I’ll call him Jonathan. He runs a sixty-person services firm in the Midlands, and he has been the founder I have spoken to more often than any other this year about the same problem. Four years ago his firm was paying around three hundred pounds for a qualified lead through Google ads. Today the same lead costs closer to two and a half thousand, and the conversion behind it has softened. He has added LinkedIn ads, hired a content lead, and put an SDR on outbound email, and each new lever has moved the needle less than the one before it.
His cost per customer has risen for eighteen quarters straight. He is spending more, working harder, and the unit economics keep getting worse. He thinks the issue is execution. He has been hiring agencies, swapping vendors, rewriting copy, retraining the SDR. None of it has bent the curve. The reason is straightforward, and it is not him. The channels themselves have shifted under his feet, all at the same time.
This post is about what those shifts actually are, what the maths now says, and which channels have quietly become the cheapest routes back to a workable cost of acquisition. Some of the answer is uncomfortable for marketing departments. Most of it is uncomfortable for anyone who has spent the last five years building a digital-first acquisition machine.
Why has paid acquisition stopped working at services-firm deal sizes?
The honest scale comes from the MSP industry’s own published numbers. The average paid ad cost for one marketing-qualified lead is now $2,428.76 across Google and Bing. At a one-in-three close rate, that’s roughly $7,286 to acquire one customer paying $2,200 to $2,500 a month in recurring revenue. Best-in-class operators are spending north of $27,000 fully-burdened to win a single logo. The numbers do not work at services-firm deal sizes.
The driver is auction maths. Paid channels price scarcity, and attention has become genuinely scarce as more firms compete for the same clicks. There is no clever bid strategy that recovers a CAC ratio when the underlying channel is overpriced for your unit economics. Spending more inside a broken channel does not bring the maths back.
For Jonathan’s firm, this is the headline shift. He has been treating cost per lead as a tactical optimisation problem when it has become a structural one. The first move is to stop trying to recover the channel and start asking which channels are now relatively undervalued.
What’s actually breaking, and why isn’t it just expensive?
Three structural shifts are running in parallel, and each is one-way. Search is being answered by AI summaries before the click happens, with eMarketer’s 2026 FAQ on AI saturation and zero-click search tracking the visibility loss for organic. Content marketing has been flooded by AI-generated volume, with 75% of content professionals saying AI has increased what they produce. Email is being summarised by Gmail and Yahoo before the recipient reads the subject line.
These shifts are durable architecture changes in how attention flows, rather than cyclical headwinds that pass. Doing more search-optimised content into a search engine that no longer sends the click is volume against a wall. Sending more cold email into inboxes summarised by AI is volume against a filter. The channels are quietly being disintermediated, which is a different problem from an overheated auction.
This matters because the standard playbook for a CAC problem is to lean harder on the channels you know. When the channels themselves have moved, leaning harder makes the problem worse. Jonathan’s firm is paying more and getting less because the surface he is paying for has shrunk.
Where are the contrarian numbers actually coming from?
GenSales’ 2024 cold calling research puts cold calling conversion at around 2% and cold email at 0.5%. That means a properly-run voice channel is roughly four times more efficient per attempt than the email channel that has eaten most of the outbound budget for the last decade. The reason is not that calls have suddenly become magical. The reason is that almost nobody is doing them well anymore.
When everyone leaves a channel, the people still using it face less competition for attention. The same logic applies to direct partnerships, named-owner referral programmes, and specific community presence. These channels were assumed dead because they did not scale the way paid acquisition did. They are now the cheapest acquisition routes left, partly because the rest of the market piled into the digital channels that have since broken.
The strategic question worth asking is which channels are now relatively undervalued versus where the rest of the market is putting its money. Today the market is piling into AI-generated content and paid acquisition. The relative-value plays are the channels it has abandoned.
What works right now for a services firm?
Four moves quietly carry the weight for the firms I see making the maths work again. The first is narrow founder-led content with a specific point of view, where the audience is small but the people who care really care. Generic content volume is a losing trade against AI-saturated search. A single founder voice with a sharp position is not, because the value sits in the position rather than the production volume.
The second is a small disciplined outbound effort where humans, not tools, do the calling and the writing. The third is partnerships and referrals treated as a real channel with named owners and tracked metrics, not as something that just happens. The fourth is presence in two or three specific communities where your buyers actually are, treated as a long-term commitment rather than a campaign.
None of these scale linearly with spend. All of them compound. They reward firms with a clear point of view and a willingness to stay narrow. They punish firms whose only acquisition muscle is paying for clicks.
How does AI fit when AI is part of the problem?
AI is genuinely causing the channel saturation. Capsule CRM’s 2026 summary of US Chamber data shows generative AI usage among small businesses moving from 23% in 2023 to 58% in 2025, with Goldman Sachs’ 2026 SMB research adding that 93% report positive impact even though only 14% have AI properly embedded. The output is everywhere. The integration is shallow. The result is a flood of AI-written content into channels that AI is also intercepting.
Using more AI to push more content into those channels makes the saturation worse, including yours. The honest AI use is upstream of the channel, not inside it. Better targeting on a small outbound list. Faster qualification of inbound enquiries. Sharper human-written outreach informed by research the AI has done first. AI helps when it makes the human work better, not when it tries to replace it with volume.
For Jonathan’s firm, this implies a different AI strategy than the one his content team is currently running. Less generated copy, more human time on a smaller, better-targeted set of conversations. The honest implication on pricing follows from the same logic. If acquisition is harder and more expensive across the board, deal sizes have to grow. That means moving up-market, narrowing the ICP, and saying no to bad-fit smaller clients more often. Cheap to win means high-margin to deliver, otherwise the unit economics still do not work.
If you are watching the same curve Jonathan is watching, the work ahead is a structural rethink of the channel mix and the ICP underneath it, rather than another agency on top of the existing stack. Book a conversation if you want to do that with someone who has been through the maths.
