Linkedin ads targeting

Start with Job Function, not Job Title

Job Title targeting is tempting - 'VP of Marketing' feels precise. It's also LinkedIn's least reliable signal because titles are self-reported and inconsistent ('Head of Growth' = same role at three companies, three titles). Job Function ('Marketing', 'Engineering', 'Sales') is canonicalized by LinkedIn and 5-10x more reliable. Layer Seniority on top to narrow within a function. Reserve Job Title for very specific ABM-style targeting.

Layer Seniority + Company Size + Industry

The four-layer combo that works for most B2B SaaS: Job Function (Marketing, Sales, IT) + Seniority (VP, Director, C-Suite) + Company Size (50-500, 500-1000, 1000-5000) + Industry (SaaS, Internet, Computer Software). Each layer should reduce the audience by 30-70%. If a layer reduces it by less than 20%, it's not adding targeting signal; if it reduces by more than 90%, you're stacking too many constraints.

# Typical layered audience build:
Job Function: Marketing, Operations
  → audience: 8M
Seniority: VP, Director, C-Suite
  → audience: 1.2M (-85%)
Company Size: 50-200, 200-500
  → audience: 280k (-77%)
Industry: SaaS, Computer Software, Internet
  → audience: 95k (-66%)
Geography: United States, Canada, UK
  → audience: 62k (-35%) [in sweet spot]

Hit the 50k-500k audience size sweet spot

Below 50k: bidding algorithm doesn't have enough impression inventory to optimize, CPMs spike. Above 500k: targeting is too loose, you'll pay for impressions on people who aren't your buyer. Sweet spot: 50k-500k. If your layered audience is 20k, loosen one constraint (broaden seniority or add a peer industry). If 1.5M, add a constraint (tighter company size or geo).

Build matched audiences for ABM and retargeting

Matched audiences let you target specific companies (Company Match), specific people (Contact Match), or website visitors (Website Audience). Upload via Campaign Manager → Plan → Audiences. Company Match needs LinkedIn Company URN or name + domain (LinkedIn matches ~60-70% of submitted accounts). Contact Match needs email or first name + last name + company (~30-50% match rate, varies by list quality).

Use lookalikes (Audience Expansion) carefully

LinkedIn's lookalike feature (now called Audience Expansion + Predictive Audiences) takes a seed audience and finds similar members. Quality of lookalikes depends entirely on quality of seed. Seed with top-quartile-LTV customers, not all customers. Expand modestly: 1-3% lookalikes outperform 5-10%. Treat lookalikes as a complement to direct targeting, not a replacement.

Exclude what doesn't fit

Most teams skip exclusions and over-pay. Exclude: existing customers (upload CRM customer list as exclusion), competitors' employees (Company Match exclusion), job titles that look right but aren't (Students, Interns, Founders if you sell to enterprise). Exclusions don't cost you reach - they cost you waste impressions.

Test one variable per audience experiment

Don't build two test audiences that vary on multiple dimensions simultaneously - you won't know what drove the difference. Keep the base audience constant; change one layer per test. 'Test A: VP+Director vs Test B: Director only' is a clean test. 'Test A: VP marketing at SaaS vs Test B: Director IT at fintech' tells you nothing.