Untangling the many strands implicit in contact optimisation
This opinion paper from Berry Thompson tackles the question of what is contact optimisation, and how can an organisation set about delivering it.
The benefits of contact optimisation, when correctly undertaken, are huge in terms of improved ROI from customer communications (we have cases of 25%+ uplift), as well as improved customer experience and reduced opt-out.
But why are so few companies fully getting to grips with it when the technology is available, and the data is often there? Is it just that managing BAU leaves little desire for change?
We believe that contact optimisation is often seen as too complicated, and requiring too much resource to introduce, when in reality planning resource is much reduced by using the technology properly.
By following a logical methodology that progresses from deciding what stage in a relationship each customer is at, to selecting from a range of potential communications which one to use for each individual, and then applying rules about contact density, we can clearly understand the processes required, and get hold of the benefits.
At Berry Thompson we provide a full contact optimisation service; proof of concept, project management, decision engine technology, and insight; we have been working in this field for over a decade, and would like to share our experience with you.
What are we trying to achieve with contact optimisation?
Forrester once defined contact optimisation as follows:
Contact optimization applications work by processing inputs, including customer data, global business rules, contact policies, predictive model scores, business constraints, and objectives to identify optimal solutions.
As a statement this is helpful but somewhat circular as it fails to explain what we mean by ‘optimal’; it does however do a good job at describing the typical inputs that feed into a contact optimisation decision engine.
A few years ago we asked some experts* what they saw as the objectives for contact optimisation with the following results:
- The primary goal is that of improving ROI from a given communications budget
- However that needed to be balanced against other customer needs such as information or support
- And the ROI itself begs the question of short term returns versus longer term customer value
We would nowadays approach the objectives of contact optimisation in a slightly more customer focused way;
- To give each customer the mix of communications, over time, that brings the best mix of brand loyalty and sales to generate longer term customer value
* Journal of Database and Customer Strategy Management, Vol16, 4, 241-245
This is a very tall order, but in this paper we will make a start at explaining how we like to go about approaching it.
But before we start, may we make a key distinction between campaign targeting, and contact optimisation. Campaign targeting is solely concerned with selecting those customers or prospects most likely to respond to a particular proposition, delivered through a pre-selected channel.
It ignores questions such as whether the customer might have been more interested in something else if it had been offered, or whether the XX% who don’t respond are disenchanted by receiving the communication, or whether the channel in question is one that the customer likes to use. Being based purely on short term returns, campaign targeting is often very sub-optimal from our customer focused perspective.
So what are the benefits of contact optimisation?
The key benefits are around generating the most longer term customer value at least cost; this is achieved by
- Being able to give each customer the proposition they are most likely to respond well to
- Balancing communications that provide short term sales with those that address the wider relationship ( e.g. re-activation of dormant customers)
- Using the channel that each customer prefers
- Managing the overall density of communications that are sent to an individual
By achieving these things, customer propensity to opt-out from receiving communications is also going to be considerably reduced – an important factor given the forthcoming EC legislation.
Furthermore there are internal benefits to the business that can be achieved by using resources better:
- If there are capacity limits on a resource such as a call centre, or a print run of catalogues, then optimisation uses that limited resource to maximise take up
- Similarly with an overall budget constraint, optimisation manages the way that budget is deployed to maximise returns
How to make the actual what, when, how decisions?
There are a number of working parts in a contact optimisation decision process, all of which need to be brought together at the point of deciding ‘who gets what, when, how’.
The flowchart on the next page describes this, but at a very conceptual level.
We can describe it as follows:
1 Decide on the state of the overall customer relationship, and how this needs to be addressed:
- Are they a new customer needing the benefits of dealing with you to be explained to them?
- Are they a mature customer whose loyalty needs to be rewarded?
- Are they at risk of attrition, and requiring a very strong offer to bring them back into an active relationship?
- Have they reached a relationship trigger point like a policy renewal?
- Are they a ‘normal’ customer to whom we should make the cross-sales proposition for which short term gains are to be greatest?
Fig.1 The conceptual elements in a contact optimisation decision process
2 Pick the best proposition to make to a customer
Within a specific customer relationship category, such as being right for cross sales offers, we next determine which proposition, by which channel, is expected to be most effective.
For cross-sales, and other environments like anti-attrition, we normally use a formula:
where P represents the propensity of an individual to respond to the proposition, V is the longer term value of the response or purchase, and C is the cost of making the communication.
Using this approach each customer will receive at any moment in time the communication that provides the best overall return.
However in some environments, such as a policy renewal or other strong trigger events, we may assume that the communication is mandatory, and we only don’t send it if the customer has opted out in some way.
