Skip to content
Menu
Home / Article

LinkedIn Ads on a small B2B budget: the audience traps

One company list in an include block turned several hundred companies into about 16 million LinkedIn members. The Campaign Manager traps we hit, and what we fixed.

Our testTwo weeks in Oct to Nov 2026 at tens of dollars a day · relaunched in November at a higher daily budget, results pendingAll our tests ↗
The short answer

On a small LinkedIn Ads budget, the audience settings decide more than the creative. Our worst mistake was a company list in an include block: LinkedIn targets every employee of a listed company, so several hundred companies became about 16 million members.

What goes wrong when a small B2B team runs LinkedIn Ads on a small daily budget, tens of dollars a day, against a named list, and which of those problems belong to the platform rather than to the team? We ran LinkedIn Ads for two weeks across October and November 2026, rebuilt the campaign, and relaunched in November. The first two weeks can't tell you whether LinkedIn Ads work for a small budget, because our conversion tracking was disconnected and most of the problems were setup mistakes we made. They do show where Campaign Manager widens an audience, and those traps apply to anyone.

Several hundred companies, 16 million people

For the relaunch we built a retargeting audience from people who had been on our site: a contact list of 2026 visitors, plus a company list of the several hundred companies seen on the site that year, most of them identified by RB2B and the rest from HubSpot. We put both in the include block of the retargeting ad sets.

In an include block, a company list targets the LinkedIn members who work at a listed company. A few of those companies were large enterprises, and the list alone came to about 16 million members. Retargeting that should have reached a few hundred people was forecast at more than 3.8 million. A second company list on one cold video ad set did the same thing, at 3.9 million members.

The forecast is how we caught it, the day after launch. The pieces of that audience were small: fewer than 300 recent site visitors, a few hundred video viewers, a few hundred target contacts. When the forecast is thousands of times the sum of the named pools, something in the include block is a company list or an expansion setting. We removed the company lists, and retargeting dropped to about 500 people.

Company lists are tempting because they match better. Ours matched at 90% or more, against 75% for our main contact list of more than 10,000 work emails. A high match rate on a company list means LinkedIn found the companies, and then it targets everyone at them. Since we caught it, our rule has been named contact lists only in any include block.

Campaign Manager settings that widen your audience

These are platform behaviours we confirmed in our account in November 2026:

  • "Narrow audience further" adds an OR inside the same block, which widens the audience. The only way we found to narrow it was with exclusions.
  • Device targeting in an include block ORs with your audiences, which adds all desktop users in your location. Use it only as an exclusion.
  • New ad sets defaulted to Audience Expansion on, Audience Network on and a $250 daily budget. Turn the first two off and set the budget on each new ad set.
  • Changing the objective or format on a draft replaces the ad set. The audience carries over; the ads don't, and the budget and expansion settings reset.
  • Daily minimums differ by objective. Brand awareness needed $18 a day per ad set; video views and lead gen accepted $10.
  • A draft with a start date in the past won't launch until you change the date.
  • A contact CSV is rejected if an "@" appears in any column other than email.

Every ad click came from a phone

PostHog and GA4 agreed: every click on our ads came from the LinkedIn app on a phone, and roughly 90% of those visitors left the landing page within seconds. Desktop visitors who reached us from organic LinkedIn posts stayed for several minutes. We tried excluding mobile, and the audience fell below 300, too small for LinkedIn to launch.

So on a small named audience you're buying phone traffic. Either the landing page has to work on a phone, or the ask happens inside LinkedIn with a lead form. The relaunch tests both.

What we set up wrong

These were ours, not LinkedIn's:

  • The campaign ran on the Engagement objective, which optimizes for engagement clicks, when we wanted website visits.
  • Conversion tracking was disconnected, so "0 conversions" in the first two weeks measured nothing.
  • The static test changed only the image, with the same headline and intro on every version, so it couldn't tell us much about copy.
  • The LinkedIn lead form wasn't synced to the CRM.
  • One long intro put one customer's numbers on another customer's ad.

Week 1 cost about $24 a click and about $250 per thousand impressions. Those figures came from a misconfigured campaign on an audience of several thousand people, with every click landing in the first three days, so treat them as a poor benchmark. Audience quality held up: about 60% of impressions went to director level or above.

On creative, a plain static styled like a phone note and a short video that opened on the customer's problem got the most clicks. A longer customer-story video got none, with average watch time around five seconds.

We also ran the account through a contractor's Campaign Manager with no API access for us, so changes were made by hand in the browser. We have since applied for Advertising API access.

The relaunch, and when we'll judge it

In November we relaunched with a handful of ad sets and about 20 ads at a higher daily budget, on the right objectives, with Audience Expansion and Audience Network off. One pair of ad sets tests a LinkedIn lead form against our landing page on the same audience and the same videos. The wider paid retest holds out half the target companies at the company level, so we can compare exposed companies with ones that never saw an ad.

We wrote the decisions down before launch. A reach check comes around mid-December, with the holidays depressing everything for two weeks after that. The first outcome read is at about eight weeks, in mid-January 2027, on target companies starting a conversation (chat, meeting, form or reply), exposed versus holdout. If exposed companies start conversations at twice the holdout rate or better, the budget goes up.

Before you spend on a small budget

  1. Put only named contact lists in include blocks. Check the forecast against the sum of your named pools.
  2. Turn off Audience Expansion and Audience Network on each new or duplicated ad set.
  3. Confirm conversion tracking fires before spending anything.
  4. Make the landing page work on a phone, or use a lead form.
  5. Hold out part of your target list, so the result is a comparison.

See account vs person identification for how the visitor list was built.

Questions buyers ask

Should you upload a company list or a contact list to LinkedIn Ads?

A contact list, at least in an include block. LinkedIn targets every employee of a listed company, so our list of several hundred companies became about 16 million members while our contact lists stayed on target.

Does "Narrow audience further" narrow a LinkedIn audience?

No, it adds an OR inside the same block and widens it. In our account the only way to narrow was with exclusions, and device targeting also works only as an exclusion.

Why are all my LinkedIn ad clicks from mobile?

On a small named audience, most delivery goes to the LinkedIn app. Every one of our ad clicks came from phones and about 90% left within seconds, so the landing page has to work on a phone or the ask should be a lead form.

What is the minimum daily budget on LinkedIn Ads?

It depends on the objective. In November 2026 we saw $18 a day per ad set for brand awareness and $10 for video views and lead gen.

Is LinkedIn Ads worth it on a small B2B budget?

We don't know yet. Our first two weeks ran with conversion tracking disconnected and several setup mistakes, so we relaunched in November 2026 with a holdout group and will judge it in mid-January 2027.

Sources

Sources and evidence behind this page