Why CX transformations fail to deliver measurable impact, even with significant AI investment

It's no secret that the pace of change in the CX world has accelerated at breakneck speed. I've been in the space for 5 years and in that time the meaning of "transforming CX" has shifted from collecting more data to personalization, from personalization to content operations, back to personalization, then to generative AI and now to agentic AI.

It feels a bit like being in the driver's seat of a Formula 1 car. You're at the wheel, going hundreds of miles per hour, cars zooming around you, and you must keep up or risk ending up in the barriers. But the rules keep changing.

As you can probably tell, I'm a big fan of F1. The speed and adrenaline are part of it of course, but what really strikes me is that it is, in every meaning of the word, a team sport. Ask any F1 fan what makes a championship-winning team and they'll tell you: it's never just the driver. It's the engine, the aerodynamics, the pit strategy, the tyre strategy, and much more.

So where is this going? AI is the Max Verstappen (or Lewis Hamilton, for those so inclined). Everyone wants these speedy drivers and believes bringing them on board will automatically fix a broken marketing stack or CX process. But that is precisely why these CX transformations fail to deliver: they are fielding a world-class driver with no team behind them.

4 reasons why CX transformations stall

The root causes tend to cluster around the same four problems.

First, the goalposts keep moving. There are no clear, stable definitions of what success looks like, and teams chase what's hot or new, driven by top-down directives or FOMO, without clear outcomes in mind.

Second, process and technology integrations are missing. Workflows aren't documented, data isn't connected at the right junctions, and tools live in silos. Individuals build workarounds like copy-paste flows between data sources, but these don't scale.

Third, nobody zooms out. Critically examining how work gets done, the processes, the skills, the roles and responsibilities, the handoffs, the baseline metrics, is unglamorous work. It's rarely what people are measured on or rewarded for. So, it frequently gets skipped.

Lastly, governance, training, and data activation are treated as afterthoughts. Without them, quality drifts, risk accumulates quietly, and generic or unchecked AI output starts reaching customers.

Introduce a star AI driver into that environment and you get a very expensive, very frustrated Max with nowhere to go.

The Championship Playbook for CX

The organizations pulling ahead are acting like F1 teams. Paying attention to the whole car, the whole strategy, the whole package. Four things in particular.

1. Process and goals: map the track before you race it

Before any agent goes live, the workflow it supports needs to be documented step by step, including key handoffs, tools, owners, and data flows. This surfaces where AI can meaningfully make a difference and, just as importantly, where it will amplify problems that already exist.

Process mapping is also the best way to set goals for your programme, because it grounds them in reality rather than aspiration. Stable, specific goals tied to measurable KPIs are what keep the programme on course, help communicate wins to leadership, and give you something concrete to course-correct against when things don't go to plan.

2. Governance: the race regulations

Every F1 team operates within a precise regulatory framework to keep competition safe, consistent, and credible. AI governance works the same way. Organizations need clarity on how AI is used, who approves it, and what the guardrails are: brand tone, legal compliance, data standards, ethical use.

Most mature organizations land on a federated model: a central team sets policy and builds organization-wide workflows, while individual teams deploy agents for day-to-day work within clearly defined guardrails. Consistency where it matters. Speed where it's needed.

3. People: practised, accountable, and ready

A pit stop looks effortless because every person has rehearsed their exact role thousands of times. AI transformation requires the same precision. Clear roles around who builds agents, who monitors quality, who escalates issues, and who owns the workflow are essential.

But 66% of CMOs say their teams lack AI skills (BCG), so role clarity must be paired with practical enablement: prompting frameworks, agent training, and guidance on how to review and supervise AI-generated content. The shift from doing the work to directing it is significant, and it doesn't happen on its own.

4. Domain-specific AI: built for this track

General-purpose copilots were built for broad productivity, but they struggle to enforce brand rules, connect to publishing workflows, or pass context across agents throughout the marketing lifecycle.

Marketing needs a platform built for marketing: where every agent automatically inherits tone standards, governance rules, and approved data sources, and where that intelligence is available where the work happens, the content management platform, the campaign workflow, the experimentation suite, the approval flow.

It's an F1 car, with an F1 team, where each member knows their role and they work together to make the marginal improvements that add up to something significant. That is what true CX transformation looks like.

In F1, as in marketing, it's the interconnectedness that makes the speed possible.

Source: Optimizely