Somewhere in your CRM there is a spreadsheet of everyone who ever asked you for a quote. Two thousand names, give or take. It represents years of work, and it is the closest thing a small B2B firm has to a proprietary asset.
So you upload it to LinkedIn to run ads at those people. Twenty minutes later the Audiences page shows a number that is nothing like two thousand, and there is no explanation attached to it.
The number is not a mistake. It is the result of your list passing through two separate filters, each of which removes people for a different reason, neither of which LinkedIn explains at the moment it happens. Once you know where the losses come from, most of them are recoverable.
The List You Upload Is Not the Audience You Get
LinkedIn publishes a worked example of this in its own help documentation, and it is worth reading slowly.
You upload a segment with 2,000 entries. Of those, 1,400 match to a member account. Some of your entries are duplicates, so those 1,400 matches collapse further. With 200 duplicates in the file, your usable audience lands at 1,200.
Two thousand became twelve hundred. Forty percent of the list is gone, and it left in two different ways.
The first loss is the match rate, which LinkedIn defines as the percentage of the entries in your source data that successfully matched to a member account or Page in the past seven days. That is the number most people look at, and it only describes the first filter.
The second loss shows up in a different column called last audience count. That figure deduplicates your entries and removes member accounts that have opted out of this kind of targeting in their ad settings. It is the honest number, the count of people you can actually reach.
These two metrics get treated as the same thing constantly, and they are not. A respectable match rate can still deliver a disappointing audience, because dedupe and opt-outs happen after matching. If your CRM holds the same person under two email addresses, both may match, and both then collapse into one reachable human.
Check both columns on the Audiences page before you build a campaign around a list. The gap between them is a direct readout of how clean your CRM is.
LinkedIn Is Checking an Email Address You Probably Do Not Have
Here is the single biggest cause of a low match rate on a contact list, and it appears in one quiet line of LinkedIn's documentation: contact targeting matches against member accounts that have verified email addresses only, which LinkedIn itself notes may result in lower match rates.
Think about what that means for the file you just uploaded.
Your CRM holds work email addresses, because that is what people type into a quote form. The address on someone's LinkedIn account is frequently the personal one they signed up with in 2012, at a job they left twice ago. Those two addresses have nothing to do with each other. The person is on LinkedIn, they are exactly who you want to reach, and your row still fails to match, because you brought a key that does not fit the lock.
You cannot fix this by trying harder on work emails. You fix it by giving LinkedIn more than one way to find the same person.
LinkedIn accepts three identifier types for contact lists, and asks for at least one:
- Email address, either plain or hashed with SHA-256. Unhashed addresses are hashed in your browser before they are stored.
- First and last name, which match considerably better when you also supply the person's company name.
- Mobile device ID for Google AID, if you happen to hold it.
The instruction that matters most is the one people skip: if you hold more than one identifier for a person, supply all of them. Two email addresses for the same contact is not a data-hygiene problem here, it is two chances to match. Put them in as separate rows and let LinkedIn find whichever one it recognizes.
Name plus company is the underrated combination for a small firm, and it works for the same reason job title targeting misses half your buyers: LinkedIn holds more than one kind of signal about a person, and you should hand it every one you have. You may not have anyone's personal email. You almost certainly have their name and where they work, sitting in the same CRM record.
There Are Several Companies Called Nike. You Meant One of Them.
Company lists fail differently, and more quietly, because a bad company match does not look like a failure. It looks like a match.
Type "Nike" into LinkedIn's company targeting box and you do not get Nike. You get Nike the retailer with ten thousand employees, and Nike Communications, which is a different business with a different logo, and Nike Australia, and a scattering of regional entities and store pages. They are all real LinkedIn Pages. They are all called Nike.
Now imagine that name arrived in a spreadsheet row with nothing beside it. LinkedIn has to pick one. It may pick the public relations firm, and your ads will run, and your reporting will show impressions and clicks, and you will be advertising running shoes to publicists.
