You set the campaign up carefully. Companies with 51 to 500 employees, director level and up, operations and finance. A lead comes in on Tuesday: a vice president at a logistics firm whose name you half recognize. You open the profile to prepare for the call, and the logistics firm turns out to be one man, working from home, with a company name that sounds a lot like a regional carrier with 400 employees.
It is not bad luck, and it is probably not the only one in your lead list. Since March of this year, LinkedIn has been matching members who never linked their profile to a real company Page to whichever Page their typed-in employer name most resembles, and then targeting them as if they work there. If you target by company size or by a list of named companies, some of the people you are paying for do not fit either. The fastest way to find out how many is not in any report. It is ten minutes spent opening your last ten lead profiles, one at a time.
This piece covers how LinkedIn decides where someone works, what changed, why your reporting cannot show it to you, how to run the ten-profile check, and what a small advertiser can actually do about it.
LinkedIn Has Three Ways to Decide Where Your Lead Works. Two of Them Are Guesses.
Every time you choose a company size or upload a list of target companies, you are trusting LinkedIn to know which company each member works for. Most advertisers assume that knowledge comes straight from the profile. It does, sometimes. LinkedIn's own help page on targeting options describes three different routes, and they are not equally reliable.
The member picked their employer from the list. When someone adds a job and selects the company from LinkedIn's typeahead, the dropdown that suggests Pages as you type, LinkedIn maps them to that Page. You can see this on any profile: the employer shows the company's logo and links through to its Page. This is the solid case. The person told LinkedIn exactly where they work, and the Page carries the company's details.
The member typed a name instead. Plenty of people skip the dropdown and type their employer as plain text. For those members, LinkedIn says it applies a "Deep Learning Model" that weighs alternative company names found on the web, whether the member's industry lines up with the company's, geography, the company's other properties, and how many of the member's first-degree connections are associated with the company. In other words, it makes an educated guess about which Page the typed text refers to. On the profile, these jobs show a generic gray placeholder where the logo would be.
The member left it blank. LinkedIn's help page goes one step further: "If a member doesn't enter a company name, LinkedIn may derive their current company based on their public IP address at ad serving time." Someone on a big company's office network could be counted as that company's employee whether or not they work there.
Company size has its own wrinkle. LinkedIn says size targeting "primarily uses Page admin account input data," meaning whatever range the company's Page admin selected. If the admin never chose one, the size is "determined by the number of member accounts associated with the Page." So a size bracket is only as good as two things: the company's own self-reporting, and LinkedIn's decision about who belongs to that company.
The first route is what you are paying for. The second and third are where the trouble lives.
Since March, LinkedIn Matches Typed Employer Names to the Closest-Sounding Company
AJ Wilcox runs B2Linked, an agency that has managed nothing but LinkedIn ads since 2014, and hosts The LinkedIn Ads Show. The October 9 episode laid out what Wilcox and a number of other advertisers started seeing in March: leads and ad interactions from people who plainly did not fit the company criteria. Not slightly off. Wildly off.
On one account Wilcox described, the campaign targeted only very senior people at companies with 5,000 or more employees. Leads started arriving from one- and two-person companies, at one point more than half of them. The team added exclusion after exclusion. Across the whole run, one lead in three came from someone outside the size range entirely. Wilcox's reasoning about what that implies is the part worth sitting with: if a third of your leads miss the target, it is a safe bet that roughly a third of your impressions and clicks miss it too. You only see the leads.
When Wilcox's team asked LinkedIn support what was going on, the answer was that this is expected behavior.
What changed, by Wilcox's account, is the matching. A member who typed "Acme" as their employer and never linked a Page can now be matched to some large, established company with Acme, or something close to it, in its name. From then on, whenever an advertiser targets by size, LinkedIn treats that person as staff at the big company. And it is not only company size. A separate B2Linked client targeted nothing but a named list of accounts and still received leads from people who worked at none of them. The best explanation they could find was a typed employer name close enough to one of the listed companies to be matched to it.
