Attributing Conversions in a Multichannel Online Marketing Environment: An
Empirical Model and a Field experiment – H.A. Li & P.K. Kannan (2014)
The multiple touches a customer makes before a conversion are rarely taken into
account when measuring campaign effectiveness across multiple channels.
Example: online purchase scenario of a sample of customers going through the purchase
decision hierarchy.
Applying the metric commonly
used in practice – the last click
metric – to the data, the firm
would attribute 50% of the
conversion to direct channel and
25% each to display and search.
However, this last-click metric
ignores the prior channel touches.
Unless these prior channel
encounters have no impact on
current visits, ignoring such
spillovers could lead to biased
estimates of attribution.
Thus, aggregate metrics used in practice do not take into account the resulting carryover
and spillover effects, nor do they reflect these effects’ relative incremental impact in
leading to website visits and conversions.
This research: Research falls within the realm of multichannel marketing. Research is
related to studies that analyse the impact of individual channels outside the website (e.g.
display ads, e-mails, search engines) in enabling conversions at the website.
Focus: Individual level customer path data from a firm in the hospitality industry.
Result: There are significant carryover and spillover effects at both the visit stage and
purchase stage, the magnitude of which varies significantly across channels.
The model
The model focuses on the decision hierarchy in the context of online purchases of high-
involvement goods or services.
Empirical Model and a Field experiment – H.A. Li & P.K. Kannan (2014)
The multiple touches a customer makes before a conversion are rarely taken into
account when measuring campaign effectiveness across multiple channels.
Example: online purchase scenario of a sample of customers going through the purchase
decision hierarchy.
Applying the metric commonly
used in practice – the last click
metric – to the data, the firm
would attribute 50% of the
conversion to direct channel and
25% each to display and search.
However, this last-click metric
ignores the prior channel touches.
Unless these prior channel
encounters have no impact on
current visits, ignoring such
spillovers could lead to biased
estimates of attribution.
Thus, aggregate metrics used in practice do not take into account the resulting carryover
and spillover effects, nor do they reflect these effects’ relative incremental impact in
leading to website visits and conversions.
This research: Research falls within the realm of multichannel marketing. Research is
related to studies that analyse the impact of individual channels outside the website (e.g.
display ads, e-mails, search engines) in enabling conversions at the website.
Focus: Individual level customer path data from a firm in the hospitality industry.
Result: There are significant carryover and spillover effects at both the visit stage and
purchase stage, the magnitude of which varies significantly across channels.
The model
The model focuses on the decision hierarchy in the context of online purchases of high-
involvement goods or services.
