A client last month showed us their LinkedIn spend: $58K a month, and they were ready to pull the plug on the channel. We've seen versions of this conversation maybe forty times in the past year and a half. Usually the channel isn't actually broken. Usually the account is just being run the way you'd run Meta, which does not work.
LinkedIn is expensive. That part is real. CPMs in the $30+ range are normal now, sometimes $50 for senior-level targeting. What that means is you have less margin for sloppy execution than on any other platform. Here's what we see go wrong, roughly in order of how often we see it.
The audience is too small
This one is counterintuitive, and I remember being wrong about it for a long time myself. A 6,000-person audience of "perfect-fit" titles at 12 target accounts sounds like precision targeting. It's not. LinkedIn's algorithm needs a pool big enough to learn from, and when you hand it 6,000 people it just shows ads to whoever is easiest to reach — which tends to be people scrolling a lot, which tends to mean interns.
Our rule of thumb is 100K+ audience size for cold prospecting. You narrow through creative and offer, not through layered filters. If you have strict ICP constraints, you use an inclusion list at the company level and leave seniority/function broad.
The exception is ABM plays where you're running to a named account list under 500 companies. That's a different motion entirely — treat those separately, don't mix them into your prospecting campaigns.
The creative is a repurposed webinar
I can spot a repurposed YouTube clip on LinkedIn from four feet away. So can everyone else. 16:9, an intro card, corporate music — the feed trains people to scroll past that in about a second.
What works right now, in our testing: 9:16 or 1:1 video, shot vertically, captions baked in, someone (ideally a founder or subject-matter expert) talking to camera like it's a FaceTime call. Production value is actively bad for performance here. We have a client whose highest-performing ad in 2026 is their CEO filmed in what is clearly a hotel room in Berlin. It's been running for three months.
The offer is a demo for cold traffic
You don't walk up to strangers at a conference and ask them to book a 45-minute call with your sales team. That's what "Book a demo" ads to cold traffic are.
Better sequencing, in general terms:
- Cold — give something useful. A benchmark report, a teardown, a calculator. No gating, or gating just with an email.
- Warm (7–30 day retargeting) — a middle step. A webinar, a product tour, an assessment.
- Hot (people who've engaged multiple times) — now you can ask for the demo.
You can compress this if you have strong brand recognition. Most B2B brands don't, and they try to compress it anyway.
You're trusting LinkedIn's own attribution
LinkedIn is, understandably, generous to itself when crediting conversions. Every platform is. The in-platform dashboard will tell you a story that usually overstates real impact by a decent margin.
We don't have a magic fix for this. What we do: tag every lead in the CRM with first-touch and last-touch UTMs, look at multi-touch paths weekly, and run the occasional pause test on specific campaigns to see whether revenue actually drops when we turn them off. Sometimes it doesn't, which is its own useful information.
Half the "high-ROAS" LinkedIn campaigns we've pause-tested didn't meaningfully move pipeline when we turned them off. The other half turned out to be the most incremental spend in the account.
So is LinkedIn worth it?
For most B2B with a $30K+ ACV, yes — if you run it like LinkedIn and not like Meta. For everything else, it's probably not your first channel. We've told plenty of clients to turn it off entirely and put the budget into Google or content. That's fine too.
If you're in the middle of burning spend right now: start with the audience size. That one change alone usually pulls CPA down 20–30% inside two weeks, and it costs nothing.