How do you move from owning Voice of Customer to facilitating it?

What does it mean to facilitate Voice of Customer instead of owning it?

Facilitating Voice of Customer means the insights team stops being the gatekeeper who produces every report and starts being the architect who lets everyone else find their own answers. Pete Daykin's framing in this webinar is direct: your job is not to know everything, it is to give people what they need. When stakeholders have understandable insight at their fingertips and trust it, they use customer feedback as their starting point rather than waiting on a quarterly report to confirm a hypothesis they have already formed.

The pay-off is scale. When finding and acting on customer insight becomes a "team sport", the number of actions colleagues can take grows exponentially, and the progress an organisation makes on customer experience accelerates with it. It is no longer a shame when a colleague finds an insight without being told it — Pete argues that is the whole point.

Why do insights teams become a bottleneck?

Insights teams become a bottleneck because the traditional Voice of Customer model is linear: collect feedback, categorise and analyse it, build a presentation, share it with stakeholders. That model puts the customer in certain inboxes rather than in every room, and the better the team is, the more requests it attracts — until the volume of requests outstrips its bandwidth. Zoe Wilson framed this as the central paradox of the owning model.

Three failures stack up. Data sits siloed across teams that each own a source. Stakeholders arrive with confirmation bias, looking for evidence for a solution they have already chosen. And insight risks becoming a "feedback graveyard" — the end of the line rather than the start of the story. As Pete put it, the value of customer feedback is inversely proportional to latency: by the time last quarter's survey is analysed, the problem has either been resolved or escalated to a crisis. Faster, democratised access is the way out.

What are the seven principles of continuous improvement for Voice of Customer?

Wordnerds adapted seven principles from the total quality management work that came out of post-war automotive manufacturing — W. Edwards Deming, and Shigeo Shingo at Toyota — and applied them to Voice of Customer. The seven: Voice of Customer should be strategic (customer-centricity in the long-term roadmap), data-driven (replacing gut feel with qual and quant evidence), integrated and accessible (a single source of truth, not siloed data), transparent (showing how data is classified and what action follows, internally and externally), employee-led (insight empowering frontline staff, not owned by a few), process-centred (repeatable, optimised workflows rather than a few CX fundamentalists), and agile (hypothesis, test, feedback loop).

These run on Deming's plan-do-check-act cycle, which Wordnerds maps to four moves: get the data centralised and standardised; check what it says through categorisation and prioritisation; act with a minimum viable test, co-created with colleagues and customers; then communicate the result back to everyone, win, lose or draw.

Why do two analysts get different answers from the same customer feedback?

Two analysts reach different conclusions from the same feedback because of three misalignments, which Izzie Johnson illustrated with two fictional hotel-experience colleagues, Susie in growth and Tom in ops. In the example, Susie reports breakfast sentiment down 18% while Tom sees a 6% drop. The first reason is different data sources — Susie analyses surveys and social, Tom looks only at surveys. The second is different definitions — Susie means food and drink at breakfast, Tom means anything from menus to wait times to seating.

The third is the deepest: non-deterministic AI. Susie used Copilot, Tom used ChatGPT, and general-purpose models will tag the same feedback differently each time they run — let alone across two different proprietary algorithms. The result is that trust in Voice of Customer erodes, which pushes teams to hoard analysis rather than share it. Wordnerds' answer is a shared, central methodology so the same classification runs identically every time.

What are themes and frameworks in Wordnerds?

Themes in Wordnerds are custom classification models you build in the platform that group data according to rules you define — from something practical like car parking to something nuanced like "guest felt ignored". The point Izzie Johnson stressed is repeatability: whether a theme is applied to one source or to social, surveys and complaints together, it runs identically every time, giving a consistent foundation rather than a one-off analysis.

Frameworks are saved, filtered views that hand-pick and group themes from the central bank — stacked like Lego, as Izzie put it — so a growth lead and an ops lead can each apply their own lens in Power BI with a few clicks and strip out everyone else's noise. The platform's newer definition-led themes make this faster: where the older context-themes method took around ten minutes to train a theme, definition-led themes take two to four, with average training time down by 60%. That combination of a central theme bank and clear definitions is what lets people self-serve without having to defend their data-gathering afterwards.

What happens to the insights team's role when everyone can access the data?

