Multichannel Online Marketing and attribution models
This blog synthesises research on managing the complexity of multichannel online marketing environments and understanding the dynamics of customer engagement. The proliferation of online channels including search engine marketing, display ads, email, and social media challenges retailers to interpret complex customer journeys. Researchers examine how various touchpoints, such as search engines, social media, and email, interact to drive website traffic and final sales. A significant focus is placed on attribution modelling, which seeks to move beyond simple "last-click" metrics to more accurately reward every interaction that contributes to a conversion. Some studies highlight the necessity of accounting for competition, noting that ignoring a rival's advertising can lead to significant errors in budget allocation.
Research suggests that classifying these channels is essential for reducing analytical complexity.
Channels can be classified along two primary dimensions to better infer consumer decision-making progress:
1. Contact Origin:
• Firm-initiated Channels (FICs): The advertiser determines the timing and exposure (e.g., Display, Retargeting, Affiliate, Email).
• Customer-initiated Channels (CICs): Potential customers trigger the communication through active search or direct action (e.g., Direct Type-In, Branded Search, Generic Search, Price Comparison).
2. Brand Usage (for CICs):
• Branded: The customer uses the retailer’s brand name to initiate contact (e.g., Direct Type-In, Branded Search).
• Generic: The customer uses unbranded inputs or third-party platforms (e.g., Generic Search, Price Comparison).
Customer Journey Dynamics and Purchase Propensity
Understanding the "Path to Purchase" requires analysing the sequence of contacts across different channel types. These transitions reflect changes in a consumer's "choice set" from awareness to active consideration.
Key Interaction Effects
• FIC to CIC Transition: A potential customer who first visits via a firm-initiated channel (e.g., a display ad) and returns via a customer-initiated channel (e.g., a search) demonstrates a narrowed choice set and a significantly increased purchase probability.
• Generic to Branded CIC Transition: Moving from generic search to branded search indicates the inclusion of the brand in the consideration set and a diminished time to purchase.
• Branded to Generic CIC Transition: This switch suggests a customer is actively seeking alternatives, which indicates a decrease in purchase probability for the original brand.
Within-Group Repetition: Repeated clicks within the same channel group (e.g., multiple display ad clicks) do not typically evoke a change in purchase probability through channel interactions.
Analysing channels in isolation leads to erroneous strategic conclusions. Digital marketing managers must abandon "last-click" heuristics in favour of sequence-based attribution. This approach accounts for the "carryover and spillover effects" where early FIC contacts facilitate the effectiveness of later CIC visits. Understanding the order of contacts is the only way to predict purchase propensity accurately. Once the path to purchase narrows the customer to the branded domain, the strategic focus must shift from macro-channel movement to micro-visual engagement.
Recommendations for online multi-channel management and attribution models
1. Moving Beyond "Last-Click" and Silo Thinking
The traditional attribution metrics like "last-click" provide biased and misleading data for decision-making. Managers relying on these metrics tend to overestimate the value of search channels while significantly undervaluing display, email, and referral channels.
• Budget Reallocation: Accurate attribution models can result in attribution changes ranging from -40% to +75% compared to last-click models, necessitating a fundamental shift in how marketing budgets are distributed.
• Holistic View: Managers must adopt a more inclusive approach that examines the interdependencies between channels, rather than measuring the isolated success of each.
2. Understanding the Drivers of Direct Traffic
Managers often observe that direct traffic has the highest conversion rates, but they should not view this traffic in a vacuum.
• The Indirect Effect: Research shows that traffic from Google paid ads, price comparison engines, and email can explain 61% of the variance in direct traffic purchases.
• First-Time Visits: Over 90% of users make their initial visit through non-direct channels, suggesting that direct traffic is often the result of earlier exposure to advertisements via other platforms. Managers should therefore credit these "assisting" channels for their role in driving future direct visits.
3. Incorporating Competitive Intelligence
Failing to account for competitor interactions leads to biased estimates of advertising effectiveness.
• Underestimation of Display/Referrals: In competitive environments, the relative contribution of display and referral channels is underestimated by a factor of two on average in single-firm models. This occurs because these channels often counter the advertising effects of rivals, a dynamic that single-firm models cannot see.
• Benchmarking: Firms should use competitive analytics to benchmark their ad effectiveness against the pack and diagnostic their messaging and copy strategies accordingly.
4. Bidding on the "Awareness" Stage
The traditional buying funnel assumes consumers move sequentially from broad to narrow queries, but empirical data suggests managers should re-evaluate this bidding strategy.
• High Performance of Broad Queries: "Awareness" queries (broad, non-branded terms) often generate more sales revenue at a lower cost than specific "Purchase" queries.
• Targeting Strategy: Managers are advised not to ignore Awareness key phrases; including these relatively generic terms alongside "long-tail" purchase terms can save costs while generating higher revenue.
5. Optimising Retargeting and Targeting Policies
While retargeting is a common industry practice, managers should be cautious about its implementation.
• Selective Intervention: Retargeting can sometimes hurt purchase probability by annoying customers. Instead, managers should use path analysis to identify specific customer segments where retargeting is most likely to increase conversion.
• Economic Value: Identifying the right customer to target with an intervention (like an extra email) can yield significant revenue increases, as the advantage of targeting is often contingent upon the specific path a customer has taken to the site.
References
Anderl, E., Schumann, J.H. and Kunz, W., 2016. Helping firms reduce complexity in multichannel online data: A new taxonomy-based approach for customer journeys. Journal of Retailing, 92(2), pp.185-203.
Cai, Y.J. and Choi, T.M., 2023. Omni-channel marketing strategy in the digital platform era. Journal of Business Research, 168, p.114197.
Filippou, G., Georgiadis, A.G. and Jha, A.K., 2024. Establishing the link: Does web traffic from various marketing channels influence direct traffic source purchases?. Marketing Letters, 35(1), pp.59-71.
Kakalejčík, L., Bucko, J. and Danko, J., 2020. Impact of direct traffic effect on online sales. Journal of Research in Interactive Marketing, 14(1), pp.17-32.
Kannan, P.K., Reinartz, W. and Verhoef, P.C., 2016. The path to purchase and attribution modeling: Introduction to special section. International Journal of Research in Marketing, 33(3), pp.449-456.
Li, H. and Kannan, P.K., 2014. Attributing conversions in a multichannel online marketing environment: An empirical model and a field experiment. Journal of marketing research, 51(1), pp.40-56.
Shankar, V. and Kushwaha, T., 2021. Omnichannel marketing: Are cross-channel effects symmetric?. International Journal of Research in Marketing, 38(2), pp.290-310.
