Data-Driven Customer Experience: Stop Guessing, Start Acting

Data-driven customer experience means using behavioural, transactional, and operational data to make deliberate decisions about how customers interact with your brand, rather than relying on assumptions about what they want. Done well, it closes the gap between what businesses think customers experience and what customers actually experience.

The gap is wider than most organisations admit. I’ve sat in enough client workshops to know that when a leadership team says “we know our customers”, what they usually mean is “we have a CRM and we ran a survey two years ago.” That’s not insight. That’s comfort.

Why Most Data-Driven CX Efforts Underdeliver

There’s a pattern I’ve seen repeat itself across industries. A business invests in a data platform, integrates a few sources, and then waits for the insights to arrive. The insights don’t arrive. What arrives instead is a dashboard that nobody looks at and a quarterly report that confirms what everyone already suspected.

The problem isn’t the data. It’s the absence of a clear question. When I was running iProspect and we were scaling from around 20 people to over 100, one of the things that kept us commercially sharp was insisting that every piece of analysis started with a business question, not a data source. What decision are we trying to make? What would change if we knew the answer? If nobody could answer those questions, the analysis didn’t get commissioned.

Most CX data programmes don’t start there. They start with “what data do we have?” and then reverse-engineer a narrative. That’s how you end up with a 47-slide deck about customer sentiment that results in zero operational change.

BCG has written about this tension between data availability and experience quality, noting that what actually shapes customer experience is often more about organisational design and decision-making than the sophistication of the analytics stack. That matches what I’ve seen. The companies with the best customer experience don’t always have the most data. They have the clearest operating model for acting on what they know.

The Difference Between Behavioural and Attitudinal Data

If you want to understand your customers, you need both types of data. But you need to know which one you’re looking at, because they tell different stories and they’re useful for different things.

Attitudinal data is what customers tell you. Survey responses, NPS scores, review ratings, social feedback. This data is valuable for understanding how customers feel and what they believe about your brand. It’s also easy to game, easy to misinterpret, and structurally biased toward the people who bother to respond.

Behavioural data is what customers actually do. Click paths, purchase sequences, return rates, support ticket volume, time-on-page, drop-off points. This data is harder to manipulate because it’s a record of action, not intention. It’s also harder to interpret because behaviour without context is just numbers.

The gap between what customers say and what they do is one of the most commercially significant gaps in any business. I worked with a retail client who had consistently high NPS scores and consistently declining repeat purchase rates. When we looked at the behavioural data, the problem was obvious: customers liked the brand but found the post-purchase experience frustrating enough to go elsewhere next time. The NPS was measuring brand warmth. It wasn’t measuring operational friction. Two different things.

This is part of a broader set of questions about how businesses understand and manage the full customer experience. If you’re building out your thinking in this area, the customer experience hub covers the strategic and operational dimensions in more depth.

Where the Data Actually Lives (And Why It’s Fragmented)

One of the structural problems with data-driven CX is that the data is almost never in one place. It’s distributed across marketing platforms, CRM systems, e-commerce platforms, customer service tools, and finance systems. Each of these is typically owned by a different team, governed by different rules, and updated on a different cadence.

In large organisations, this fragmentation is the norm. I’ve been in businesses where the marketing team was optimising ad spend based on attributed conversions that didn’t match the actual revenue numbers in the finance system. Both teams thought they were right. Neither was wrong, exactly. They were just looking at different parts of the same reality through different lenses.

The consequence for customer experience is that no single team has a complete picture of what a customer has experienced. Marketing knows about acquisition. Sales knows about conversion. Customer service knows about complaints. Finance knows about payment behaviour. But nobody is stitching those together into a coherent view of the customer relationship.

Forrester has been consistent in pointing out that practical CX improvement requires cross-functional data ownership, not just cross-functional goodwill. You can run all the workshops you want. If the data is siloed, the insights will be siloed too.

Solving this doesn’t require a multi-year data transformation programme. It requires identifying the three or four data sources that, if connected, would give you a materially better view of the customer. Start there. Build the connective tissue between those sources first before adding complexity.

Personalisation: The Gap Between Ambition and Execution

Personalisation is probably the most over-promised and under-delivered capability in customer experience. Every marketing technology vendor sells it. Most businesses are still doing it badly.

I remember sitting through a pitch from a major network agency that was selling an AI-driven personalised creative solution. The headline claim was a 90% reduction in CPA and triple the conversion rate. The room was impressed. I wasn’t, because when I pushed on the baseline, it turned out they’d replaced genuinely poor creative with something competent. That’s not personalisation success. That’s a low starting point making the numbers look dramatic. The AI was doing real work, but the story being told around it was inflated beyond what the data supported.

Real personalisation at scale requires three things that most organisations haven’t sorted out: clean, connected data; clear decision rules about what triggers what; and the operational capacity to actually deliver different experiences to different people. Most businesses have a partial version of one of those three things.

