The single LinkedIn setting silently tripling your ad costs — and it's not your budget, your bid, or your creative.

SIGNAL · Issue 01 — Paid Social · LinkedIn ABM · Bid Strategy

SIGNAL is Ziggy’s field-notes series on the experiments we run and what the data tells us. This issue has been anonymised at the client’s request; every percentage and ratio is reported exactly as measured.

Every account-based programme carries a tension that rarely shows up in a kick-off deck: the tighter you target, the more you pay to be seen. For one of our enterprise travel-technology clients, that tension stopped being theoretical in the spring of 2026. Across two flagship audiences — a broad agency-prospecting pool and a deliberately narrow airline pool — the cost to reach a thousand people (CPM) didn’t just drift upward. It compounded.

By late May, the airline audience was costing close to $1,478 CPM in its worst week — more than four times where it had started in early March. The agency audience had climbed from roughly $49 to a $314 peak over the same window. Spend was holding flat or rising while the number of people we actually reached was shrinking week after week. On one campaign LinkedIn audited for us, cost had risen from about $1,000 across its first five weeks to $12,564 in a single recent week, for a fraction of the delivery.

This is the anatomy of that problem — and the test that reversed it in a fortnight.

The setup: a small audience is a feature — until the auction turns it against you

Account-based marketing on LinkedIn lives or dies on precision. For this client we were targeting named accounts and tightly-bounded job functions, which is exactly the point: speak only to the people who can buy. One airline campaign LinkedIn inspected on our behalf contained roughly 430 addressable members. That is a deliberately small universe.

The trouble is what a small universe does inside a real-time auction. Fewer members means fewer ad-eligible sessions each day, and the same handful of high-value professionals are being chased by every other advertiser who has bought that profile. Demand stays constant; supply is tiny. Price goes up.

Weekly CPM by audience. Both climb through April and May, peaking the week of 25 May (agency $314, airline $1,478). The dashed line marks the bid-strategy switch on 5 June — within a week both collapse to near their March starting points.

The diagnosis: it wasn’t the audience. It was how we were bidding into it.

We took the worst-performing campaign to LinkedIn’s technical support team to pressure-test our own read of the data. Their conclusion matched ours, and it pointed at the optimisation goal rather than the targeting. The campaigns were running on Maximum Delivery bidding, optimised toward Reach — that is, toward touching as many unique members as possible.

In a large audience, optimising for unique reach is sensible. In an audience of a few hundred, it becomes a trap. Once the system has shown the ad to most of the available members, it has to bid harder and harder to find the next new face — and Maximum Delivery will keep raising bids automatically to chase that goal. The result is a feedback loop that prices itself into the stratosphere.

The compounding loop, step by step

1. Tiny audience, scarce inventory. ~430 members means very few daily ad sessions to bid on, with many advertisers competing for the same people.
2. Reach-based goal punishes saturation. Optimising for unique members forces the system to keep finding new faces — increasingly expensive as the pool is used up.
3. Max Delivery auto-raises bids. To keep winning scarce impressions, the system lifts bids on its own. CPM climbs with no ceiling.
4. A low daily budget throttles delivery. A ~$70/day cap meant pacing logic restricted serving, concentrating spend on the priciest impressions.
5. Middling creative relevance compounds it. Lower relevance scores lose auctions, nudging the required bid higher still.

Impressions told the story plainly. One campaign that had served around 510 impressions a week in mid-March was delivering under 80 a week by late May. We were paying a steeper and steeper premium for a thinner and thinner slice of delivery.

Widening the audience was off the table — precision was the whole strategy. So the only lever left was to stop paying for unique reach we could no longer afford.

The hypothesis: trade a sliver of incremental reach for control of the price

The fix required changing what the campaign was optimising for. Maximum Delivery doesn’t allow a cost ceiling, so the first move was to switch the campaign optimisation goal from Reach to Impressions. That single change unlocks Cost Cap bidding — where you set a target average cost and the system holds bids near it, rather than chasing unique members at any price.

We launched the switch on Friday 5 June, paired with frequency caps to keep the same person from being over-served now that the goal had moved from unique reach toward impressions. The trade-off we accepted, knowingly: against a set account list, optimising for impressions lifts frequency rather than net-new reach. For ABM, where the job is repeated, relevant exposure to a fixed list of accounts, that is an acceptable — arguably preferable — exchange.

The result: more delivery, far less money

The effect was immediate and large. Comparing the fortnight before the switch (22 May–4 June) with the fortnight after (5–18 June), on matched windows:

Across the two audiences, total impressions rose +97% (72,142 → 142,063) while spend fell −24% ($29,761 → $22,591). The programme stopped overpaying for scarce unique reach and started buying volume at a controlled price.

The guardrail: did cheaper impressions mean a worse audience?

The obvious risk with a cost cap is that you buy down-market — winning cheap impressions by reaching more junior, less relevant people. So we checked the seniority mix before declaring victory. We saw a small drop at the most senior tier, which redistributed into Manager, Director and VP — squarely the client’s decision-making sweet spot — with no leakage into junior or entry-level seniorities.

More telling: click-through rate rose among the audiences we reached. The most plausible reading is that we were no longer re-serving the same saturated handful of people; we were reaching net-new members who hadn’t yet seen the ad and were more likely to engage. Lower cost and higher engagement, from the same targeting.

The transferable lessonIn a small, tightly-targeted audience, optimising for unique reach is a CPM trap. The algorithm will spend without limit to find the next new person — long after the economics have stopped making sense.

The moment an ABM audience is narrow enough that saturation is likely, the right instinct is to optimise for impressions under a cost cap, accept marginally higher frequency, and let the price come back to earth.

What we’re watching next

• Frequency discipline. Average frequency is holding below 3, but we’re reviewing it at ad-set level in case any pocket needs a tighter cap.
• 30-day penetration. A full month of data will show the real impact on offline account penetration — the metric that actually matters for ABM.
• Re-investing the savings. With CPMs controlled, headroom opens to lift CPM targets deliberately where they expand reach into genuinely net-new accounts, and to test off-platform formats (CTV, audience networks) for audiences that are simply expensive to reach on-platform.

Methodology. Figures are drawn from the client’s live paid-media data warehouse (LinkedIn Ads delivery), covering 2 March–21 June 2026. CPM is calculated as total cost ÷ impressions × 1,000. “Before” and “after” comparisons use matched 14-day windows (22 May–4 June vs. 5–18 June) to control for the day the bid strategy changed (5 June). Audiences are aggregated from the campaign-level agency and airline prospecting segments. Client identity and absolute account-level figures have been anonymised; all percentage changes and ratios are reported as measured.

This is how we run paid media — diagnosis first, data always.

Ziggy is the B2B demand generation agency behind some of the world’s largest technology companies. If your CPMs are climbing and no one can tell you why, that’s a conversation we like having. Schedule a call →