Algorithmic vs rules-based attribution
Attribution models split into two families. Rules-based models apply fixed, human-chosen weights to touchpoints by position — last-click, first-click, linear, time-decay, U/W-shaped. Algorithmic (data-driven) models learn credit from observed conversion paths using methods like Shapley values or Markov chains. This page contrasts the two and explains when each is appropriate.
What this means
Rules-based attribution encodes a belief about which touches matter — last-click says only the closer matters, linear says all matter equally, time-decay says recent matters more. The weights are chosen by people and stay fixed regardless of the data.
Algorithmic attribution instead estimates credit from the data itself. It compares converting and non-converting paths and assigns credit based on how channels' presence changes conversion likelihood, using techniques such as Shapley value averaging or Markov removal effects. Google's data-driven attribution is the best-known production example.
Trade-offs
Rules-based models are transparent, reproducible, need little data, and are easy to explain to stakeholders — but their weights are arbitrary and can systematically misvalue channels. Algorithmic models adapt to your actual journeys and avoid hand-picked weights, but they need sufficient conversion volume, are harder to audit, and can quietly understate channels that are undertracked.
Neither family proves causation on its own. Both describe how recorded credit is distributed; for causal claims you still need incrementality experiments or marketing-mix modeling.
- Rules-based: fixed weights, transparent, low data need
- Algorithmic: learned weights, adaptive, data-hungry
- Neither is causal proof without experiments
How it appears in analytics and logs
If credit follows a fixed positional formula, the model is rules-based; if credit shifts as the underlying path data changes, it is algorithmic. Each implies different trust and data conditions.
Diagnostic use case
Decide between a transparent fixed rule and a data-driven model by weighing interpretability and data requirements against the risk of arbitrary weights.
What WebmasterID can help detect
WebmasterID provides first-party path data that can feed either approach, so you can apply a simple rule for clarity or supply an algorithmic model with the journeys it needs.
Common mistakes
- Choosing an algorithmic model without enough conversion volume.
- Assuming data-driven credit is automatically more 'true'.
- Treating either family's output as causal lift.
Privacy and accuracy notes
Both families operate on touchpoint paths, not identity. Algorithmic models need more path volume but still work on aggregated, de-identified journeys.
Related pages
- Data-driven attribution: promise and caveats
Data-driven attribution (DDA) assigns credit using a model trained on a site's own conversion paths rather than a fixed rule like last-click. Done well it credits assist touches more fairly. Its caveats are real: it needs enough conversion volume, it is a model not a measurement, and it cannot see touches that were never tracked.
- Shapley value attribution
Shapley value attribution applies a concept from cooperative game theory: it treats channels as players in a coalition and assigns each one credit equal to its average marginal contribution across all possible orderings of channels. The result is a principled, order-independent way to split conversion credit. It underpins data-driven attribution in several analytics products.
- Markov chain attribution
Markov chain attribution models customer journeys as a probabilistic graph of transitions between channel states, ending in conversion or null. Each channel's credit is derived from its 'removal effect' — how much the overall conversion probability falls if that channel (and its transitions) are removed from the graph. It is a leading algorithmic alternative to Shapley-based attribution.
- Attribution analytics
First-party paths for either attribution approach.
Sources and verification notes
- Google Analytics Help — About attribution and attribution modelsDistinguishes rules-based models from data-driven (algorithmic) attribution.
Last reviewed 2026-06-24. Facts are checked against primary/official sources where available; uncertain specifics are marked “Data not yet verified” rather than guessed.