Why would LinkedIn do this? Wilcox's explanation is simple, and it is a read rather than anything LinkedIn has said. Members who never link to a company Page cannot be reached by company size or company name targeting, so they are inventory LinkedIn cannot sell to B2B advertisers. Those members tend to be at very small, often one-person companies. Matching them to established companies turns them into sellable reach, and it inflates the audience and impression counts you see.
Wilcox cited one advertiser in the show's community, a high spender, who has tracked it closely and raised it with their LinkedIn rep. Starting in March, 30% of their leads suddenly did not match their targeting and had to be thrown out, which pushed their cost per qualified lead up by 30%. If LinkedIn's prices had fallen by a similar amount over the same stretch, it would roughly wash out. They have not. Wilcox says cost per click and cost per thousand impressions have gone up over the same period.
Here is what that looks like on a small account. Say you spend $3,000 a month at $150 per lead. That buys 20 leads, and on paper your cost per lead is $150. If 30% of those leads work somewhere you never targeted, 6 of them are not real prospects. You have 14 leads worth a phone call, and each one actually cost you about $214. Nothing in Campaign Manager changed. The real price of a customer went up by more than $60 per lead, and the only place it shows is in your sales team's patience.
That gap is why we push every LinkedIn client to judge the channel on qualified leads, not reported ones. We have written about how to know whether a LinkedIn lead at three times the price of a Google lead is actually cheap, and that math only works if the leads you count are real.
Your Demographics Report Repeats LinkedIn's Wrong Company Guess
The natural response is to check the reporting. LinkedIn's demographics report breaks down who saw and clicked your ads by job title, seniority, company size and company name. We have recommended it ourselves. In our piece on the demographics report, one of the four questions we told you to ask was whether you bought the right size of company.
For this particular problem, that check no longer works, and we would rather say so plainly than leave the advice standing. The report shows the company LinkedIn decided each member works for. If the matching put a one-person consultancy inside a 400-person carrier, the report counts that click under the carrier, in the 201 to 500 bracket, exactly where you wanted it. As Wilcox points out, the demographics will tell you people are coming from companies of 5,000 and up, because that is what LinkedIn believes. The mismatch is only visible when a human opens the profile and looks.
The report is still worth reading for everything else. Job titles and seniority come from what members write about themselves, and the report remains the best quick way to see whether you bought researchers or decision makers. It just cannot audit the very company match it is built on.
This also cuts against a piece of standard LinkedIn advice. The agency Understory, in a playbook video published October 2, put it this way: "LinkedIn has very good data on the company size," with industry and revenue being the weaker fields. That has been the common view for years, and for members who link their employer properly, it is probably still true. Industry was already known to be shaky, because each company picks its own label, which we covered in why LinkedIn's industry targeting puts companies in the wrong box. What changed in March is that company size, the filter advertisers trusted most, now rests partly on a guess.
Ten Profiles Will Tell You More Than the Dashboard
The check Wilcox recommends takes about ten minutes, and it is the single most useful thing you can do this week if you target by company size or company list. Here is how to run it.
1. Pull your ten most recent leads. Download them from your lead gen form in Campaign Manager, or pull the last ten LinkedIn-sourced leads from your CRM. If you do not run lead gen forms, use the people who most recently engaged with your ads and whose names you can see.
2. Find each person on LinkedIn. Search their name and the employer they gave you.
3. Look at the employer line on their current job. A company logo that links through to a Page means they picked the company from the list. A gray placeholder with plain text means they typed it, and LinkedIn's model decided which company they meant.
4. Check the real company against your targeting. Click through to the company's Page, or search for it if there is no link. Look at the employee count, and if you run a list, whether this is actually a company on it. A placeholder employer that reads like a famous name usually belongs to a much smaller business, or to the person themselves.
5. Compare the email domain. If the lead gave a work email, the domain is a second opinion. A vice president at a 400-person carrier does not usually write from a personal Gmail address or from a one-page website that matches their own surname.
6. Count. Wilcox's threshold: if close to 30% or more of what you check does not match your targeting, you have been affected. In a sample of ten, that is three. One odd lead in ten is normal noise. Three is a pattern.
If your campaigns target a named list of companies, run the same check and ask one question per lead: does this person actually work at a company on my list? A lead from a company you never listed is worth a closer look either way. Some are innocent, like a consultant with two current jobs or someone who changed employers last month. A typed employer name that happens to resemble one of your target accounts is the one to note.
