In a recent Dynata survey on customer loyalty, 74% of consumers indicated that feeling valued and understood both factor more into brand loyalty than across-the-board discounts and perks. Those terms (valued and understood) were defined as a customer knowing their worth to a brand extends beyond a transactional value, and that the brand demonstrates that it knows the customer’s individual preferences and behaviors vs. those of another customer.
Survey respondents also stated that they typically are truly loyal to fewer than three brands, a clear indication that meeting the expectation for a personal understanding is a prerequisite for attracting and keeping loyal customers.
Recommended: Latest Report: “A Cookieless World: The Shifting MarTech Landscape”
Receiving relevant product and service recommendations was cited by 52% of respondents as the No. 1 way they feel valued by a brand, with 44% attaching value to a brand’s offering seamless navigation between channels, specifically in-store and online.
While customers have strong feelings about what they want, they’re less concerned with the how. The onus is on the brand to provide a differentiated experience. And that is where machine learning comes into play. Because to truly understand a customer on an individual level, marketers simply can no longer rely on manual audience segmentation processes that fail on several fronts. First, by their arbitrary nature, marketer-driven segments at a macro level are simply educated guesses on what’s important about a customer.
- Does being a male between 18-34 accurately define a person’s interests?
- Second, manual segments are limited by human imagination. How many different reasons for differentiating an audience can a marketing team think of?
- And third, a list-based segment does not account for the dynamic nature of a customer journey.
A customer may be targeted for a campaign because she is in a segment of other women in her age group with similar style, color and size preferences. A relevant, personalized email offer quickly becomes irrelevant, however, if the customer purchases a product at full price while the email discount offer is sitting in her inbox.
Discover what Really Matters to a Customer
A personal understanding across a broad range of customers – at scale – is the job of automated machine learning. Offline, self-training machine learning models take out the guesswork by analyzing customer cohorts, single customer views, behavioral and transactional patterns and other data sets that provide a better understanding of customers at the individual level. Unlike manual rules for segmenting an audience, machine learning is able to analyze large data sets without needing an obvious correlation between what you’re trying to discover and the information in the data.
An unsupervised machine learning model is a common use case for audience segmentation precisely because it eliminates arbitrary cut-offs (males 18-34, etc.).
Instead, a model will segment an audience based only on what’s important in the data, ensuring that insights truly reflect what matters to a customer. Granular audience segmentation is made possible by running dozens or hundreds of models simultaneously; the more models, the more confidence marketers have in knowing precisely why a customer belongs to one segment vs another.
Keep Pace with Dynamic Customer Journeys
As to the problem of models decaying with time, especially as customer journeys become more dynamic, the solution is to run online operational models with a constant influx of new data. Traditionally, companies invest heavily in building offline models using static data, having to repeat the process whenever a current model becomes outdated – when there’s new customer data, market conditions change or when a business use case changes.
In-line, operational self-training models with 24/7 testing and built to handle a steady influx of new data can detect and incorporate changes into subsequent production models, ensuring that insights about an individual customer are always up to date. This can be done while also providing a degree of control to the marketer in that parameters may be used to constrain the models, e.g., the models need to deliver within a range of targeted business objectives or business rules. Because every customer action, every engagement, device, transaction, and behavior are accounted for on every channel and updated in real time, a brand can move with a customer through each journey stage.
Customers will likely not know or care that automated machine learning is the secret weapon behind every hyper-personalized experience and a brand’s ability to consistently demonstrate a deep, personal understanding. But when a next-best-action appears at the right time, on the right channel and in the cadence of a unique customer journey, customers will know that the brand takes the time and effort to understand them as an individual.
In the Dynata survey, 64% of consumers said that they would rather purchase a product from a brand that knows them, and – some good news for brands – one-third (32%) of customers said they will even overlook a single bad customer experience if they feel that a brand is trying to understand them as a customer.
Doing nothing, though, is not an option. Customers are clear – a brand must exhibit a consistent, personalized experience to earn loyalty.
Martech News: Martech SaaS Platform Qualifio Expands Vision for Zero-party Data and Marketing Analytics
[To share your insights with us, please write to sghosh@martechseries.com]
Comments are closed.