Automate the journey, preserve the relationship

How do you make member support more efficient without losing the personal connection people pay for? For me, the opportunity is to give members more ways to get things done, while making personal support easy to reach when their circumstances call for it.

Working with a health and fitness client, Enfuse helped develop automated self-service journeys and a central, member-facing chatbot. Our team led process design, requirements and product delivery alongside the teams building the technology. The ambition was to make the app a single front door for support without losing the personal attention members valued.

Long queues and capable teams working beyond sustainable capacity make a compelling case for automation. But removing contact is not the same as removing the need for care. The challenge is to make routine tasks easier while preserving choice, judgement and responsibility when circumstances demand more. So how can you make that happen?

Give members a choice that works for them.

We helped give members a range of ways to self-serve. For straightforward tasks, that convenience is valuable. But members may already recognise that their situation is unusual and that a conversation with someone is the most useful starting point. A premium experience needs room for those different preferences and circumstances.

Together, we developed a more flexible approach: here is how to reach the right team, but we may also be able to resolve this here. With this philosophy, chatbot engagement has grown steadily over the twelve months since launch. Other factors may also have contributed, so I would not attribute that growth to one decision alone, but the lesson is that offering personal support and encouraging digital adoption does not have to mean significant trade-offs.

Gartner reported in 2024 that 64% of surveyed customers would prefer companies not to use AI for customer service. Difficulty reaching a person was their leading concern. That is not proof that every automated experience is unwelcome; it is a reason to make the value of automation clear while reassuring people that help is available.

Recognise the person, not just the request.

Consider a member who wants to suspend their membership. A routine request may be best handled through a clear self-service flow, surfaced by a chatbot through a direct link. The same request following bereavement may need personal attention. The transaction has not changed, but the situation has.

We configured a large language model to perform automated triage: interpret the request, recognise sensitive circumstances and select an appropriate response or route. Historical interactions and business specialists’ expertise informed that judgement. This meant we modelled part of an experienced agent’s reasoning, instead of simply triaging through loose classifications and keyword matching.

Live integrations supplied factual context: the relevant booking, subscription or existing request. Knowing someone’s account details is not the same as knowing how to help them; useful personalisation needs both. Historical conversations also shaped the chatbot’s tone, preserving clear explanations and acknowledgement of the member’s experience. The aim was to reflect the brand’s service style, not disguise automation as a person.

A handover needs context and someone who can act.

Automation will not remove escalation, and the moment a member moves between channels is often where the experience is most vulnerable. Salesforce research found that 85% of customers expected consistent interactions across departments regardless of contact entry point. To a member, an app conversation followed by contact with a support team is one relationship. Internal boundaries should not become their problem.

Our team designed chatbot handovers to pass an AI-generated conversation summary alongside relevant member information. The receiving colleague could focus on helping rather than reconstructing the history. A useful handover lets the member continue the conversation, rather than start again.

But information alone is not enough. Feedback illustrates why. Some comments need to reach the relevant team or location; others describe an experience that needs putting right. Recognising that difference helps the business respond with the attention the member needs.

The receiving colleague needs authority to put things right or involve someone who can. The first person cannot always fix everything, but someone must own the outcome and keep the member informed.

For the member, that ownership should be visible: who is helping, what happens next and whether they need to do anything else. Behind the scenes, teams need an agreed way to work together. Passing a request between departments should move it towards resolution, not simply move it off someone’s list.

Decide how the business will deliver the promise.

This is where automation becomes an operating-model decision, not just a product feature. It changes where work enters, who handles it, what they can decide and who remains responsible across teams. Those choices need to be made by the business, not left to the technology. Service design helps connect the two: testing not only what appears on screen, but what colleagues can realistically deliver behind it.

For a gym group, consider the balance between consistent membership policies and local discretion. If a member’s circumstances do not fit the standard rule, can they be helped through standard self-serve journeys? Do they need a local team? Or must a specialist decide how to proceed? Clear boundaries give colleagues confidence to respond appropriately.

One of my clearest delivery lessons is that process understanding and high-quality data are not optional preparation. They determine what the experience can reliably deliver. Establish which information to trust, who maintains it and who may use it, whether building a chatbot or self-service flow.

Frequent requests, clear rules, dependable information and repetitive effort can make a strong automation case. But alongside “what could we save?”, always ask “what are we unwilling to compromise?”. That could be execution accuracy, so some journeys need simplifying first. It could be a requirement for human approval that necessitates partial automation. Or it could be preserving a concierge-style experience that lets customers determine their own contact journey.

Those choices also shape the support colleagues need to provide. Where human judgement or personal attention matters, people need the skills, time and authority to help, not just training on a tool. Involving them in the design helps preserve their expertise, while clear guidance gives them confidence to step in when an automated recommendation is not appropriate.

That means planning for the work that remains, not simply counting the contacts removed. Fewer routine questions may leave a greater proportion of complex or sensitive cases, each requiring more attention. Released capacity can create space for that care, provided expectations and performance measures reflect the outcomes colleagues are being asked to deliver, rather than speed alone.

Learn from the experience, not just the dashboard.

Technical performance and service targets matter, but an automated step can complete successfully while the member remains confused. User acceptance testing, pilots and post-launch feedback should examine the experience beyond the straightforward case.

Alongside product reporting, we developed bespoke measures, including AI-assisted conversation assessments shaped by business specialists’ manual reviews. These help assess interaction quality and satisfaction, complementing direct member feedback rather than replacing it. A well-handled escalation can be a better outcome than an automated conversation that goes nowhere.

Someone must turn that insight into improvement: identify recurring friction, bring the relevant teams together, change policy, process, guidance or automation, then check the result. That makes reporting a practical way to keep improving the service.

The questions are practical: did the member achieve what they needed, repeat information unnecessarily or leave unsure what would happen next? Looking at those outcomes alongside technical measures helps distinguish a completed transaction from a helpful experience. It also keeps improvement focused on what members value, rather than adding features for their own sake.

Our role at Enfuse was to connect business expertise with processes, requirements and delivery decisions, working with the teams building the technology. Every implementation is bespoke. The platform names matter less than whether the organisation can deliver the experience it promises.

Choose what to do with the capacity released.

We helped a business reduce inbound manual contact by 20%. That creates a choice about how to use the capacity released. Depending on business goals, it can support greater efficiency, more attention for complex cases, or both.

Retaining capacity can help protect the personal service members value. Any change to team size needs to account for the remaining workload, so the business preserves the benefits automation was intended to create. The decision should follow what it wants to achieve and protect.

So before asking how much customer contact you can automate, ask something harder: can we recognise the member and their circumstances, preserve that understanding across the journey, give people the authority to act when judgement is needed and remain accountable for the outcome?

If the answer is yes, automation stops being simply a route to lower cost. It becomes a way to make the organisation easier to deal with and the relationship harder to leave.