This is why the columns you left blank matter more on the company side than the contact side. LinkedIn matches company rows on any of:
- Company name
- Company website
- Company email domain
- LinkedIn Company Page URL
- Stock symbol
The Page URL is the one to chase. It is LinkedIn's own identifier for that organization, so it removes the guesswork entirely, and LinkedIn explicitly recommends including it for better match rate and accuracy. Finding the Page URL for a hundred target accounts is a genuinely dull afternoon. It is also the difference between an ABM campaign and an expensive coincidence.
There is a shortcut worth knowing for smaller lists. If your target list is short enough to type by hand, skip the upload entirely and enter company names directly in the campaign builder. You see every ambiguous match as you go, you pick the right entity yourself, and you skip the matching wait. As a rule of thumb, if typing the names would take under half an hour, type them. Above that, build the file properly.
Typing them by hand has a second benefit that the upload path does not give you. As you add companies, LinkedIn suggests similar ones beside them. Start with Nike and it offers Under Armour and Adidas. For a firm still working out which accounts to chase, that panel is free prospecting.
300 Is the Number That Bites You Three Times
Three hundred is the floor everywhere in LinkedIn's audience system, and small B2B advertisers meet it in three separate places without realizing it is the same number.
Three hundred rows to upload. Your file needs at least 300 rows or the upload fails outright. The ceiling, for context, is 20MB or 300,000 records.
Three hundred matched member accounts to be usable. Uploading is not qualifying. Your list must match a minimum of 300 member accounts before it can run in an active ad set. Below that, the Audiences page shows < 300 members in the last audience count column and greys out the list name. It sits there looking like an audience and cannot be used as one.
Three hundred in the ad set, after targeting is applied. This is the one that catches people. The minimum ad set audience size is 300, and location is a required targeting facet. So a list that cleared 300 matches nationally can drop back under the floor the moment you narrow to your state, because location is not optional and it subtracts.
One more piece of diagnostic trivia that saves an afternoon of confusion: if your list matched fewer than 300 member accounts, LinkedIn displays the match rate as < 5%. That is a floor on the display, not a measurement. A list showing < 5% has not necessarily matched terribly. It has matched fewer than 300, and LinkedIn stopped reporting precisely.
If you land under the floor, you have three moves. It is worth reading these next to what a healthy LinkedIn audience size actually looks like, because clearing 300 is the floor, not the target. Apply several matched audience segments to the same ad set to clear 300 between them. Switch from a contact list to a company list, since matching organizations is far more forgiving than matching individuals. Or widen the geography you were about to narrow.
You Attached the File. You Did Not Upload It.
The mechanics of getting a list into LinkedIn are unremarkable except for the places they quietly go wrong.
Download LinkedIn's template first and paste your data into it. Do not rename the headers on your own export to match. Changing or removing the template headers causes the upload to error, and the error message will not tell you that is why.
Then the trap that catches almost everyone once. Selecting your file makes it appear on screen, which looks exactly like it worked. It has not uploaded. You have to scroll down and press Agree and upload. People walk away from a file that never went anywhere and come back the next day wondering why there is no audience.
After that, wait. LinkedIn says it might take up to 48 hours to generate your audience, and on rare occasions longer. In practice 24 to 72 hours covers most lists, with bigger files at the slow end. There is nothing to fix during this window, so do not re-upload and create a duplicate segment.
You can attach a segment to an ad set while it is still building. It will start delivering on its own once it crosses 300 matched accounts, which is the right move when you are waiting on a list and want the campaign ready to go.
One structural limit to plan around: one list per audience segment. If you have three lists, you are creating three audiences. There is no merging on upload.
And then the clock nobody mentions, which is genuinely important for a small firm running campaigns in bursts. LinkedIn deletes contact list data automatically within 30 days. The audience built from that list is deleted within 90 days if it is not edited or used in an active ad set. Build a list in January, run a campaign for a month, pause everything for the quarter, and the audience you built will not be waiting for you in May. If a list matters, keep it attached to something live or plan on rebuilding it.
Your List Already Wrote You a Report. Almost Nobody Opens It.
Upload a company list and LinkedIn generates an engagement report against it that most advertisers never look at, which is a shame, because it answers the question everyone actually has.
It shows you which companies matched and which ones did not. That is your data quality problem itemised by name, rather than expressed as one demoralising percentage. Fifteen unmatched accounts you can go and fix by hand is a completely different task from a 70% match rate you can only feel bad about.