When everyone can access the data, the insights team moves from gatekeeper to facilitator and orchestrator — you become the architecture of the system rather than the builder of every solution. Pete Daykin's framing: you set up the standardised structure, ensure data quality and consistency, build the continuous-improvement machine and make sure people are executing on it. You become the guardian of the feedback loop — did the hypothesis get tested, did the data improve, what did we learn, what's next — and the currency of the business becomes how fast it can turn insight into learning.

The early outcomes back this up. Wordnerds is seeing customers who reported spending up to 80% of their time in data and reports report a three-to-four-times reduction in ad hoc requests, fewer challenges to "black box" categorisation, and stakeholders shifting from "how many complaints about X?" to the higher-order question of whether the fix they put in place is working.

Full Webinar Transcript

Zoe Wilson: Good afternoon everyone, and thank you for joining us today. You're here for our live webinar, "The Customer Always in the Room", where we're going to be talking about how you can move from owning Voice of Customer to facilitating it while still maintaining your expert control. The chat is open now, and it'll be open throughout, so please do pop a message in — let us know who you are, what your role is and where you're joining from.

Since it's Pancake Day, we're going to have a bit of fun, and we'd love to know your top topping. We've been having a debate at Wordnerds — lemon and sugar, the classic, has come out on top already, but personally I'm going for a white chocolate and raspberry variation tonight. Give us some inspiration and settle the debate.

While everyone's joining, let me introduce our speakers. I'm Zoe, your Customer Success Manager. I work with customers at Wordnerds across a range of different industries to help them get value from using the platform. Today we've also got our CEO and co-founder Pete. Pete spends the majority of his life talking to CX and insights professionals and is especially invested in what best-in-class Voice of Customer programmes look like, and in how Wordnerds can be a valued partner in these. And we've got Izzie, our Product Manager. Izzie is really on the front line of developing the platform, especially leading on our newest release of definition-led themes and how we can integrate with Power BI. So Izzie's going to be sharing a lot of that expertise today, including a live demo. We've also got the wonderful Katie, who'll be in the chat today.

A few housekeeping things. Please do continue the intros in the chat. I'm also going to bring up a poll, just to frame the discussion: we'd love to know where you're at with Voice of Customer and sharing it in your organisation — ranging from having an interactive Voice of Customer dashboard accessible across your organisation, all the way to Voice of Customer being siloed in individual departments, maybe even manual tagging. We'll be doing a live Q&A at the end, so if anything comes to you as we go, please pop questions in the chat. A quick note that anything you see data-wise on screen today is all publicly available data, and we're recording the session so we can share it directly afterwards.

So, framing what you'll get out of today: we'll start with some theory around the continuous-improvement approach, where it comes from and why it matters for Voice of Customer programmes. Then we'll move from theory into practice and look at how you can actually build this system using Wordnerds to turn insights into action — centralising data, making it available across your organisation and empowering your stakeholders to take action. That will include the demo. Then we'll finish on what this actually enables you to do, and of course the Q&A.

To start this off, I'm going to refer to a story some people might be familiar with. Jeff Bezos, the founder of Amazon, would famously mandate that an empty chair be present in every Amazon meeting. The chair represents the customer, and the idea is that it reminds everyone the customer should be at the centre of all decisions. As Voice of Customer nerds, we are fans of this idea — but there's a problem, which is that the customer isn't actually there. This is just executives in a boardroom making assumptions about what they think the customer's opinions are. There's a quote from Jeff himself where he says they'd think about what kind of opinions the customer might have, or how decisions would affect the customer — but there's nothing about bringing the real customer voice, the real customer data, into that thinking.

The question becomes: how can you actually build a structure where real Voice of Customer data is in every room, constantly being referred to, to drive real decisions? Before we unpack our model, let's look at why the old structures can't quite support this. If your Voice of Customer programme is more linear, it might look like this: you collect customer feedback, your insights team categorises and analyses it, they produce a presentation — perhaps a PowerPoint — and that's shared with stakeholders. This isn't necessarily bad. It's just not doing the job of putting the customer in every room. Instead it puts the customer in certain inboxes of certain people, and others are either ignorant of these great insights, or it's unclear and they're receiving noise from lots of different places.

