A credible contact centre AI business case starts with the customer problem, not the tool. Baseline what contacts cost and why customers call, count governance and change costs from day one, and prove value in a focused pilot before you scale. Most UK firms using AI report modest gains so far, so boards want evidence rather than promises.
That last point is now on the record. The government’s AI Adoption Survey 2026, published on GOV.UK on 29 October, found that a third of UK businesses now use at least one AI technology, double the 2025 figure. It also found that the gains are mostly incremental, and that a lack of a clear business case or proven return is one of the main things holding firms back. For contact centre leaders in London and the South East, who are being asked to do more with the same teams, the question has shifted from “should we use AI?” to “how do we show it is working?”
What does the latest UK data say about AI returns?
Two government sources published this year tell a consistent story: adoption is widening faster than it is deepening.
- Adoption has doubled. In the AI Adoption Survey 2026, 32% of UK businesses were using at least one AI technology, up from 16% in 2025. Larger firms were ahead, at 52%.
- The gains are real but modest. Half of adopters (50%) reported an overall rise in productivity, but the survey describes these gains as generally modest. On costs, 24% reported a reduction, 64% had seen no immediate financial effect and 9% reported an increase.
- The business case is the sticking point. Among businesses not yet using AI, 65% cited the lack of a clear business case or proven return on investment.
- Headcount has barely moved. The ONS article Artificial intelligence in UK businesses: 2023 to 2026 found that AI use among firms with 10 or more employees rose from around 12% to around 35% since late 2023, but that this has not yet translated into widespread changes in overall workforce headcount.
None of this says AI fails. It says most organisations have bought the tools faster than they have built the measurement around them. That gap is where a contact centre business case either earns trust or loses it.
Why do so many contact centre AI business cases fall short?
The CCMA’s good practice guide, Delivering AI Project Success in the Contact Centre, names the most common mistake plainly: choosing an AI solution before understanding the problem. Three patterns follow from it.
- The case counts licences, not change. The guide warns that projects which account only for technology licensing will fail. Training time, knowledge clean-up, integration work and ongoing optimisation are real costs.
- There is no baseline. If you cannot say what a contact costs today, how long it takes and how often customers come back, you cannot prove anything has improved.
- Deployment is mistaken for adoption. A tool that is switched on but quietly worked around by advisers delivers nothing. The guide suggests treating adoption data as a live performance measure.
How do you build a contact centre AI business case a board will back?
Senior leaders across the capital tend to face the same audience: a finance director who has seen several technology promises already this year. A case that holds up usually has five parts.
1. Start with why customers contact you
Intent analysis comes first. Look at the real reasons behind contact, by volume and by value, and be honest about which ones should never be automated. Complaints, bereavement, financial difficulty and anything involving a vulnerable customer usually belong with a person, whatever the technology can do.
2. Set a baseline before you pick a tool
Agree the starting numbers with finance: cost per contact, handling time, repeat contact rate, transfer rate, adviser attrition and customer effort or satisfaction. Track them before, during and after selection so the comparison is clean.
3. Cost the whole change, not just the software
Build in governance, integration with CRM and knowledge systems, training time that is protected rather than squeezed, and a planned dip in productivity while teams adjust. The CCMA guide suggests allowing an embedding period of six to eight weeks, and at least six months for adoption to become sustainable.
4. Define success in the board’s language
Pick a small number of outcomes that matter commercially: capacity released without new hiring, fewer repeat contacts, faster resolution, lower attrition or better retention of customers at risk. Agree in advance what result would justify scaling and what would stop the project.
5. Pilot narrowly, then scale on evidence
Choose one or two contact types with clear volume and low risk, run a focused pilot, and publish the results internally, good or bad. Evidence from your own operation will carry more weight with the board than any vendor case study.
What should you measure once AI is live?
Containment or deflection rates on their own can flatter a project. A customer who gives up on a bot and rings back the next day has been “contained” once and cost you twice. A balanced scorecard is more useful:
- Resolution: did the customer’s issue get fixed, in any channel, without a repeat contact?
- Effort: did the journey get easier, measured through customer effort or satisfaction scores?
- Escalation quality: when AI hands over to an adviser, does the context travel with the customer?
- Adviser experience: are advisers using the tools, and has the work they handle become more manageable or more draining?
- Cost and capacity: what has changed in cost per contact and in hours released for higher-value work?
Oversight belongs on the scorecard too. The AI Adoption Survey found that 81% of businesses apply some level of human input or oversight to AI-generated outputs. In regulated sectors such as financial services, that oversight is part of the case, not an extra. Our earlier piece on AI governance and the FCA’s good practice covers what that looks like in practice.
Where London leaders are working through this together
At the launch of London Contact Centre Forum (LCCF), part of the Contact Centre Forum network, leaders kept raising the same challenge: improve service and manage costs at the same time. That conversation shaped the Forum’s first event since launch, Making CX Work: Connecting People, Process and AI, on Thursday 26 November 2026 in a private box at Twickenham Stadium. Its three member-chosen themes are growing without growing headcount, introducing AI people can trust, and proving the value of customer operations.
On Thursday 10 December 2026, the Operational Excellence 2027 Challenges Workshop runs from 10am to 1pm at Zoom London HQ, 71 High Holborn, for leaders who want to work through practical challenges with peers before budgets are set.
Beyond events, LCCF brings the experience of Jane Thomas and the wider Contact Centre Forum network to London: Forum on Tour site visits, monthly user groups for operations, planning, training and HR, and the Friday Update, a 30-minute weekly conversation with colleagues across the network. It is a place to compare notes with people who have already made the case, and with those who learned from a pilot that did not go to plan.
Frequently asked questions
What should a contact centre AI business case include?
It should include the customer problem you are solving, baseline metrics agreed with finance, the full cost of change including governance, integration and training, clear success criteria, and a pilot plan. Cases that only count software licences tend to overpromise and struggle to show a return once the tool is live.
How long does it take to see a return from AI in a contact centre?
It depends on the use case, but expect a settling-in period before results are reliable. CCMA guidance suggests six to eight weeks to embed a new tool and at least six months for adoption to become sustainable. Plan for a short productivity dip and measure over a realistic window.
Is containment rate a good measure of AI success?
Not on its own. Containment shows how many contacts stayed with the AI, not whether the customer’s issue was solved. Pair it with repeat contact rate, resolution, customer effort and the quality of handovers to advisers, so a high containment figure does not hide customers coming back through another channel.
How many UK businesses are using AI?
The AI Adoption Survey 2026 found that 32% of UK businesses used at least one AI technology, up from 16% in 2025. The ONS puts the figure at around 35% for firms with 10 or more employees. Both sources suggest the impact so far has been limited.
Join the conversation in London
If you lead customer operations, CX or contact centre transformation in London or the South East, we would love you to join us. Request an invitation to an upcoming LCCF event, or talk to Jane about membership and what it could offer your whole organisation.