It also shows an engagement level for each matched company, from very high down to very low, alongside impressions. That is a rough read on how aware each organization already is of your brand. For a small firm doing outreach alongside ads, the high-engagement rows are the warmest names on your list, and they are sitting in an ads dashboard rather than a sales tool.
Separately, the list insights view breaks your matched audience down by job function, seniority and location. Use it as a sanity check the day your list finishes building, because it catches the failure that is otherwise invisible: a list that matched plenty of people, just not the ones you meant. If you uploaded 400 target accounts expecting operations directors and the seniority breakdown is mostly entry level, your audience is technically healthy and strategically useless. Better to learn that before the budget runs.
A List Is a Seed, Not an Audience
Uploading an account list does not target the buyer at those companies. It targets everyone at those companies, from the summer intern to the president.
On LinkedIn, where clicks are expensive, that is an expensive way to be approximately right. The fix is to treat the list as the first layer rather than the whole audience, then narrow with the attributes LinkedIn is good at: job function, seniority, sometimes skills. Your company list says which organizations. The attributes say which desk.
That is genuinely powerful for a small B2B firm, because it is the one thing this platform does that no other channel really matches. You bring the account list. LinkedIn brings the org chart. If you are building that list from scratch rather than exporting it, targeting companies by name is the other half of this.
Then turn off audience expansion. It is a checkbox in the campaign builder, LinkedIn prefers it on, and it shows your ads to people with similar attributes to your target audience. On broad attribute targeting there is a reasonable argument for it, since job titles are a free-text field and LinkedIn's buckets genuinely miss people.
On a list you built deliberately, it undoes the work. You spent an afternoon deciding which 200 companies deserve your budget. Audience expansion's entire job is to go and find companies that were not on it.
Two independent practitioners we read while researching this arrived at the same instruction unprompted, in almost the same words: uncheck it. When the platform's default and the people who run these accounts daily disagree, back the practitioners.
The Safety Net Under a Short List Is Gone
This all used to matter less, because there was a fallback. If your list was small, you built a lookalike audience from it and let LinkedIn find more people like them.
LinkedIn discontinued lookalike audiences on 29 February 2024. They cannot be created or edited. Existing ones stopped refreshing and became static.
What replaced them raises the stakes on the list itself. Predictive audiences combine your source data with LinkedIn's AI to build an audience of members judged more likely to convert, which is a meaningfully different promise from the old "people who look similar." But look at what a predictive audience needs to exist:
- A source, which can be a contact or company list, a conversion, a Lead Gen Form, or a retargeting audience.
- If that source is a list, it must hold between 300 and 300,000 rows, with a status of Ready or Updating.
- If it is a conversion, form or retargeting audience, it needs more than 300 member accounts, though you may combine several sources to clear the bar.
There is a hard cap of five predictive audiences per ad account, and they cannot be shared between accounts. Note also that enabling one disables audience expansion on that campaign, which is LinkedIn's own admission that the two mechanisms do the same job.
Read that list of requirements again as a small firm. The tool that used to rescue a thin list now requires a healthy list to run at all. Your CRM export is no longer just one audience among many. It is the entry requirement for LinkedIn's own expansion machinery, and every unmatched row is a row that no longer counts toward it.
Which turns list quality from housekeeping into strategy. The half hour you spend adding Company Page URLs, or the second email address for a contact, is not tidying. It is the difference between having access to the platform's best targeting and being locked out of it.
A last thing worth knowing, since it decides how hard you should work on this: a matched list is also the cleanest way to tell LinkedIn which of your leads were actually any good, and small professional audiences wear out faster than most owners expect, so the list you build now is not a one-time job.
Start with what you already own. Export your customers and your closed-lost quotes, put them in LinkedIn's template properly, fill in every column you can, and look honestly at both numbers when it finishes building. That gap between what you uploaded and what you can reach is the most useful thing LinkedIn will tell you about your own data all year.
We run this on client accounts every week, and if you would rather hand the spreadsheet to someone who has done it a hundred times, we can take it from here. If you would rather run it yourself, everything above is the whole method.
