Free Marketing Attribution Calculator

The fundamental problem

A customer touched five channels before they bought. Which channel deserves the credit? There is no factual answer. Counterfactual attribution ("would they have bought without channel X?") is impossible to measure for a single conversion. So every attribution model is a heuristic that picks a rule and applies it consistently. The rule is always wrong about some fraction of conversions. The art is picking a rule whose wrongness is acceptable for your business.

The math, top to bottom

For N touches with days_before_conversion[i]:

first_touch:    weights = [1, 0, 0, ..., 0]
last_touch:     weights = [0, 0, ..., 0, 1]
linear:         weights = [1/N for each]
time_decay:     w[i] = 0.5 ^ (days_before[i] / half_life)
                normalize so sum(weights) = 1
u_shaped (N>=2):
  weights = [0.4, ..., 0.4]
  middle touches share 0.2 equally

Each weight is multiplied by the deal value to assign credit per touch. The sum across touches always equals 100 percent of the deal value. The disagreement between models is purely about how that 100 percent gets distributed.

How to choose

• First-touch: Use when budgeting brand and awareness channels. Avoid when reporting on closing channels.
• Last-touch: Use for closing channel ROI (which paid keyword closed?). Avoid as a primary report because it overweights direct, brand, and retargeting.
• Linear: Use when you have no theory about which touches matter more. Useful as a comparison baseline.
• Time-decay: Use for short B2C cycles where recency dominates. Tune half-life to match your typical buying window.
• U-shaped: Use for B2B journeys with clear introduction and close stages. Most B2B SaaS teams default to this.

Why models disagree the most on direct touches

Direct traffic (someone typing your URL or clicking a saved bookmark) almost always gets last-touch attribution. The customer did not arrive from a campaign, so the visit appears unsourced and last-touch credits the conversion to "direct". First-touch and U-shaped both correctly give credit back to whatever channel introduced the customer weeks earlier. The difference between last-touch direct credit and U-shaped direct credit is often the single most revealing comparison in attribution.

The tradeoff between fidelity and operability

Data-driven attribution (Google's DDA, Adobe Algorithmic) is the most accurate model in theory because it learns weights from your conversion history. The downside: it requires hundreds of conversions to train, results are uninterpretable ("the model says LinkedIn gets 23 percent because of complex interaction effects"), and weights drift as the model retrains. Most teams settle on U-shaped or time-decay for their primary reporting and use DDA as a sanity check.

A worked example

Acme Co. saw a LinkedIn ad 45 days ago, opened a newsletter 21 days ago, clicked a Google Ads CPC link 7 days ago, and came back via direct today to convert. Deal value: $5,000. First-touch credits all $5,000 to LinkedIn. Last-touch credits all $5,000 to Direct. Linear gives each touch $1,250. Time-decay (7-day half-life) puts the most credit on Direct ($2,375), then Google Ads ($1,750), then Email ($625), then LinkedIn ($250). U-shaped gives LinkedIn and Direct $2,000 each, Email and Google Ads $500 each. Same journey, five very different stories. The right one for your team depends on whether you are budgeting for awareness (first or U-shape), measuring closing channels (last), or trying to balance both (linear or time-decay).