The most effective personalisation I’ve seen in practice is usually modest in scope and precise in execution. A financial services client we worked with didn’t try to personalise everything. They identified five specific moments in the customer lifecycle where a personalised message had a measurable impact on retention, and they built reliable data pipelines and content workflows around those five moments. That’s it. No grand unified personalisation engine. Just five well-executed interventions that moved the commercial needle.

HubSpot’s work on collecting customer feedback through social channels is a useful reminder that personalisation signals don’t only come from your own platforms. What customers say publicly, and how they respond to your content in social contexts, is behavioural data too. It just requires a different collection mechanism.

The Touchpoints Businesses Consistently Underinvest In

If you mapped where most businesses invest their CX budget against where customers are most attentive, you’d find a consistent mismatch. The investment goes into acquisition-facing touchpoints: website design, advertising, onboarding flows. The underinvestment is almost always in post-purchase and service touchpoints, which is exactly where customers are most engaged and most likely to form lasting impressions.

Transactional emails are a good example. Order confirmations, shipping notifications, account updates. These have open rates that most marketing campaigns would envy, because customers actually want to read them. They’re functional communications that arrive at a moment of genuine interest. And yet most businesses treat them as operational outputs rather than experience touchpoints. The design is minimal, the content is generic, and there’s no strategic thinking about what the customer needs at that moment or what the business could usefully say.

Optimizely’s analysis of transactional emails as a CX and revenue driver makes the commercial case clearly. These communications are not just operational hygiene. When they’re designed with the customer’s next question in mind, they reduce support contacts, increase repeat purchase rates, and improve the overall perception of the brand.

Customer support is another systematically underinvested area. Vidyard’s work on using personalised video in customer support is interesting precisely because it addresses the moment when a customer is most frustrated and most likely to churn. A personalised video response from a support agent costs more than a templated email, but it signals something the email doesn’t: that a human being looked at your specific problem. That signal has commercial value that’s hard to put in a spreadsheet but very easy to see in retention data.

Connecting CX Data to Commercial Outcomes

This is where most CX programmes either earn their keep or become expensive theatre. The question isn’t whether your customers are satisfied. The question is whether their satisfaction is connected to the commercial outcomes the business cares about.

I’ve judged the Effie Awards, which means I’ve reviewed a lot of work that claims to have driven business results. One of the things that separates genuinely effective marketing from the rest is the quality of the chain of evidence. Not just “we improved NPS and revenue went up”, but a coherent account of why the experience improvement led to the commercial outcome, with data at each link in the chain.

BCG’s research on the consumer voice in customer experience identified that experience quality has a measurable relationship with commercial performance, but the relationship is not automatic. It depends on whether the experience improvements are concentrated in the moments that actually drive customer decisions, rather than spread thinly across every touchpoint in an attempt to optimise everything simultaneously.

The practical implication is that you need to know which touchpoints have the highest leverage on the commercial outcomes you care about. For a subscription business, that might be the onboarding experience and the moment a customer first encounters a billing problem. For a retailer, it might be the delivery experience and the return process. These are different for every business, and they can only be identified by connecting CX data to commercial data at a granular level.

Video content is increasingly part of how businesses explain complex processes and reduce friction at high-stakes moments. Vidyard’s support experience tools reflect a broader shift toward using richer media formats not just for marketing but for operational CX, particularly in moments where text alone creates ambiguity or frustration.

Building a Data-Driven CX Capability Without Starting from Scratch

Most businesses reading this are not starting from zero. They have data. They have some tools. They probably have a CRM that’s partially implemented and a customer service platform that generates reports nobody reads. The question is how to turn that into something that actually changes how decisions get made.

Start with the commercial question, not the data inventory. What is the single most important thing that, if you understood it better, would change how you run the business? For most businesses, this comes back to one of three things: why customers leave, why customers don’t buy again, or why certain customers are significantly more valuable than others. Pick one. Build the data model around answering that question specifically.

Then identify the minimum data connections required to answer it. You probably don’t need a data warehouse and a machine learning team. You might need a clean export from your CRM, a join to your transaction data, and someone who can write a decent SQL query. The sophistication of the infrastructure should be proportionate to the sophistication of the question.

Video is increasingly being used to gather qualitative context around quantitative data. HubSpot’s overview of video in customer experience covers how businesses are using video feedback and video-based support to capture the texture of customer experience that survey scores don’t convey. That combination of quantitative signals and qualitative context is often what’s missing from CX analysis that feels technically complete but commercially useless.

Once you have a working model for one question, the capability compounds. The data governance you build, the cross-functional relationships you establish, and the habit of connecting CX data to commercial outcomes, these transfer to the next question and the one after that.

The businesses that get this right don’t have a CX data programme. They have a way of working that uses data to make better decisions about customers. That’s a cultural and operational shift as much as a technical one, and it’s the harder part to build. But it’s also the part that’s hardest for competitors to copy.

There’s more context on how experience strategy connects to broader marketing effectiveness in the customer experience section of The Marketing Juice, where the commercial and operational dimensions of CX are covered across a range of business contexts.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions