Ali, you’ve worked in CRM for nearly 20 years and say the objective has always been broadly the same – maximising customer lifetime value. If we’ve known the goal for so long, why is it still so difficult to achieve?
Because gambling CRM is incredibly complex. Most operators have multiple products, and those products are constantly changing. In sports, the content changes every day. New casino games are launched all the time. Customer preferences shift, and entirely new behaviours emerge – crash games are a good example of something that became hugely important relatively quickly.
The challenge isn’t just understanding all of those signals. It is deciding what to do with them.
At the same time, CRM teams have commercial targets, competitors launching promotions, internal stakeholders wanting particular campaigns pushed and different parts of the business competing for the customer’s attention. So while everyone talks about maximising lifetime value, it is very easy for the organisation to optimise for the campaign, the journey or the immediate commercial requirement instead.
Historically, it has also been extremely difficult to build a genuinely dynamic 360-degree understanding of an individual customer and then use that understanding to continuously optimise what happens next.
Operators now have onboarding, early-life, retention, churn and reactivation journeys, supported by increasingly sophisticated models. Has adding more journeys and more intelligence created another problem – fragmentation?
Absolutely. I’ve seen a huge improvement in the sophistication of CRM over the years. Automated journeys are better, segmentation is better and increasingly sophisticated models sit behind them.
But the result is often a huge collection of journeys covering onboarding, early life, retention, cross-sell, upsell, churn prevention, reactivation and everything in between. Each journey can have different models, different triggers, different segments and predetermined campaign flows. So the individual components are becoming more intelligent, but the overall system can become incredibly complicated and fragmented.
The problem becomes: who is looking across all of those journeys and deciding what is actually best for the customer? There is also a significant operational overhead. CRM teams spend enormous amounts of time building, maintaining and optimising the machinery rather than thinking about the experience they ultimately want to create.
A customer can generate signals relevant to several journeys at once, while the operator has different commercial objectives competing for attention. How does traditional CRM decide which one should take priority?
Usually the journey does. Traditional CRM tends to divide the customer lifecycle into different stages and create criteria determining when somebody enters a particular journey.
Take a customer who hasn’t played for a few weeks and then suddenly returns. They may immediately enter a reactivation or re-onboarding journey, perhaps telling them about new features, games or promotions. But simply because somebody has returned doesn’t mean that is necessarily the right thing to do next. There may be another action that is far more relevant to that individual customer. That is the limitation of compartmentalising customers into journeys. The system is effectively saying: “You meet these criteria, therefore this journey applies.”
What we really want to ask is: “Given everything we currently know about this customer, what is the best thing to do next?”
Is that why you believe operators need to move away from thinking primarily about customer journeys and instead build a dynamic understanding of each individual player?
Yes. Journeys are useful organisational constructs, but customers don’t behave like journeys. A customer can simultaneously be high value, showing signs of churn, interested in a new product, changing their playing behaviour and becoming more or less responsive to certain types of promotion. Those things don’t happen neatly one after another.
So instead of deciding which box the customer currently belongs in, the better approach is to continuously understand that individual and decide what action makes most sense given their current context. That is a very different way of thinking about CRM.
What does “dynamic understanding” mean in practice, and how is it different from the single customer view or increasingly granular segmentation operators already use?
A single customer view is important, but having the data and knowing what to do with it are two different things. Traditionally, a single customer view might tell you the customer’s value, the games they play, their deposits, profit and loss, frequency of play and lots of other attributes. Those fields may change over time, but the way they are used is often relatively static. They feed segments, models or predetermined journeys.
A dynamic understanding means continuously looking at those signals in combination and asking how a change in behaviour affects what we believe about that customer’s future value and what we should do next. The important part isn’t simply having the 360-degree customer view. It is having a dynamic decisioning capability sitting on top of it.
If one thing changes, what does that mean for everything else we know about the customer, and does it change the action we should take? That is where it becomes much more powerful than increasingly granular segmentation.
Once you have that understanding, how does optimising for lifetime value change the next-best-action decision compared with optimising for an immediate response, deposit or engagement?
It changes the time horizon. If you optimise for an immediate deposit or click, the obvious decision can often be to communicate now or put an incentive in front of somebody. But that doesn’t necessarily maximise the long-term relationship. There is a lot of discussion about real-time CRM in gambling, but actually a huge amount of good CRM isn’t about responding instantly. It is about understanding behavioural patterns.
Imagine a customer who normally plays blackjack or poker on a Friday night. Occasionally they lose a hand, stop playing and leave. You don’t necessarily need to immediately react and say, effectively, “We’ve noticed you’ve stopped – come back.” That may simply be completely normal behaviour for that customer.
The more useful signal is when they start deviating from their normal pattern. So there is absolutely a place for real-time intervention, but there is also enormous value in understanding historical behaviour, normal playing patterns and predicted future behaviour.
Optimising lifetime value means choosing the action – and the timing – that is most likely to improve the overall customer relationship, rather than simply maximising the next response.
One possible action is to do nothing. With so much emphasis on real-time CRM and hyper-personalisation, are operators in danger of assuming better technology should mean communicating more frequently rather than making better decisions?
Definitely. For years, there has been enormous pressure in CRM to communicate more.
How many customers have we contacted? How many campaigns have we sent? What are our competitors doing? How many times have we spoken to this customer this week? But that isn’t how people behave.
If you repeatedly send somebody irrelevant messages, they don’t necessarily unsubscribe. Often they simply stop paying attention. Once that happens, you’ve damaged the future value of that communication channel. I’ve heard the argument before that brands need to maximise their “mental availability” with customers. In CRM, I think relevance is much more important.
Relevance beats frequency. Sometimes the best possible action is doing nothing. Don’t spend money unnecessarily. Don’t send somebody an offer they don’t need. Don’t talk about something they have only a marginal interest in. Wait until there is something genuinely relevant to say.
For us, “do nothing” is a completely legitimate next-best action.
What is the Axom system learning about individual customers that the existing journey-based approach fails to comprehend?
The key difference is contextual learning over time. Axom builds an individual understanding of each customer and continuously learns from what happens. We observe what actions were taken, what happened afterwards, what tends to work for that person, what doesn’t work, how their behaviour changes and how all of those things interact over time.
The longer we work with a customer base, the richer that context becomes and the better the decisioning becomes. It is similar to the reason AI assistants become more useful when they have context and memory. We’re applying that principle at an individual customer level for businesses.
Rather than repeatedly asking whether somebody currently qualifies for a particular journey, we’re continuously building an understanding of that person and using that context to determine the best next action. It gives us a depth of understanding of individual customer behaviour that I haven’t seen from traditional journey-based CRM.
If the decisioning becomes increasingly autonomous, what happens to the CRM team? Does its role shift from building and maintaining journeys towards creating better actions and experiences for the system to choose from?
I think that’s exactly where the opportunity is. CRM has already been moving towards automation for years. Models already trigger campaigns and decide which customers enter journeys, so autonomy itself isn’t really new. What changes is the level at which the optimisation happens.
Rather than CRM teams spending huge amounts of time maintaining journeys, changing rules and continually requesting another model from data science, they can spend more time thinking about the actual customer experience. If I were this customer, what would I find valuable? Could we create a better promotion? A new communication channel? A more interesting product experience? A different type of reward? It moves CRM away from maintaining machinery and towards creating better experiences and better actions for the system to choose from.
To me, that is a much more interesting use of a CRM team’s time and expertise.
Axom isn’t trying to replace the underlying CRM platform, but to provide an intelligence layer across it. Longer term, could that dynamic decisioning layer effectively become the central CRM brain, with traditional journeys and segmentation playing a much smaller role?
Yes, that’s very much how we see it developing. Axom is deliberately focused on decisioning. We don’t want to rebuild the infrastructure operators already have, and we don’t execute the campaigns ourselves. The operator keeps its existing CRM platform, communication channels, data infrastructure and actions. Axom sits across that ecosystem and decides the best action for each individual customer.
Longer term, I think that central decisioning brain becomes increasingly important because ultimately somebody – or something – needs to optimise all of those competing possibilities against the same objective: maximising net customer lifetime value. I do think traditional segmentation and journeys will therefore play a smaller role.
They won’t disappear completely. There will always be situations where the business knows something the system couldn’t reasonably anticipate – a new product launch, a major tournament, a regulatory communication or a particular campaign the operator explicitly wants to run. But rather than journeys being the centre of CRM and intelligence sitting around them, I think that relationship increasingly reverses. The central brain becomes the primary decision-maker, while journeys and campaigns become some of the tools available to it.