Looking deeper at the challenges of this model: firstly, siloed data — different teams own different sources, or feedback is trapped in a single department. Next, your insights team are likely amazing, doing a lot with limited resource — but what happens when the volume of requests outweighs their bandwidth? They become an unintentional bottleneck. The analysis starts with a pile of messy data that needs consistent categorisation; if you use different technology in different departments, or generic categorisation models, you get a lack of clarity and a consequent lack of confidence. Then, when you present to stakeholders operating with years of experience, that brings brilliance but also preconceptions — before they even ask a question they're often looking for information to confirm assumptions they already hold, so you're fighting confirmation bias. And finally, there's the constant risk of what we call a "feedback graveyard": if the process is linear, insight becomes the outcome at the end of the line, when really it should only be the start of the story.

So today we want to take that linear model and make it better — to move to a facilitation approach, ensuring customer feedback moves from departmental intelligence to strategic infrastructure, where the customer is really in every room. Before I hand to Pete, the poll results show a real mix — someone at each stage, from siloed feedback to static reports through to live interactive dashboards. So I'm keen to hear as we go through whether this approach resonates. I'll hand over to Pete.

Pete Daykin: Thanks, Zoe. Before I start — I want to go to Sophie's house for tea: cheese and Marmite on pancakes. What an absolute pervert. I love it.

Before I get too far in, I want to take a step back and explain some of our thinking — how we've arrived at this journey towards continual improvement and facilitation. I know a lot of you on the call are customers of ours. It starts about two years ago, where we began to see a shift in the maturity of Voice of Customer. Prior to that, the problem was "we're drowning in data, we don't know what our customers are telling us", and the race was to use technology to analyse it all at scale. Then there was a progression: we're not drowning in data anymore, we've got insights — but we've still got loads of insights and we don't really know what to do with them. And then, about eighteen months to two years ago, people who were getting really good results started saying: the insights are great, but what I'm less sure about is our ability to do meaningful things with them. There's no point just telling us things we haven't fixed. It's this idea of data to insights to action.

We are not consultants — we don't go into businesses and tell them how to improve the operational side of what they do. But we do believe we have a role to play by lining things up the right way. So we looked at how you turn insights into continuous improvement, and — being the dull, nerdy people we are — we went to the library and read a lot of books about it. The stuff we found most interesting came out of the automotive industry in the post-war era, the seventies and eighties: the work started by W. Edwards Deming and repopularised by Shigeo Shingo at Toyota, and all the total quality management ideas — Six Sigma and the rest. We were fascinated by how factories went from hit-and-miss quality to making quality a vital part of everything they did, and the business transformation that delivered it.

What we found was seven key principles we believe are transferable from quality in car manufacturing to customer experience and Voice of Customer in a modern insights team. Strategic means customer-centricity as a fundamental component of the organisation's long-term roadmap — Voice of Customer prioritised in every interaction and every meeting, the thing Bezos was reaching for with the empty seat. To be strategic it has to be data-driven: replacing gut feel and opinion with insight from both qual and quant feedback. To be data-driven, data has to be integrated and accessible — no more marketing owning social, CX owning complaints, the app team owning app chat; we need a single source of truth everyone can access. When it's integrated it also needs to be transparent: you share how data is processed, classified and interpreted, and crucially what actions are being taken — internally and externally, because confidence comes from everyone knowing what's going on.

It needs to be employee-led: customer insight isn't the domain of the insights team anymore, it should empower frontline staff to lead change. And if that's going to happen it needs to be process-centred — not the domain of a few radical CX fundamentalists, but the output of repeatable, optimised workflows built into the fabric of how your system operates. And finally it needs to be agile — not in the loose "flexible" sense, but the pure sense: a process that embraces iterative, incremental change and feedback loops. You look at something in the data, form a hypothesis, run a test, look at the data again — better, worse, or no change — and adjust what you've done in light of it.

In terms of how we deliver that, we went back to Deming's plan-do-check-act cycle and asked what it means for someone in an insights team working with customer experience data. It starts with the data: getting it in the right place, centralised and accessible, bringing in all the sources where customers speak to you, and standardising it — both the data itself, so the same things are called the same name across sources, and the methodologies, because in the nightmare scenario Zoe painted, different teams use different platforms with different sentiment analysis that isn't comparable. So our first job as insights professionals is to bring all of that together.

After that we do the checking: what's the data telling us? For us that's in two parts. The analysis piece is categorisation first — and as many of you know, we've fundamentally changed how we deliver categorisation, moving from context themes, where people ticked or crossed things in and out of data sets, to something quicker and more transparent. Then prioritisation: even great categorisation just gives you lots of things customers are telling you. We go back to the plan — what are you trying to solve — and look at which feedback moves the needle on it. The second part of checking is democratisation: putting this in the hands of the colleagues closest to the customer, the systems and the data, so they can act on it.

Then we move into action. This is the bit we weren't really thinking about a couple of years ago. The best practice we see in advanced teams is their ability to take customer insight and turn it into a test — and by test I mean a minimum viable test. We think the data is telling us this; what's the smallest possible thing we can do to see whether we can improve it? If we can co-create that with our customers, and if other people in the organisation can have those conversations and design changes without us, that's great. We measure it as we go back round the loop, and win, lose or draw, the final stage is communicate: constantly telling people, inside and outside, "we surveyed you, you said this, we took it to mean that, we did this, and now more of you are saying this." If you create that virtuous cycle, people don't expect you to be perfect — they know there are problems and constraints everywhere — but they really appreciate seeing genuine efforts to make things better. Often that's all they want. At the end of the loop you look again at your objective, reassess whether it's still a priority, and keep going. That's what delivers continuous improvement.

What was interesting to us was the pivot around democratisation. There's always been an unspoken assumption that insights teams should own Voice of Customer — that you're the gatekeepers, stakeholders come to you with requests, you produce reports and tell them what to do. We think that model is changing, not because insights teams aren't good, but because of the volume of data. Customers speak everywhere now — surveys, complaints, social media — and decision-making needs to be faster; waiting for quarterly reports is no longer feasible. The value of customer feedback is inversely proportional to latency: the quicker you get it, the more valuable it is. By the time you've analysed last quarter's survey, the problems have either been resolved or escalated to a crisis. So the counter-move isn't to defend your territory and take more control — it's to do the opposite, to facilitate access and democratise that learning.

That means letting customers speak in their own voice, not forced into your surveys; taking it wherever they're speaking; not ranking channels, because people tell you different things in different places. Gather it consistently, analyse it robustly, then democratise it, so stakeholders can investigate for themselves and Voice of Customer becomes something always there rather than something they request. It's no shame when someone finds a customer insight without telling you — that's the whole point. A lot of people worry that if they're not telling people the insights, they're not doing their job. But if you let people find their own insights, your job stops being to know everything and becomes to give people what they need. When people wait for reports, they've already developed a hypothesis and a confirmation bias, and the data becomes an uncomfortable inconvenience. When they've got understandable insight at their fingertips and confidence in its veracity, they use customer feedback as their starting point.

You're making this a team sport. By making action a team sport you increase exponentially the actions colleagues can take, and you massively accelerate the progress you make as an organisation on customer experience. That's the conclusion we reached. Everything we've done from a Power BI and a definition-led themes perspective has been in pursuit of this new world — where the insights team builds the themes, the job for a few, so that activity can become the team sport of the many. Zoe.

Zoe Wilson: Thanks, Pete — so interesting, and really good to have the theory behind it. Just before I hand to Izzie, back to our poll: the most popular answer now is the insights team dealing with individual requests, which speaks to your point. It's almost a paradox — the better you are at your job, the more requests you get, and the harder it gets to manage them. It's much more efficient to get the data out in front of the people who'll use it so they can investigate for themselves. So Izzie is going to talk us through how you can actually build this — from centralising the data to creating the categorisations, all the way through to a BI demo.

Izzie Johnson: Thanks, Zoe, and thanks, Pete — I scribbled so much down. One thing that stuck with me about the total quality management process is how much space it gives to both employees and process, in balance, and I think that's one of the reasons the model works so well. In this age of data democratisation, speaking for myself, it sometimes feels like we take two steps forward and one step back because we don't have the right blend of process or rules of engagement for our people. That's a lovely segue, because I wanted to start by asking: yes, we're a data-driven generation with people power — but is everyone in the organisation following the same rules of the road, or do we all have different processes for Voice of Customer?

