Sixty-person services firm. The founder shows me his ad spend chart. Four years ago, paid Google ads delivered qualified leads at three hundred pounds each. Two years ago, two thousand. Today, two and a half thousand and falling in conversion. He’s added LinkedIn ads. He’s added a content team. He’s added an SDR for outbound email. Each new lever has moved the needle less than the last. He’s spending more, working harder, and his cost per customer has gone up every quarter for eighteen months.
He thinks the issue is execution. The channels themselves have shifted under him, and the shift is structural rather than cyclical.
Why are the channels actually breaking?
The cost of acquiring customers through digital channels has roughly doubled in the last two years across most service categories. The cause isn’t a temporary advertising market shift. It’s structural. Search is being answered by AI summaries before the click. Content marketing has been flooded by AI-generated volume. Email is being summarised by AI before the human reads it. Paid ads have priced in the new scarcity. Each shift is one-way.
Search first. Google’s AI Overviews now answer many search queries in the result page itself, with no click through. The historical assumption underlying organic content marketing, that ranking for a query brought traffic, is no longer reliable. Organic search visibility has been shrinking for eighteen months.
Content second. Seventy-five percent of content professionals report that AI has increased the volume they produce. The result is a content market saturated with competently-written, search-optimised, indistinguishable articles. The cost of standing out has gone up. The cost of producing has come down.
Email third. Gmail and Yahoo are now summarising messages before the human reads them. Cold outreach that depended on the recipient seeing the subject line and the first sentence increasingly competes with an AI-written summary that may or may not surface the message.
Paid ads fourth. The MSP industry data is the cleanest illustration. Average paid ad cost for a marketing-qualified lead now sits over twenty-four hundred dollars on Google and Bing. At a one-in-three close rate, that’s seven thousand to acquire a customer paying around two thousand monthly recurring. For most service businesses outside enterprise deal sizes, the unit economics on paid have stopped working.
What’s now relatively undervalued?
While the market has piled into paid ads and AI-generated content, several channels have been quietly abandoned and are now structurally undervalued. Cold calling now converts at around two percent. Cold email at half a percent. Cold calling, the channel everyone has spent the last decade calling dead, is four times more efficient than the channel everyone treated as the future. The cause is scarcity: almost nobody is doing it well any more.
The framing that matters is relative value. When the market crowds into one channel, the cost of acquisition rises until the unit economics break. The channels the market has abandoned hold their conversion rates, sometimes for years.
Cold calling is the clearest example. The two percent conversion figure isn’t new. What’s new is that almost nobody is doing it. The prospect on the other end of the call has been getting almost no calls for a year. Their attention is genuinely available, in a way it isn’t for the eight emails and three LinkedIn messages that arrived that morning.
Direct partnerships are the second undervalued channel. Two or three carefully chosen non-competing firms who serve the same client base, treated as a real channel with named owners and tracked metrics, produces better economics than equivalent spend on paid ads. Most firms treat partnerships as something that “just happens”, which is why most firms get nothing from them.
Founder-led narrow content is the third. Specific point-of-view writing on a narrow problem area, where the audience is small but the people who care really care. The economics work because the cost of producing is low relative to the conversion rate of the people who do find it.
Specific community presence is the fourth. Two or three places where your buyers actually are, treated as a multi-year commitment. Specific subreddits, specific Slack groups, specific industry events where the same people show up year on year.
What four moves actually work right now?
The four moves that work for service businesses right now are best treated as a portfolio. Narrow founder-led content for the inbound flywheel. A small disciplined human-led outbound effort for immediate pipeline. Direct partnerships for the recurring referral channel. Specific community presence for the long compounding awareness lever. None of these alone replaces what paid ads used to do.
The discipline matters more than the choice. Each of these channels works only when run consistently for six months or more. Most firms try one of them for a quarter, see modest results, conclude it doesn’t work, and reach for the next thing. The pattern that produces results across all four is unglamorous: pick the version that fits your business, name an owner, give it a fixed weekly time commitment, and run it without changes long enough to know whether it pays back.
The AI question lands here. AI is not the answer to a CAC problem caused largely by AI saturation. Using AI to produce more inbound content into a saturated content market makes the saturation worse, including your own contribution to it. The AI uses that do help are sharper targeting, faster qualification, and writing better human outreach. AI as the augmentation layer.
The firms quietly winning right now are running the four-channel portfolio with discipline and using AI as augmentation. They look regressive on the surface, because they’re calling people and showing up at events and writing two-page positions on narrow problems. The maths is what’s modern.
What does this mean for pricing and ICP?
If acquisition is structurally harder and more expensive across the board, deal sizes have to rise to compensate. That implication is the part most founders defer because it’s the most uncomfortable. Moving up-market. Narrowing the ideal client profile. Saying no to bad-fit smaller clients more often. The unit economics of an expensive-to-acquire customer only work when the customer pays back over a long enough relationship at a high enough margin.
The maths is unforgiving. If your fully-loaded CAC has gone from three thousand to seven thousand, and your average customer paid you fifteen thousand a year before, the payback period has shifted from a quarter to half a year. If your gross margin on delivery is fifty percent, you’ve moved from CAC payback inside the first project to needing the customer through to year two. That changes who you should be targeting.
Most services firms respond by pushing harder on the smaller deals because they’re easier to win. The maths runs the wrong way. Smaller deals are now more expensive to acquire as a proportion of the deal value, and they pay back over a shorter relationship.
Narrowing the ICP feels risky during a tight market. The intuition is that you can’t afford to say no. The maths says you can’t afford not to. The clients who pay back the new CAC are larger, more specific, more expensive to acquire individually but cheaper to retain, and worth refusing the rest of the market to keep finding.
The four-channel portfolio reinforces this. Cold calling, partnerships, narrow content, and community presence all favour deeper relationships with fewer prospects. The market shape these channels naturally produce is the market shape the new economics require.
The playbook most services firms have been running has structurally broken. The firms quietly winning are the ones running channels the market has abandoned. The work is uncomfortable. Cold calling. Two-page positions on narrow problems. Two or three real partnerships. Showing up where your buyers actually are. The maths is on its side.
If your cost per customer has been rising for the last six quarters and the channels you’re adding aren’t paying back, book a conversation.
