 job a month, it's good work; if it produces one a quarter, you're better off elsewhere.
Most of that time disappears into research rather than writing: opening each site, testing it on a phone, deciding whether there's anything worth mentioning. That's the number to attack, because it's the one that makes the whole activity feel unaffordable.
This is the practical argument for scoring prospects before you write to them. A LeadCrawlr scan of a trade and area returns every business with its website scored 1–10 and the specific failures listed, which collapses the research step and removes the businesses that were never worth contacting.
The effect on the numbers is at both ends: fewer wasted messages, and a higher reply rate on the ones you do send, because every message has something true in it.
Outreach is a slow-feedback activity. Messages sent this week produce conversations next week and signed jobs the month after, so judging it weekly guarantees you'll quit during a normal quiet patch. Set a number — a hundred messages, properly recorded — and make the decision when you get there, not before.
LeadCrawlr scans Google Maps, scores every business website 1–10, and flags the ones that need you most. $49/year, unlimited scans.
Get LeadCrawlr — $49 / year