To illustrate this, let me introduce two fictional personas who both work in travel and hospitality, both focused on hotel guest experience, but with very distinct roles. First, Susie, in the growth team — she's focused on fixing friction in the end-to-end customer journey, from booking through arrival, the stay and check-out, to drive loyalty and retention. Then Tom in the ops team — he's monitoring CX data for issues that regularly lead to complaints, so he can mobilise teams ahead of escalations and oversee incident management. Two people, both in guest experience, both analysing guest data, but — spoiler — with very different approaches.

Here's a made-up conversation between them, exaggerated to highlight how incongruences in our approaches can quickly cause misalignment. Pete asked me to do different voices for each person; I just can't find it in me. Susie says to Tom: "I ran our survey and social data through Copilot and sentiment around breakfast is down 18% this month." Tom says: "Huh — when I analyse surveys I see a 6% drop, with lots of comments about menus not being up to date." So here's challenge number one: different data sources. Susie's looking at social and surveys, Tom only at surveys. People leave feedback in many places, often where they're most comfortable, including tweeting about a poor breakfast — so it's no wonder they've reached different conclusions.

Then Susie says she was looking at comments about food and drink at breakfast only, not the menus. Tom was looking at any mention of breakfast — menus, wait time, seating. That's challenge number two: different definitions of breakfast. Tom's map is much broader than Susie's. Finally Susie says: if we agree on one definition of breakfast, can you rerun your analysis? Tom says yes — but even with the same data and the same prompt, there's no guarantee the AI will tag feedback the same way again, and he was using ChatGPT, not Copilot. This is a huge problem in Voice of Customer. These two AI systems are non-deterministic: run a prompt again and again and you get different results each time, and across two different proprietary algorithms you get even more variation. So there's no hope of repeated analysis reaching consistent conclusions, and no visibility of whether different data sources are weighted or treated the same — it's all an AI black box.

What this shows, albeit to the extreme, is that misalignments stack up really quickly. Going back to insights being a team sport, it erodes our ability to make it one, because trust in Voice of Customer is eroded, and our trust in ourselves to present data defensibly is eroded. It's natural that this pushes us to want to own analysis rather than share and facilitate it, for fear of inconsistent conclusions — and, going back to Zoe's feedback graveyard, that's exactly what we want to avoid in an era where we have more data than ever and more responsibility than ever to honour the feedback people give us.

Which leads nicely into why a central methodology for CX analysis is so important. Let me explain more about the Wordnerds platform and how it tackles these challenges through themes. For those less familiar, themes are, in a nutshell, a custom classification model you create in the platform that groups your data according to the rules defined in that theme. They can be about anything — weird, wonderful and colourful, like Lego pieces. Practically, you might have a theme about car parking that matches all verbatim mentioning car parking; a broad one like comments about restaurant staff, regardless of breakfast or lunch; or something nuanced and emotional like "guest felt ignored". So we already have different layers of themes.

Question one: with so many types of theme, how do themes build alignment across many different stakeholders in a Voice of Customer team? The beauty is that whether you apply a single theme to one data source, like social, or a mix of social, surveys and complaints, the theme works the same way every single time — a repeatable, consistent programme. That classification runs identically regardless of data source and every time it's run. Question two: how does that facilitate teams to self-serve and be autonomous? That brings me to another feature, frameworks.

All those themes — those Lego pieces — are created and stored centrally in Wordnerds for everyone in your organisation to interrogate. From that central store, themes can be hand-picked, grouped and saved into filtered views we call frameworks. Going back to Susie in growth, she might have a framework focused only on the end-to-end customer journey — picking everything to do with the check-in process, perhaps leaving out facilities, which isn't part of her remit. Tom in ops, looking at incident management, might include room cleanliness, parking, noise — a few things from different categories he can stack like Lego to create his own custom framework. When these two open their Power BI, they can apply that framework with a few clicks and filter every visual to just the themes that matter to their role, removing all the noise of other departments, and deep-dive into sentiment analysis focused exclusively on their lens.

Finally — Pete mentioned this — we recently launched a feature called definition-led themes, which uses some very clever AI mechanics to help you build a theme by providing and then refining a very clear definition. Earlier, when Susie and Tom couldn't agree on breakfast, we have a similar theme in our demo project called "quality of breakfast", where you can see very clearly from its definition that we want all sentiment about breakfast — good, bad or neutral — but only mentions that clearly involve breakfast or food, excluding general service that doesn't include food. I've just spotted a few typos in this definition — let's make it a teaching moment: the AI helpers here don't care about my typos. It'll still go and find that theme programmatically, despite me being a human who can't always type.

