Bridging Marketing Theory and Big Data Analytics: The Taxonomy of Marketing Attribution

Part of: Digital Marketing Topic 2 — Media Strategy, Platforms & Channels · Reading Citation: Buhalis, D., & Volchek, K. (2021). Bridging marketing theory and big data analytics: The taxonomy of marketing attribution. International Journal of Information Management, 56, 102253. Key concepts: Marketing Attribution, Customer Journey, Touchpoint, Multi-Touch Attribution, Data-Driven Attribution, Purchase Funnel

---

TL;DR

As consumers touch a brand across many channels and devices, allocating credit for a conversion to the right touchpoint becomes hard — and the terminology for doing so ("multi-channel," "omni-channel," "advanced," "algorithmic") has become inconsistent and overlapping. This conceptual paper builds a five-dimensional taxonomy that systematically names and describes any Marketing Attribution method. It combines a deductive conceptual framework (rooted in consumer decision-making and the capabilities of big-data analytics) with an inductive systematic literature review of 62 sources. The taxonomy's punchline: even today's "advanced" methods still fail to systematically capture the context of decision-making, so attribution remains imperfect.

Why It's on the Reading List

It is the analytics backbone of Topic 2 (Media Strategy, Platforms & Channels): it teaches how marketers decide which channel/touchpoint earned a sale, why single-touch models (first/last click) are biased, and the vocabulary (single- vs multi-touch, rule-based vs data-driven, cross-digital/platform/channel) you need to talk precisely about measurement.

Background & Research Question

Mobile and wearable devices have exploded the number of Touchpoints between consumers and brands. Big data and advanced analytics let firms estimate marketing ROI far more granularly — but rapid innovation has produced heterogeneous, overlapping terms with no unified scheme. No framework yet exhaustively summarises all available attribution methods and their capabilities. Research aim: develop a comprehensive tool for naming and describing marketing attribution methods, and assess their ability to realistically allocate value along the Customer Journey.

Key Concepts & Definitions

Definition — Marketing attribution

A strategy of determining the value of marketing communications and allocating that value to the touchpoints along a customer journey. Its distinctive feature is the use of individual-level, high-frequency big data plus advanced analytics.

Definition — Touchpoint

A single interaction between a customer and a brand, experienced via a channel as a marketing communication (e.g. an ad). Each touchpoint can have a positive, negative or neutral effect on the decision to convert.

Definition — Carryover vs spillover effects

Carryover = a prior interaction's overlapping effect within one channel. Spillover = the overlapping effect between different channels. Touchpoint sequences can be synergic or antagonistic — exposure is not simply cumulative.

The conceptual framework (Fig. 1) defines attribution by facilitators (the data and analytics that enable it) and capabilities (what it can account for in consumer behaviour), all assessed against the Marketing Mix and Purchase Funnel (awareness → interest → desire → action).

Main Arguments / Findings — The Five-Dimensional Taxonomy

The taxonomy is a second-order hierarchy: first order = facilitating parameters + resulting capabilities; second order = mutually-exclusive classes. Any real method is described by picking one class on each of the five dimensions.

Dimension detail

1. Single- vs multi-touch.

• Single-touch (a.k.a. single-channel) assigns the whole conversion value to one touchpoint. Earliest forms: first-click and last-click. Simple and available, accurate for short journeys, but biased for long ones (ignores timing, sequence and causality).
• Multi-touch distributes value across several touchpoints — more realistic, generally more accurate.

2. Fractional vs incremental.

• Fractional assigns proportionate value to each touchpoint independently of the others (easy ROI; often equated with "rule-based"). Does not capture synergy.
• Incremental / synergic accounts for the cumulative effect between touchpoints — more realistic but needs complex modelling.

3. Channel scope.

4. Rule-based vs data-driven.

• Rule-based applies predefined assumptions: e.g. U-shape (40% first + 40% last + 20% middle), time-decay (more weight the closer to conversion), weighted/linear. Simple, cheap, no advanced analytics — but heuristic and blind to journey dynamics.
• Data-driven uses individual-level data to empirically determine each touchpoint's role (regressions, machine learning, cooperative game theory). More accurate, especially for long journeys.

5. Standardised vs custom (new terms proposed). The authors propose "standardised attribution" (fixed/predefined computation) and "custom attribution" (model fitted to the specific dataset) to replace the muddled use of "rule-based," "algorithmic" and "data-driven." Custom attribution can model all events, customer heterogeneity and overlap — but is costly and sometimes performs no better than simple rules.

Reading the map (Fig. 3)

The five dimensions form a map: identify one parameter and the likely others follow. E.g. a cross-device method "will likely be multi-channel, custom-made and data-driven, but can apply either fractional or incremental value allocation."

What current methods still miss

No reviewed method systematically accounts for the context of decision-making (internal context: demographics, culture, personality; external context: location, time, weather, social setting). "Advanced" is often used as a marketing label, not a precise capability. The full analytical capacity of attribution has not yet been met.

Methodology

Conceptual paper using taxonomy development (Nickerson et al., 2013), combining:

• Deductive reasoning → conceptual framework from big-data analytics + consumer behaviour theory.
• Inductive reasoning → systematic literature review: keyword search of academic databases (Scopus, ScienceDirect, Google Scholar) plus industry white papers and vendor reports, supplemented by snowball sampling. Located 164 academic + 31 industry sources; after screening (post-2005, relevance, quality), 62 sources were analysed via qualitative content analysis with two rounds of descriptive coding (by title, then by description) and triangulation.

Framework / Model

The output is the taxonomy (Fig. 2) + the five-dimensional map (Fig. 3). Three hypothesised customer-journey cases illustrate why context matters:

• Health insurance — long, multistage, emotional, credence service; company-initiated communications get over-credited because attitudinal loyalty isn't modelled.
• Low-cost airline tickets — short, utilitarian + relational + hedonic motives; value attributable very early at the "desire" stage.
• Dining choice — review/metasearch driven; external context (weather, traffic, social party) can disrupt earlier touchpoints, making allocation nearly impossible.

Implications for Marketers

• Choose attribution by journey length: single-touch is fine for short journeys; long, multi-channel journeys need multi-touch, data-driven, cross-channel methods.
• Don't trust the word "advanced" or "omni-channel" — describe a method by all five dimensions.
• Recognise that even custom/data-driven attribution can't yet capture context, so treat outputs as estimates and complement with judgement.
• Cross-platform and cross-channel attribution improve as cloud and account-synchronisation grow; offline value attribution is still evolving.

Exam Takeaways

Likely exam points

• Five dimensions: (1) single vs multi-touch, (2) fractional vs incremental, (3) cross-digital/platform/channel, (4) rule-based vs data-driven, (5) standardised vs custom.
• First-click / last-click = simplest single-touch; U-shape and time-decay are classic rule-based weightings.
• Single-touch is biased for long journeys (ignores timing, sequence, causality).
• Authors coined "standardised" and "custom" attribution to fix terminological overlap.
• Key limitation: no method systematically models the context of decision-making.

Summary

• Attribution = allocating conversion value across customer-journey touchpoints.
• The paper's contribution is a unifying five-dimensional taxonomy + map.
• Methods range from simple rule-based single-touch to complex custom data-driven multi-touch.
• Terminology is messy; context-awareness is the frontier attribution hasn't reached.

Related Notes

• Digital Marketing — subject hub
• 02b-Moorman et al - Why Marketers Are Returning to Traditional Advertising
• Marketing Attribution
• Customer Journey
• Purchase Funnel
• Multi-Touch Attribution