Call Attribution

Call attribution answers the fundamental question: 'Which of my marketing efforts actually generated this phone call and was it valuable?' Without attribution, a business running Google Ads, Facebook, and organic SEO has no idea which channel is actually driving calls and revenue.

There are three layers of call attribution:

**Click-level attribution** uses GCLID (from Google Ads) or equivalent IDs from other platforms to connect a call to the exact click that generated it. This is the most precise: you know the call came from 'personal injury lawyer near me' in California on March 15 at 2:47pm. Click-level attribution is only possible with DNI on your website, where you capture the GCLID before the call happens.

**Campaign-level attribution** groups calls by broader marketing source: Google Ads, Facebook, organic search, direct calls, etc. This is coarser but doesn't require any special tracking infrastructure. You manually assign calls to campaigns based on which number they called or which landing page they came from.

**Probabilistic attribution** uses machine learning to estimate which click likely generated a call based on anonymized signals: geography, time of day, device type. This is useful for calls where the GCLID was lost (some iOS traffic) but is inherently approximate.

The best attribution systems use all three in combination. DNI captures GCLID on every website click, enabling precise click-level attribution. For calls that arrive through other channels (direct dials, 3rd party referrals, brand searches), campaign-level tracking via number assignment or landing page IVR data bridges the gap. Probabilistic matching fills remaining gaps where possible.

Once you have attribution, you can measure: cost-per-call per channel, conversion rate per keyword, lifetime value of calls from each buyer, and optimal bid strategy. This is the foundation of data-driven media buying.