So, in summary: number one, these centralised theme banks with clear definitions provide that single source of truth for Voice of Customer. And in having that centralised theme bank, you can create repeatable workflows for anybody in your organisation that ensure themes are applied consistently, time and time again. It's this combination of process that enables your people to self-serve Voice of Customer — to find and explore insights without feeling they then have to defend their data-gathering, because they're pulling themes from the same pool as everyone else. That's the mechanics of the continuous-improvement machine in our platform.

Let me share my screen for a quick BI demo. Zoe, tell me when you can see my screen — I can't see you afterwards. What I'm showing, for anyone who hasn't seen our BI before, is a specific page in our travel-and-hospitality industry template called journey frameworks. I've filtered the data down to the back end of the 2025-26 financial period for this demo, and applied the customer-journey framework — so think about Susie in growth, looking at friction in the end-to-end journey. This is her framework applied to all the data on this page.

One thing I didn't mention earlier: yes, we have themes you can train, but in all the reviews and data you upload there's also metadata — a review might have a hotel type, a hotel name, a city. That's not unstructured feedback, it's metadata, and we can use it to create slices. So while looking at a top-level framework, you can limit it down to a certain hotel type. I'll limit this to package holidays. A few things I see straight away as Susie: a lot of volume, just over 34,000 for Q1 and Q2 of this year, which is great for a growth role — a thirty-thousand-foot view of all the themes and feedback relevant to her. Just over 40%, good to know. And I can see the overall sentiment of my package-holiday feedback is 78, on the star-rating KPI we've pre-coded for this customer.

On this graph, the Y axis is sentiment and the X axis is the different stages Susie has coded into categories — stage one is booking and pre-arrival, stage two arrival and check-in, and so on. There are more reviews referencing booking, arrival or post-stay retention, judging by the size of the bubble. But what I see straight away is that down in rooms and suites — a smaller bubble, less volume — the sentiment for that part of my end-to-end journey for package holidays is really low. Clicking into it, it's 1% of my volume for this period, but in a growth CX role I can think about the strong correlations between overall journey sentiment and retention down the line, so it's in Susie's interest to figure out what's happening here.

I've clicked on this and it's pulled up a sentiment chart for all verbatim referencing this category — 51 for all of those — and I can see all the themes grouped within rooms and suites: bedroom quality, cleanliness, and more nuanced ones with less volume. Now I can deep-dive. Something is definitely happening in bedroom quality, so I'll drill into these very negative sentiments and through to the verbatim. Number one, straight away — a Tui hotel in Tenerife, hundreds of bedbugs. Someone's definitely had a rubbish time with their accommodation. So now I can go through all the verbatim, find what I want, and take it into further BI pages if I feel inclined. That's a whistle-stop tour — happy to answer more questions, but otherwise I'll hand back to Zoe.

Zoe Wilson: Amazing — thank you, Izzie. A really great snapshot that illustrated the point: depending on who you are and what you want to see, as long as you've got those building blocks set up in your themes, you can completely customise how it comes out in BI. It's always a challenge during a BI demo because there's so much you could show, but hopefully that shows that just selecting a different framework gives you a completely customised view. So, to wrap up before the Q&A — and do pop any questions in the chat — Pete is going to take us through what this enables, in terms of moving from the more tactical to strategic, embedded infrastructure.

Pete Daykin: Thanks, Zoe. It's brilliant, isn't it — we've practised this three times and you only spot the typos in the live version. A great save. Before the "what this enables" piece, it's worth surfacing some anxieties. Definition-led themes hit our beta testers a few months ago, and we've been rolling out Power BI dashboards for a year or so. We get three main questions at the start of this process.

