First-Party Data on LinkedIn Ads: Why 21% of Your Company List Never Reaches an Ad

Where we left off

In the first article of this series I wrote about broad targeting and why it damages lead quality more than it raises cost. The natural next question is: so what should you target instead?

In most of the enterprise accounts I work on, the answer is already inside the company: in the customer relationship management (CRM) system, in customer lists, in the lists of target companies sales has been building for years. The hard part is getting that data into LinkedIn without losing a good part of it on the way, and then not using it all as one big list.

One of our clients, an Italian manufacturing company, had a CRM that tracked the whole sales flow, from the moment a lead came in to the moment the deal closed, including its value. That CRM was only used internally, to show the numbers. None of it was in the LinkedIn account. The campaigns were all pure prospecting, on a target that was far too broad for the budget available, so each company saw the ads too rarely.

On LinkedIn the target is fundamental, as I wrote in part one: an audience that is too small or too big is useless. The right frequency matters just as much.

The data you already own

First-party data is what you collect and own yourself. On LinkedIn that means the signals the platform gathers for you: website visitors through the Insight Tag, people who opened or submitted a Lead Gen Form, video and document viewers, Company Page followers and visitors. It also includes what you own outside LinkedIn, mainly your CRM, HubSpot or Salesforce in most of our accounts, and your marketing automation tool.

The list in this article is a list of target companies we put together internally. The names are the full legal names of the companies, which, as you will see, matters a lot.

In my experience lists break in two different places. Contact lists struggle because people log into LinkedIn with a personal email and use their work email everywhere else. Company lists struggle because your systems usually hold the legal name of the company, and LinkedIn holds the brand name.

What happened to our company list

::: figure src: /images/company_list_match_rate.png alt: Company list audience detail, 118,466 matched and 31,370 unmatched :::

The list had 149,836 companies. LinkedIn matched 118,466 of them, so 79%. The other 31,370 did not match.

That does not mean those companies are not on LinkedIn. Many of them are. Matching is often hard because the legal name of a company is not the name of its LinkedIn page. The platform tries to harmonise the two, but it is not always possible, and so part of the data gets lost.

In my experience 79% is a perfectly normal number, I would say about average. Going much higher is hard, not impossible, but hard, for exactly the reasons above. I actually expected less.

The matches need checking too, though. Once, in a list of small and medium-sized enterprises (SMEs), there was a company called Mario Ferrari, and LinkedIn matched it to Ferrari, the car brand.

The matched tab helps a lot here, because it shows the name you uploaded next to the company LinkedIn picked:

IBM S.R.L. going to IBM is exactly what you want. Siemens and Deloitte need a closer look.

Siemens Mobility ended up on the global Siemens page. So the campaign is no longer reaching the Italian mobility business that was on our list, it is reaching Siemens in general, and only location and job targeting keep it from going much wider. The same happened to Deloitte's Italian tax services company, which went to the global Deloitte page.

When I went through the unmatched records, the problem was clear. Many names did not just end in S.R.L.: they carried the full legal form spelled out, SOCIETA' A RESPONSABILITA' LIMITATA SEMPLIFICATA (simplified limited liability company). That is never part of a LinkedIn page name, so we removed it and uploaded the list again. Even so, many of the remaining problems involved small companies.

What we do now before uploading a company or contact list:

• remove S.R.L., S.P.A., the legal form when it is spelled out in full, and anything in brackets from the company name

• add the website domain

• one company or one contact per row

• for contact lists, send every field you have, not only the email

• after matching, open the matched tab and sort it by company size and industry; the companies that do not belong usually stand out quickly

• fix the file and upload it again

::: figure src: /images/list_upload_dialog.png alt: LinkedIn list upload dialog recommending 10,000 to 300,000 records :::

LinkedIn suggests between 10,000 and 300,000 records for the best results. In my experience, the bigger the list, the better. You narrow it down afterwards with the other targeting criteria.

One website audience is not enough

::: figure src: /images/website_audience_timeframes.png alt: Website audience timeframe options in Campaign Manager :::

When I open a new account, most of the time there are no website audiences at all. When there is one, it is a single list: all visitors in the last 180 days, with no segmentation.

Inside that audience there is someone who looked at a product page yesterday and someone who read one blog post five months ago. LinkedIn treats them the same, and so does the ad they see.

What I do instead is split visitors by two things: which pages they visited and how long ago.

Fifteen audiences sounds like a lot. They are quick to set up and cost nothing until you use them. The useful part is that the time window tells you something about where a company is in the buying process, and in B2B that process can take months. In my experience, people who saw one of our ads in the last 14 days are considerably more likely to convert than older audiences, as long as the targeting is right.