3 Manage the contact density that a customer receives
Contact density is frequently ignored by direct marketers, and is of the greatest importance. We know how response declines when offers are repeated at too short an interval, and equally ROI depends on getting the overall level of customer contact right for the level of the customer relationship.
Work we have done in the home shopping industry, based on live tests, showed how high demand groups may continue to respond well to frequent communications, whereas less attached customers required far fewer communications to reach their order peak.
Fig. 2 A chart showing how customers with different expected spend quintiles provide different cost to sales ratios from different levels of communications costs
Contact density rules can be set as absolutes (e.g. don’t communicate more than x times per month through y channel), and/or they can be managed through adjusting the P or propensity to respond according to the interval since the previous communication. The advantage of the latter approach is that higher propensity customers will naturally get more communications than lower propensity ones.
4 Impose overall business rules and constraints
In any organisation resources are finite, and hence, after all efforts have been made to answer the ‘who gets what, when, how’ question, we need to enter Darwinian territory.
With limited channel capacities, or communications budgets, we have to get the best return from the resources we have. This will mean ranking at any moment in time, all the potential communications opportunities for all customers, and selecting the most productive.
That may be at a channel level, such as for an outbound telemarketing centre with fixed capacity, or where there is an existing and fixed print run for a catalogue; equally it may be that a product manager has a finite budget to spend on email marketing and we have to pick the ‘best’ within that.
Or it may be at an overall communications plan level, where we know the budget for a month or a year, and are trying to spend it in the best possible way.
All of which implies that we have a common currency, longer term customer value, that we can use to arbitrate between alternative communication approaches. This becomes hard when we are, say, choosing between spending money on education, or early stage customer relationship development, versus sales. And these decisions are made at an individual communication level, not at a category level. In other words one customer’s development may be prioritised above another customer’s potential response to an up-sales opportunity.
However we are of the view that it is better to arbitrarily attribute a value to softer communications, so that they can compete full on with sales, rather than not attempting this at all.
Dealing with the on-line customer?
With the correct technology in place we can now interact with an on-line customer in real time, sending them offers and web pages that are personalised to their content interest; software packages like Idio are designed to fulfil this. We can call this content optimisation.
Additionally we can respond, not in real time, and send confirmatory or explanatory emails, or other communications, at a later point in time, when we can identify the browser on our database. A classic example is the dropped basket follow-up.
Both activities fall under the heading of contact optimisation, but there are significant differences between the two.
Deciding which best content to serve a browser, in real time, is normally done by putting in place a rule set based on previous browser history by that IP address; rarely are there propensity models in place to support this, but there may be a browsers’ segmentation that categorizes browsers into groups and allows them to be treated differently.
However dealing not in real time, with browsers who can be identified in our underlying database, and hence for whom there are alternative channels of communication, brings us back into the reach of contact optimisation as so far described in this paper; the additional information we have is the trigger or call to action derived from the browsing activity.
The way we deal with it may be separated from the main flow of arbitrating between potential communications, because the need for response is so great that the communication becomes mandatory, or it may be that the type of response is arbitrated from a range of alternative potential customer needs. In other words responding to the browsing trigger may be less important than making a new and unconnected offer.
Clearly web interactions with customers, and non-customer IP addresses, raises a whole new domain for contact optimisation, but we do not regard it as being outside the overall contact optimisation decision making process, rather it is an interesting extension to the conventional machinery.
Using contact optimisation in planning mode
There is a massive time dimension to communications planning, and, as we have seen above, contact density has a major impact on customer responsiveness and communications ROI.
This then immediately begs the question of how to plan forward over a series of time periods to get the best customer value generating experience from our communications.
The plain answer is that we cannot; customers’ needs and attitudes change through time, and the information we can glean from changing transactional, and even browsing patterns, will alter the way we want to communicate with customers.
However plans need to be made, budgets need to be set, and even resourcing needs to be managed if for instance we are to have the right number of staff in our call centre responding to calls.
What we can do is to start from the premise that even if at an individual level customers’ needs are constantly changing, when viewed in aggregate they remain reasonably static; the needs of a mother who has just given birth are very specific to her, but the number of births in a year is relatively constant at a population level.
Hence by using customer data as available today, a contact optimisation decision engine can still plan forward; this requires the engine to remember which communications each individual receives in time periods A, B, and C etc. and then using that to influence what happens in period D.
Clearly different times of year will have different seasonal requirements, and these need to be factored in, potentially by changing the ‘P’ scores in different months. Equally the business may have different sales targets at different times of year, which need to be allowed for.
Our experience is that organisations using a contact optimisation decision engine for planning purposes get a massive benefit, both in reducing the time taken to plan, but also through building much more sensitive plans that take account of contact density, and managing the customer experience through time.