The first: "I'm not an expert Power BI user — what if I read the data wrong and people think I can't do my job?" That's an entirely reasonable apprehension. But your job as an insights person is to provide the structure and the definitions — I love the Lego-block analogy, that they're different colours and can be stacked in different ways, and as an insights team you decide what's in a block and how blocks are grouped. For different people they'll be different things; this stuff has to be customisable, or people won't self-serve. It's your job to understand what views they need and organise that process — you're building the architecture. You don't need to build any dashboards; that's our job. We have BI specialists in house — Frankie is a wizard. We can give those to you, and often people have their own BI expertise in house and love building on top of our data. Consumed in the browser, Power BI is just like any other dashboard — you filter, follow your nose, click on things. And we're doing our gamma release of Power BI next month, with tooltips that explain everything, supported by training content the CS team will roll out. If you've got questions, go straight to your Customer Success Manager.

The second apprehension: "If I make Voice of Customer transparent, will people start questioning how I've categorised things, interpret my data differently, think I'm rubbish at my job?" The answer is a resounding no. With people creating their own definition-led themes and putting them in dashboards — and I have the exotic Dutch voice of a customer of ours ringing in my ears here — the beauty of definition-led themes is they're quicker, lower friction and easier, but because you've got the explanation of what's in and what's out, they're really transparent. If you take a little time to rebuild your classification model using definition-led themes, you become super-confident about what your data is and isn't showing. In that context, questions about themes become a useful opportunity to debate whether you were right to include something, or whether you need separate themes for different nuances — all constructive conversations. People are much less likely to decide you're wrong.

More than that, the whole point of continuous improvement is that it doesn't matter if you're wrong. Your goal as an insights manager isn't to interpret the data perfectly first time. Your job is to see something, build a reasonable hypothesis and test it — or see that someone else is testing it — quickly, look at the data again, and change what you're doing based on the test. That's a massive mind-shift, away from insight as the final product and into customer feedback as an ongoing source for continuously steering the organisation. You can be wrong, that's fine, as long as you're constantly correcting based on the data coming in.

The last one we get is: "If everyone can access the insights, what's my role? It's not my job to make changes across the business." That's definitely the case. In this new world of continuous improvement, the role of an insights team moves from gatekeeper to facilitator and orchestrator. You're the architecture of the system, not the builder of every solution. You set up the standardised structure, ensure data quality and consistency, build the continuous-improvement machine and make sure people are executing on it — effectively enabling other people to be effective, which massively multiplies your impact. You're the connector between insights and actions, not the doer, and you become the guardian of the feedback loop: did the hypothesis get tested, did the data improve, what did we learn, what's next? What's really interesting is that the currency of the business becomes learning — how fast you can turn insights into learning becomes the new thing you measure. The more you learn, the more you improve customer experience and the better all your other KPIs become. It's a really powerful leading indicator.

So what does it mean if you get it right? I'll whistle through this to leave time for questions. We're already seeing people who reported spending up to 80% of their time in data and reports talking about a three-to-four-times reduction in ad hoc requests — think of the time that gives you back. People report fewer challenges to black-box, generic categorisation, and more confidence in explainable, tailored insights. People stop coming in with "how many complaints have we got about X?" — and it's interesting in the poll that this is still the predominant way most people here share insights. It turns answering ad hoc questions into helping colleagues see whether the fix they put in place is working, which is a much higher-order question. It moves insight from the end of your job to the beginning of a measurable continuous-improvement cycle. And it stops the confirmation-bias problem of people coming late to the data with their minds made up — turning Voice of Customer into the starting point that informs all the little micro-decisions. That's the shift from gut-led to data-led.

It sounds easy and clean when we put it like this, and we know it's not in practice — this stuff is messy, customer feedback is messy, Wordnerds is never 100% accurate, there's always a degree of interpretation. But hopefully the process lets you build a robust methodology and system that maximises your chance of trying enough things on the questions that matter most, to improve outcomes for your customers. At the end of the day, that's what all of us are here for. Zoe, back to you.

Zoe Wilson: Great, thanks Pete. Interestingly, a couple of the questions we've got speak to other anxieties around this process. Before we get to them, I'll give people a chance to add more questions. While we do, a few resources we'll be making available to help with the shift to a more democratised, strategic approach. First, the Power BI gamma release Pete referenced — we've done a lot of learning, customer testing and feedback at Wordnerds, so thank you to everyone involved; if your CSM hasn't been in touch about that, they will be. We've also been working on a theme-bank assessment framework, so if you've already got your theme bank set up, now — with the Power BI release in mind — is an excellent time to review and optimise it; the way we talk and the things we talk about advance, so your theme bank may need a review. This assessment framework will help you look at what you've got, where gaps might be, and how to optimise. Related to that, we're running a series of workshops in March to help you make this shift, and we'll send invites in the next week. And finally, all this will come together in a digital resource pack, where we'll also chunk down clips from this webinar and key takeaways so you've got everything in one place.

