Every customer interaction leaves a footprint of customer’s likes, dislikes, attitudes and behaviour in form of data. Successful marketers are those who cannot only make use of these footprints to understand customer’s journey but can steer the journey to meet marketing objectives. With increased and sporadic customer touchpoints across online and offline platforms, marketers are facing an increased challenge in building a seamless customer journey. Data Management Platform or DMP is precisely the toolset that can assist marketers in achieving this task.
So what is Data Management Platform or DMP? As per MarTechtoday.com data management platform is software that houses audience and campaign data — yep, a data warehouse — from all kinds of information sources. In digital advertising, these sources include publisher’s websites and apps on which advertisers buy advertising. A DMP offers a central location for marketers to access and manage data like mobile identifiers and cookie IDs to create targeting segments for their digital advertising campaigns.
Now the purpose of this article is not to delve deeper into the capabilities of DMPs. The purpose is how the success of Data Management Platform in any organisation would rely on its capability in delivering in 3 primary areas: Improving Reach, Optimising Media Spends and Personalisation. Before deep diving into each of these areas individually I would like to call out on how these marketing aspects function in a customer’s buying journey. I am taking help of AIDA model that stands for Attention, Interest, Desire and Action for this purpose. The AIDA model is widely used in marketing and advertising to describe the steps or stages that occur from the time when a consumer first becomes aware of a product or brand through to when the consumer trials a product or makes a purchase decision.
Improving the reach of a brand or products would happen in the awareness phase of the AIDA model. During this phase, marketers try to cosy up with the audiences who have never been with the brand before or to introduce the brand’s existing customers with new offerings. Data Management Platforms or DMPs have a marketplace of 3rd Party data sources that can help marketers buy audiences and their corresponding behavioural or attitudinal data for improving reach. Using DMP’s ‘Lookalike’ modelling feature marketers can create segments of audiences from 3rd party data sources that look similar to their existing audiences. DMPs can assist marketers to measure the effectiveness of awareness campaigns in meeting its objectives for these audience segments.
For creating a use case in this article I would take help of popular automobile brand Honda because I drive a Honda car. So in our situation, the use case could be “To generate awareness of Honda’s upcoming Amaze facelift”. It is always best to quantify what success looks like. It can increase traffic to the new car’s content pages by x%.
Media Optimisation would happen in most of the phases of the AIDA model provided the customers, a marketer is trying to reach, is reachable on the desired marketing technology. For example from his/her behaviour, we know that a prospect is susceptible to respond to an email. However, if we don’t have his/her email id we cannot reach out via email. In such cases, we would need to see if we can reach out to the prospect via another medium.
Now the overall process of achieving media optimisation can be divided into 5 steps:
Step 1: Segment the media campaigns based on products
The first step would be creating segments of products. This can be created based on the features/category of the product. For example for Honda the category would be SUV like Honda CR-V, Sedan like Honda City or Hatchback like Honda Jazz & features could be car segments like a compact sedan (Honda Amaze), mid-size sedan (Honda City) and premium sedan (Honda Accord).
At this step, the use case can be increasing online leads for a compact SUV like Honda WR-V by 2%.
Step 2: Segment users based on attitude and/or behaviour
Once the product definitions/segments are in place the next step would be to segment the customers the marketers intend to reach out. This can be based on customer’s demographic attributes like location (customers from Delhi/NCR) or age & income (Millennials who work with a Technology company) or lifestyle attributes (Road enthusiasts, own a compact sedan for last 5 years, etc). The segments can also be based on behaviour (those who have searched on compact SUVs in last 1 month). In most cases, the segments would not be clean and would involve overlap. A DMP can help marketers to gain these insights into these segments and really understand their customers better.
Step 3: Understand the buying phase of Target User Segments
The next step in the process would be to map product segments from step 1 to customer segments from step 2. This will give us a clear picture of our target customers and help us quantify how we are performing for a customer segment.
Extending the car use case from Step 1 “Increase online leads for Honda WR-V by targeting Millennials who are at middle management level with and own a small size car for the last 5 years”.
One caveat in creating customer segments is that we always have a trade-off in reach vs accuracy with DMPs. The finer definition you create the smaller would the user segment be and hence the response could be low. The base size of customer segment for each product category would vary depending upon the type of the product.
After identifying the target segment it is better to define at what stage of buying journey are the customers. We need to have a separate message and medium for the customers who are in interest phase compared to the customers who are in desire phase of the buying journey. This would essentially entail segmenting the target customer segments into sub-segments based on their recent behaviour.
Step 4: Map target audiences with marketing campaigns
So now we have a good understanding of our product segments, target customer segments and customer sub-segments based their buying stage. The next step is to map these customer sub-segments with marketing campaigns being executed for the respective buying stage. We all know that different marketing mediums have different response rates based on the stage of AIDA. Hence the icing on the cake here would be to have sub-segment of customers based on their behaviour to a specific marketing medium.
For example, certain customers are okay with receiving a call from customer care while certain others respond better to email campaigns. Every brand may not have this information initially but it can be built over a period of time-based using a DMP. I would like to finally re-iterate that there is always a trade-off in reach vs accuracy while creating sub sub-segments of users.
Step 5: Build your ideal media journey
With continuous use of DMPs and garnering insights of customer behaviour and attitudes for various segments and buying stages, we can create an ideal media journey for every identified use case. This would help us getting insights into media timing, its sequence and the frequency. It would ultimately lead to reducing the loss in media spends by employing tactics like impression suppression or frequency capping.
While Reach and Media Optimisation may be limited to certain phases of AIDA model Personalisation spans across the phases. However, the scope of personalisation would increase as the user traverses through the various phases of AIDA. The idea is to learn from customer’s behaviour and attitudes with every interaction. Also when the users are in the initial stages of the AIDA model they would experience a brand on any of owned, earned & paid media. Hence personalisation would have a mix of content as well as ad personalisation. However, once the user reaches the lower funnel personalisation would mostly be content and experience personalisation. Fortunately, DMP provides support in both the situation by providing customer insights for aiding personalisation.
Data Privacy & GDPR
Although GDPR is a whole different mammoth and a topic of a different discussion I would like to briefly touch upon it in this article since data and privacy go hand in hand. Now my belief is that DMPs are aggregators of information and hence the approval of how the data would be used is with the original data services collecting the data. However, we can’t wash hands just like that. So DMPs usually provide Data export controls to ensure that data does not violate the privacy regulation enforced on underlying data sources.
The use cases I mentioned were relatively straightforward. However, the complexity can increase based on industry verticals. So for a consumer durables company like LG, the use cases can be tricky as same users can be a target for multiple product offerings (I can be targeted for microwave, refrigerator or a mobile phone). The brands should always keep in mind not to intrude into personal space by bombing multiple marketing messages. Hence the success of data would rely not only on the tools but the team working on those tools.
Try deep learning using MATLAB