Ten profiles is a small sample, and it is meant to be. You are not measuring the exact rate. You are finding out whether to worry. If three of ten fail, pull the next twenty before you change anything, so you know whether you are looking at a 10% problem or a 40% one.
What You Can Do While LinkedIn Decides Whether to Fix It
There is no setting that turns the matching off. Wilcox's proposals, an opt-out on each ad set, an exclusion segment for fuzzy-matched members, a discount on mismatched traffic, or simply reverting the change, are requests to LinkedIn, not options you can click today. Until one of them happens, the defenses are about catching bad leads early and making sure they do not distort what you pay for.
Ask the question yourself on the form. LinkedIn lead gen forms pre-fill contact and professional details from the member's profile, which makes them easy to submit and is a large part of why they convert. It also means a pre-filled field repeats whatever the profile says. A custom question is different, because the member answers it. LinkedIn allows up to three custom questions per form, and they are required. A multiple-choice question such as "How many people work at your company?" costs you a little submission rate and gives you a qualifier that does not depend on LinkedIn's matching. One detail matters: LinkedIn pre-selects the first answer by default, so put the answer you do not want first. If the smallest bracket is the default, a lazy click disqualifies itself instead of slipping through as a match.
Validate the work email. If you include the work email field, LinkedIn offers a Validate work email option that blocks the common free email domains like Gmail, Yahoo and Hotmail. LinkedIn says it may lower submission rates, and it will. It also forces a domain onto every lead, and the domain is the cheapest company check there is.
Count qualified leads, not reported ones. Score every LinkedIn lead in your CRM as a fit or not a fit, and measure the campaign on cost per qualified lead. Then send only the qualified ones back to LinkedIn as conversions, so the bidding stops chasing more people like the junk. We walked through that setup in teaching LinkedIn which leads are actually worth having. It was good practice before March. Now it is the only number that tells you the truth.
Make the company filter one layer of several. Pair company size with job function and seniority, so that a company match alone cannot qualify someone for your ad. To be honest about the limits, this narrows the damage rather than ending it. Wilcox's team added exclusion after exclusion on the account where a third of leads missed, and the problem persisted, because the people being let in carry senior titles of their own. What layering does is stop the matching from being the only gate.
Build company lists from Page URLs. If you upload a target account list, include each company's LinkedIn Page URL, LinkedIn's own identifier for the company. As we explained in why an uploaded list shrinks once LinkedIn matches it, a bare company name can match the wrong Page entirely. That fixes the matching on your side of the upload. It does not fix LinkedIn's decision about which members belong to the company, so keep running the ten-profile check.
Report it to LinkedIn, with evidence. Most small advertisers do not have a LinkedIn rep, but anyone can open a support ticket in Campaign Manager. Include the profiles that failed your check, your targeting settings, and the dates. Ask whether you can receive credit for the off-target traffic. You may get the same "expected behavior" answer. The more advertisers who document it, the harder that answer is to keep giving, which is exactly why Wilcox asked listeners to raise it.
Company Size Still Works. It Just Can't Work Alone.
None of this means company size targeting is useless. For members who linked their employer properly, it still does what it says, and on most accounts that is still the majority of the audience. What changed is that it can no longer be the only thing standing between your budget and the wrong people.
That matters more for a small account than a large one. An enterprise advertiser spending six figures a month can absorb 30% waste for a while and still collect enough real leads to keep sales busy. At $3,000 a month, 30% is the difference between a channel that pays for itself and one that gets cut at the next budget review. Small budgets do not have room for a filter that quietly lets in a third of the wrong people.
So the setup that holds up now looks like this: company size or a company list as one layer, job function and seniority on top, a qualifying question on the form, work email validated, and every lead scored as a fit or not before anyone decides the campaign is working. Then the ten-profile check, once a month, so you know when LinkedIn's matching moves again.
If you would rather have someone else run that audit, score the leads and restructure the targeting for you, that is everyday work for us in LinkedIn ads management. And if you would rather do it yourself, start with ten profiles this week. They will tell you more about your campaign than anything in the dashboard.
