Managing Director Laughlin Consultancy
Share this content

Targeted marketing: Do event triggers trump propensity models?

2nd Sep 2016
Managing Director Laughlin Consultancy
Share this content
Target

When I say the words ‘targeting your marketing’, what comes to mind? It seems, for many, these words prompt ideas about personalisation or sophisticated algorithms.

In today’s digital, mobile shopping world, we’ve all become used to varying levels of personalised communication. It can work well, when firms get it right. But it can also feel patronising or creepy when they don’t. I’m particularly interested in that second association, and the apparent assumption that targeted marketing is all about statistical models.

Data science

The growth in popularity of data science, and the use of statistics in marketing, has led to a number of benefits. More companies now A/B test their communication targeting and many have more robust attribution for effectiveness measurement.

But, alongside these benefits, a few assumptions appear to have crept in. One of those is that a propensity model is what is needed, to target the people most likely to respond. Certainly this is a step forward from subjective business rules, or targeting based on broad demographic segments. But is it the best weapon for marketers to use in their battle for customers?

Event triggers

In an oft-quoted soundbite, when asked what prime ministers fear most, Harold MacMillan answered “events, dear boy, events”. Should marketers think in the same way? At least they should be aware of the power of communicating relevantly at the right time (especially if their competitors are doing so already).

In recent years, much less has been written about the role of ‘events’ as triggers to target people with marketing. So, arriving like the proverbial London buses, it was encouraging to see two reports at once, recently published on this topic:

  1. In The new rules of engagement, the DMA shares research showing that relevancy for consumers (especially related to when they receive marketing) trumps personalisation. For instance, 40% were interested in a service that reminded them about upcoming birthdays and made relevant suggestions (just in time).
  2. In “Life events – the hidden marketing key to solving customer churn”, research from Royal Mail Data Services shows a growing appreciation of the importance of event triggers. Over 70% of marketers surveyed recognised the importance of them using such triggers to provide a reason to engage with customers (compared to only 33% last year).

Working for over 25 years in financial services, I’ve seen marketing practices change a great deal over that time. Whenever I've seen improvement it has been related to the realisation as to the role of triggers as well as models. Both when leading my own teams and when helping clients, the optimal targeting solution has often proved to be a relevant trigger, plus a model. In some industries, including general insurance, an event trigger (like ‘renewal date’) is so predictive of response that a model is not needed. 

In some industries, including general insurance, an event trigger (like ‘renewal date’) is so predictive of response that a model is not needed. 

So, what exactly do I mean by using the phrase ‘event trigger’? Well, in this context, ‘trigger' just means that the event identified should be a trigger to take action (normally in the form of relevant communications (marketing or service) related to the event.  The word ‘event’ in this context is worth dwelling upon a little longer, as it can take a few forms.

Types of events, that can be useful as triggers, include:

  • Major life events (birth, first job, marriage, home moving, children, divorce, retirement, death).
  • Product events (renewal date, preferential rate expiry, offer expiry, maturity).
  • Customer action events (large credit, large withdrawal, taking-out/changing/cancelling other products).
  • Major purchase events (car, property, investment, expensive travel).
  • Dissatisfaction events (complaint, reducing payment, going dormant, cancellation).
  • Plus, many others (esp. related to product innovations and ‘Internet of Things' in future).

Same, same but different

Why should identifying and acting on these events be more successful than using a robust statistical model?

Think for a moment about your own experience. Do you expect to be marketed, for products or services, just because other people ‘like you’ purchased? Given many propensity models are built with basic variables, like age, affluence and geography as the most predictive inputs, does that work for you? Are you just like others of your age and affluence who live in a similar neighbourhoods?

Consider the impact of either: (a) being marketed as a moderately affluent single man for a sports car aimed at your demographic; (b) being marketed on the event of recently having married with relevant products for preparing for your life together (perhaps a bigger car, joint account or Help to Buy ISA, etc).

Now, when working with clients to help them migrate to using event triggers, a few common objections are raised. Few disagree with the premise I have laid out so far, but many often tend to suggest it is a ‘counsel of perfection’ given their circumstances.

A common concern is volume. How could they generate sufficient triggers, to be comparable to the volume achieved by simply adjusting which deciles are used on a propensity model (filtering entire marketable audience)? It is true that volumes marketed normally reduce, but often with improved returns. Plus, when all the different types of events outlined above are considered, most businesses can generate a much larger number of event triggers than they expect (especially once options to also purchase data are explored).

That last point is where data providers come into the picture. Data brokers and data providers have matured as a market in recent years. A few charlatans have fallen by the wayside and the major players are now used to clients requiring robust evaluation and realistic contracts. I’ve advised a number of insight or analytics teams on designing tests to evaluate the value-add that external data could provide to their marketing. With the growing importance of event triggers, it's an option I’d encourage all marketers to explore.

Further reading: How life events can solve customer churn

You might also be interested in

Replies (0)

Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.