Now into the Q&A. The first question — thank you, Sarah: "Something I get asked is, our culture isn't data-driven. Will this actually get used? What's your take?" I think this is a good one for you, Pete.

Pete Daykin: Yeah. Nobody sets out not to be data-driven — we all want to be. So usually when I hear "our culture isn't data-driven", it's because either people don't know how to access the data, they don't understand it when they access it, or they don't believe what the data is telling them is true. For me, this is the democratisation, transparency and confidence piece. By getting this stuff out there so you literally go to a URL, see the dashboard and filter it, we hopefully get around the accessibility problem; with definition-led themes we bring in the transparency, so people understand how we've categorised it. Then, in terms of acting on it, it's up to them — they'll hopefully believe it because they understand what's gone into it, and the rest is on them. There's a limit to what we can do: we can lead the horse to water but we can't make it drink. As insights people — in the general sense, not just Wordnerds — there's a point at which we've done everything we can. But it's a great question, and in that situation, concentrate on the quality of data.

Zoe Wilson: Great tips, thank you Pete. Next question — I think this is a good one for Izzie. Thank you, Patricia, who asks: "At the moment we don't have capacity for a massive rebuild. This sounds resource-intensive. What would you say to that?"

Izzie Johnson: I take your point, Patricia. I assume you're someone who's used our context themes historically. For reference, we're super proud of the context-theme tech we used to train themes — but it used to take upwards of maybe ten minutes to train a theme, and now, with definition-led, we're down to between two and four minutes. That process has really expedited how long it takes to create a theme and your theme bank, so it's much less labour than it used to be, and you can have as many tabs open as you want, with various themes running at once. So yes, there is some upfront work, but it's much simplified — our latest stats say average training time for a theme is down by 60% with definition-led themes. There are also benefits to spending time setting this up, because in the long term the time you save from having a theme bank with clear definitions will pay dividends. So I'd challenge back: doesn't that justify a little bit of resource-intensive effort at the beginning? I think yes, it does.

Zoe Wilson: It's such a good point — and now that we've done a lot of the learning around Power BI, that output is there and done for you, with hopefully ready-to-share dashboards. Thank you for that. Next, a question from Sophie: "It's reassuring to know we're thinking in the right way about Voice of Customer. It's quite the challenge to get the whole business thinking the same way, especially around democratisation and co-creation led by a customer-experience strategy. It takes a long time to get everyone on board, and longer still to get things embedded as the new way we do things around here." Pete, I can see you're going to jump in.

Pete Daykin: Yeah. What I'd say is: I don't want anybody on this call thinking this is now the new gold standard of Voice of Customer and there's a bunch of people already magically here. We work with people at lots of different stages of maturity, and we're bringing on new customers all the time — it's a real journey for everybody. Like any organisational change, it starts with a few people grasping something, telling some other people, showing some other people, and people seeing the outcome of what they do. That process takes a bit of time. If I'm not wrong, Sophie comes from Town and Country — this is something Theo did really well early on in Voice of Customer, where he took some really small projects and ran a few things just to get people on board. When people saw the power of this stuff, that's when Voice of Customer in Town and Country really became a thing people valued. This is just the next iteration of that — it's evolution, not revolution. Different people arrive at it at different times, and that's okay; you're not going to convert everybody at once. Just keep fighting the good fight, and if you do it and it works, eventually people cotton on, and that's when these things really take off.

Zoe Wilson: Yeah, absolutely — thanks so much, Pete. I know we're at time, so I'll wrap up by saying thank you so much to everyone for joining. As we said, the recording will be in your inbox shortly, and we'll be in touch in the next few days with the resource pack, the theme-bank assessment framework, invites to the March workshops. If you've got questions that come to you afterwards, please reach out — whether to your Customer Success Manager, or Izzie or Pete, I'm sure they'd be happy to follow up. Thanks all for your engagement and questions today. Have a great rest of the afternoon, and enjoy your pancakes — I've been inspired by the toppings, so I'm going to the shops to get some more for a variation. Thanks everyone.