2023-01-10 | Article | Insights
Since the outbreak of the Coronavirus, consumers have increasingly shifted from stationary (brick-and-mortar channels) to online retail. According to studies, online shopping is still (post-pandemic) preferred because it is convenient, allows social distancing, and offers a greater selection of products with detailed product information. The severity of the global dynamics resulted from the pandemic made it essential for retailers to understand the forces impacting shopping behavior with 40% of users being motivated by promotional offers to purchase goods from online channels as revealed by the International Omni-Channel Retail Report conducted by YouGov in 2021.¹
Although it is intuitively comprehensible that promotional offers or reduced prices, in general, motivate users to convert and buy, reducing prices naturally collides with the economic purpose of businesses, making it imperative to be taken with a grain of salt.
With Behavioral Couponing you can increase your couponing efficiency by using Artificial Intelligence modeling. Similarly to the more widely known behavioral targeting, Behavioral Couponing aims to interpret user signals to identify the affinity and intentions of users and then uses that knowledge to determine where and what type of a coupon will be most effective at motivating the user to buy. By personalizing your vouchers, you can reduce waste and increase ROI over conventional couponing practices. Vouchers constitute a large part of marketing budgets today, yet are not optimized the same way as other digital advertising. The following actions might be relevant to incentivize via a coupon:
With Google Cloud (GC) and Google Marketing Platform (GMP) combined, you can develop and implement your Behavioral Couponing strategy with ease. Both tools are designed to seamlessly connect with each other sharing available identifiers while preserving user privacy. Three key steps can be divided towards establishing your Behavioral Couponing using GC and GMP:
(1) Collect Behavior Signals: Record different data points along the user journey that indicate a user’s interest such as landing page/ user entry point, product detail page, viewed content and the like.
(2) Train AI Models: To calculate user coupon affinity the collected behavior signals are used within one or many machine learning models - either in real-time or with a batch processing method depending on your use case. Examples of ML models to optimize your coupon delivery strategy are Conversion Probability Prediction and Category Interest Prediction.
(3) Activate your Insights: User affinity for coupons can be used both on- and off-site for optimization. Onsite optimization measures may include discount adjustments, voucher placement as well as restrictions, whereas offsite optimization can involve adjusting the course of your remarketing strategy, building and addressing similar audiences to extend the effect of the gathered insights around coupon effectiveness to more potential clients as well as dynamic creative optimization to put the right creative including discount in front of the right user.
For step one of the above, a state-of-the-art tracking implementation on your website (and further properties) is essential. In the current stage of the industry, this entails using a server-side GTM implementation of Google Analytics 4 running inside Google Cloud on Cloud Run. By scaling automatically up and down, Cloud Run makes sure the server-side GTM always has enough resources to handle all incoming requests for both behavioral tracking and conversion tracking.
The data generated from GTM on Cloud Run can be passed into two databases inside Google Cloud: BigQuery and Firestore. Firestore’s near-real-time read-and-write capabilities make the database the perfect match for short-term user record data. With asynchronous variables recently released in server-side Google Tag Manager, data from GTM can be passed into this fast and scalable NoSQL database. To read data from Firestore, Google introduced a new Firestore Lookup variable in server-side GTM.
Long-term user record data can be exported to BigQuery, Google Cloud’s analytics database. In BigQuery user record data can be merged with offline data for a more complete picture of the available user touch points and therefore stronger insights to use for the Behavioral Couponing strategy.
BigQuery integrates smoothly with Vertex AI, Google Cloud's AI platform with ready-to-use pre-trained models (step two). This is where the offline model training takes place and where the real-time propensity predictions are made.
From Vertex AI, the coupon information is fed into and processed with Dataflow to enable category inference, discount selection, and a coupon budget check. Dataflow also has access to short-term user record data in Firestore. The processed data is returned to a Cloud Run service, which provides the optimal coupon for the current user state via an HTTP endpoint. The coupon can now be used in Optimize 360 or onsite with a direct integration (step three).
From BigQuery, data can be imported in Google Analytics to build custom segments for off-site personalization. These segments are available for remarketing campaigns in Google Ads, or DV360. For dynamic creatives, a Campaign Manager 360 account is needed.
Today, there is no doubt that the era of online shopping is in full swing which especially came about as a result of the lockdowns of the past couple of years. This may have overwhelmed your digital operations leading to spreading discounts on a scattergun principle to somewhat keep up with consumer demands. However, with the tools available today, you can strengthen your approach to couponing by adopting a behavior-based strategy for both, onsite and offsite personalization. The application of artificial intelligence, machine learning, and automation can serve you tools that can help your business keep up with the shifting trends in consumer expectations while preserving your desirable margins enabling business continuity.
¹ INTERNATIONAL OMNI-CHANNEL RETAIL REPORT 2021, Shopping in the pandemic and the implications for the future, yougov.com/retail