Firmographic Audience Architecture for LinkedIn Campaigns

The targeting decision that moves your number is almost never which job titles you picked. It is how many audiences you are willing to run at the same time. LinkedIn needs at least 300 matched companies or contacts before a list will activate, and an audience taking less than roughly a fifth of your weekly budget cannot be read inside a month. So we build three to five audiences per account rather than forty, and we tell you which of them we would drop first. The list work nobody bills for: Account lists arrive from Apollo, ZoomInfo or a sales spreadsheet and they are wrong in predictable ways. We normalise to registrable domain, roll subsidiaries into the parent where LinkedIn recognises the parent and split them where it does not, drop free-mail and reseller domains, and reconcile against the company records LinkedIn actually holds. A South African manufacturer trading under three names is often one live page and two ghosts, and matching against the ghosts quietly removes a third of your target market. The upload limit is 300,000 rows. The useful list for a capital equipment seller is usually under 900. Where the firmographics mislead in this market: Company size comes from the employee count on the page, so a 200-person foundry with a neglected page sits in the 11 to 50 bucket and a size filter deletes it from your buy. Function mapping puts a stores controller, a buyer and a procurement officer in the same bucket as a retail purchasing clerk. Seniority is inferred from title text, and local industrial titles, Technical Director, Works Manager, Financial Director, do not map the way American ones do. We run title-based and function-based versions of the same audience against each other instead of assuming either is right. Exclusions come first: Current customers, open opportunities, your own staff, your distributors and resellers, recruiters, agencies, and the competitors who click your ads to read your copy. On most accounts the exclusion lists change cost per qualified enquiry more than the inclusion lists do, and they are the first thing we rebuild. Settings we switch off, and why: Audience expansion, on every campaign, because it quietly returns the reach you just paid us to remove. The LinkedIn Audience Network on anything with a form behind it. Member interests and traits as primary targeting, which are fine as a test cell and never as the core buy. Predictive audiences need a source of at least 300 members, and it has to be built from converters rather than page visitors, or it inherits whatever noise the tag collected. What you receive: An audience map giving each segment, its LinkedIn size, the exclusions applied, the budget share it can carry and the date on which we will judge it. Matched lists loaded in your own account, under your ownership. A refresh specification saying which lists are rebuilt monthly, from which source, by whom. And a written note of the segments we considered and rejected with the reason, because that is the argument you will be having internally in six weeks. What we will not do: Scrape LinkedIn for member data or sell you a contact list. Build audiences for an account that cannot exclude its existing customers, since the resulting cost per lead is fiction. Split a segment below the size at which its result can be read, however good the story sounds in a deck. Who should not buy this: Anyone whose problem is the offer. If enquiries stall after the first call, targeting is not what is broken, and a sharper audience will deliver the same disappointing conversation to more people at a higher cost per head.