There's a particular Tuesday that happens in a lot of founder-led businesses. The quarter is soft, the pipeline is thinner than it should be, and somebody has sent over a proposal for £4,000 a month plus spend. It looks like the decisive thing to do, and everything else on the list looks like admin.
The proposal isn't wrong. The position it occupies on the list is.
Ask a founder what they're doing about a slow quarter and you'll get four or five items: new agency, rebuild the site, try LinkedIn, get someone doing outbound, maybe a chatbot. Every one is defensible. The order is almost always wrong, and the order is worth more than any individual item on the list.
So why aren't the leads converting?
In a business under about £10m it's rarely the ad account, and rarely the leads themselves. Five things break conversion upstream of the campaign, roughly in this order of frequency:
- Nobody owns the first reply. The enquiry lands somewhere three people can see and belongs to none of them. Most common cause, cheapest fix.
- There's no visible proof. The buyer checks you before they take your call, finds six reviews from 2023, and doesn't take it.
- Follow-up stops after two attempts. Most deals in this market need more than that and most sequences have fewer.
- The person on the call has no process. A different conversation every time, and no way to tell a good one from a bad one afterwards.
- The offer doesn't match the traffic. You're paying to reach people who aren't ready to buy the thing you're asking them to buy.
Increasing the budget addresses none of those. It multiplies all of them.
Additive versus compounding
Nobody selling you one of those services will mention this, and it's the part that decides the outcome: those fixes have returns that depend on each other. Run them in the order they occurred to you and you get four separate small improvements over nine months, several of which partly cancel out. Run them in dependency order and each one raises the ceiling on the one after it.
| Fix | Return on its own | What it multiplies |
|---|---|---|
| Work the old database | Cash, within weeks | Nothing, but it funds the rest |
| Collect proof systematically | Small, slow | Every lead from every channel, forever |
| Close the follow-up leaks | Moderate | Every lead from every channel, forever |
| Train the people who close | Moderate | Everything downstream of a booked call |
| Buy new demand | Large, but expensive | Nothing. It's the last multiplicand. |
Read the right-hand column and the sequence writes itself. Traffic multiplies nothing; it's the thing that gets multiplied. Which is why it belongs at the end.
We named our delivery model after Aristotle for that reason, which is a slightly pretentious way of saying the parts only add up if you assemble them in order. The sequence below hasn't changed in two years of running it.
The repair order
1. Database reactivation: money you already own
Before anything new is built or bought, go into the database. Old enquiries, quotes that went quiet, closed-lost deals, customers who lapsed in 2024 and were never contacted again.
You have already paid to acquire every one of those names, at whatever your cost per acquisition was at the time. Most businesses trading longer than three years have somewhere between four hundred and four thousand of them sitting in a CRM that nobody has opened since the person who cared about it left.
The arithmetic is worth doing on the back of an envelope right now. Take the count of contacts who enquired and never bought. Multiply by your average deal value. Multiply by a conversion rate you'd be embarrassed to miss, say 1%. For a business with 1,200 dead enquiries and a £6,000 average deal, that's £72,000 sitting in a spreadsheet nobody opens, available at a media cost of zero.
Done properly, this means a real reason to be in touch rather than a "just checking in" blast: a change in what you offer, a piece of work you've finished, a price that's about to move. Done badly it burns the list, which is the one irreversible mistake in this whole sequence.
There's a second reason it goes first, and it's commercial rather than technical: it produces cash before any new machinery has to be built, and it generates the baseline data that makes everything afterwards measurable. If you don't know what was happening before, no improvement you make later can be proved, and by next spring you'll be arguing with your agency about attribution instead of counting money.
2. Proof
Reviews, referrals, case studies, whatever your particular market checks before it will agree to speak to you. Almost nobody collects this systematically. They wait for it to happen, and then wonder why the competitor with 94 Google reviews gets the call.
It goes second because it's arithmetic rather than sentiment. Every lead you generate from this point forward, whether organic, paid or referred, converts at a better rate because of work you did once. TrustRadius's January 2026 survey of 1,862 buyers found 74% relying on reviews when they choose a supplier, and product demos, free trials and prior experience ranked alongside them. Gartner's finding that B2B buyers spend only around 17% of their total buying time with all suppliers combined says the same thing from the other side: most of the decision happens where you aren't, using material you didn't write.
The mechanism is a process, not a campaign. One named person, one trigger point in delivery, one script, tracked. Ten reviews collected deliberately over eight weeks beats a burst of thirty from a client-wide email that never happens again.
3. The leaks
How long a new enquiry sits before a human replies. What happens to calls nobody answers. Whether the form on the site actually delivers anywhere useful, which is worth testing today, because a surprising number of them don't.
This is where most of the recoverable revenue in a small business hides, because it isn't anybody's job and therefore never gets measured. The 2011 Harvard Business Review audit of 2,241 US companies is still the cleanest public number on it: 23% never responded to a web enquiry at all, and among those that did, the average took 42 hours. Fifteen years and a lot of software later, the businesses we look at have not moved as far from that as anyone would like.
Fix it before you buy traffic, not after, or you are paying for every lead twice: once to the platform, and once again when it goes cold in an inbox three people can see.
4. The people who close
There is no sense pushing more volume at a team that can't convert what it already gets. Most salespeople in businesses under ten staff have never had a day of formal training or a written process, which is not a criticism of them.
Grade three real recorded calls before you change anything. Actual calls, scored against something written down, so that whatever you change afterwards is provably better rather than differently opinionated. Three is enough to find the pattern and few enough that it happens this week.
5. New demand
Ads, outbound, whatever channel you were going to start with in January.
By the time you get here the money goes into a business where follow-up is fast, the proof is visible, the database is warm and somebody competent picks up the phone. Identical spend, different result, and you're buying demand from a position of profit rather than hope.