China's 1.4 billion consumers—with their 1.6 billion smartphone subscriptions—expect and want personalised, one-to-one interactions with the brands they favour. But how can this level of personalisation be attained in a country with so many consumers? How do you build toward the elusive segment-of-one?
The challenge of attaining the segment-of-one
The ability to use available data to deliver the right offer at the right time to the right customer is the holy grail for any brand. Key to this type of personalised service is the capacity to acquire, store and process huge amounts of data at low cost.
China is ahead of the game when it comes to the latter, but needs to improve when it comes to market segmentation. Traditional segmentations are static, hard to update and incompatible with always-on consumers.
Consider, for instance, a typical example from the FMCG world. In one case we looked at recently, a global FMCG company in China was using segmentation dimensions such as ‘has travelled abroad’, ‘stay-at-home’, or ‘level of education’. While these segments made sense to the company, they were not scalable; other media teams didn’t have the same information within their targets. As a result, the segmentations couldn’t be used, and basic levels of personalised communication were impossible.
So what can brands do?
Starting segmentation with personas
One solution lies in creating fictional personas—customised segments based on the attributes and values of various consumer ‘types’. Such personas are highly specific to each business and can be as varied as ‘logo addict’ in luxury or ‘coupon clipper’ or ‘suburban mom’ in retail.
To develop these segments, brands just need to look at the data their consumers are leaving behind. The process begins by grouping consumers based on their apparent behaviours. Supermarkets, for example, might segment their consumers as ‘quick shoppers’ (people who value getting products efficiently and are focused on day-to-day needs), or ‘bargain hunters’ (people who react to promotions and low prices).
After grouping in this way, brands can then identify archetypal consumers, and feed samples into AI engines to test and learn from them over time. Crucially, the most effective AI engines also include ‘back doors’ which allow users to reclassify consumers and evolve personas as behaviours change (for example ‘bargain hunters’ could become ‘quick shoppers’).
Enriching segmentation with life stage and engagement
Another useful segmentation type is life-stage segmentation. This categorises people into segments according to where they are in their life cycles; this approach generally requires a combination of socio-demographic and transaction data collection.
For instance, a nutrition company might provide a service to track the weight of children and provide feedback on their progress. Over the years, it will be able to capture information on the age of the children and tag the family members accordingly, then target the families with relevant messages as the children get older.
Lastly, segmentation by engagement represents where the consumer stands in their decision journey. It is also called awareness-interest-purchase, or cold-warm-hot. Many leading ecosystems in China, working with players specialised in extracting data, now follow consumers along their journey and can target them specifically with ever more personalised messages.
Pushing towards the segment-of-one
To achieve a segment-of-one targeting then, marketers should combine a top-down approach that structures personas, life stage and engagement segmentations with a bottom-up approach that validates the structures and provides new dimensions (‘tags’) to test.
The key is to start small and set the right level of expectation: keep excitement low at the beginning and focus on delivering positive business impact. This will most certainly involve building a cross-functional team that combines marketing, sales, data and IT profiles and knows how to prioritise based on business needs.
With the team in place, marketers can then begin testing and learning, improving complexity gradually, measuring improvement, and knowing when to stop experimenting and scale to cash in results.
China may be ahead of the world when it comes to data abundance and maturity in machine learning and AI, but most companies there are still trying to crack personalisation. While it will require extensive testing and learning iterations to reach the ultimate segment-of-one objective, companies that have already started on this journey are enjoying a strong competitive advantage.
Xavier Mussard is partner with Artefact Asia.