Re:Wire Episode 02 – Sanchia Neilson on Data Fluency, Unlearning, and Getting Ready for AI

Re:Wire, the Experience Haus podcast
Episode 02

Host: Amit Patel, Founder, Experience Haus
Guest: Sanchia Neilson, Co-Founder of Data Everyday; former Head of Data & Analytics at Digitas (Publicis Groupe)

In this episode of Re:Wire, Amit sits down with Sanchia Neilson to talk about one of the most overlooked pieces of workflow design: data. Sanchia has spent 20 years in data and tech, most recently as Head of Data & Analytics at Digitas, before leaving corporate life to co-found Data Everyday with Manny Moreno, a venture built around making data accessible, understandable, and a lot less frightening.

The conversation covers why so many people freeze up around data, who actually owns it inside an organisation, the three pillars Sanchia uses to bridge the gap between teams and data specialists, and what “unlearning” really means when you’re trying to redesign a workflow from the ground up. It closes with a practical look at what it takes to get messy, real-world data ready for AI-powered workflows.

Please note, this transcript has been lightly edited for clarity and length.

From a Maths Background to Head of Data

Amit: Welcome, everyone, to the latest episode of Re:Wire, our podcast here at Experience Haus. I’m Amit Patel, founder of Experience Haus, and today we’re tackling a really important topic: data. To help us get into it, I’m joined in the studio by Sanchia. Thank you so much for joining us this morning.

Let’s start with a bit of background: how did you get into data?

Sanchia: Thank you so much for having me, it’s great to be here. I’ve been working in data and tech for about 20 years. I’ve got a background in maths, numbers have always been my thing, and since then I’ve worked in-house and agency-side, in design and build agencies. Most recently I was Head of Data at Digitas, part of Publicis Groupe, a very large advertising holding company.

Since then I’ve left corporate life and started my own thing with my co-founder, Manny Moreno. We’ve built something called Data Everyday. It’s all about making people feel more confident with data: helping them find their own story in their numbers, and helping them stop being so scared of it. It doesn’t have to be big and scary. It can be little and powerful.

Amit: I’ve also got a degree in maths, so that’s something we have in common, and we’ll definitely get Manny on the podcast at some point too. Hello to Manny, we’ll get you on, no doubt.

Why People Struggle to See the Value of Data

Amit: At Experience Haus, over the last 12 to 18 months we’ve pivoted strongly into workflow design with all kinds of organisations. One thing I’ve noticed is a lack of awareness, or a lack of ability to recognise, how important data is throughout the whole journey of designing a workflow. What have you seen, in terms of people’s ability to recognise the importance of data?

Sanchia: It’s a great question, and it transcends departments: it’s not just for designers or senior leaders, it’s for everyone. What we’ve noticed is that people go straight into the data if they have access to it. Now, with AI, data is more available and accessible to everybody, so people think, “I’ll quickly dig into this and make a dashboard,” without really thinking through what they’re trying to do.

One of the biggest challenges is that people never ask the right question, and that’s only being made worse by easier access to data and to tools that analyse it. You’ve got conversational analytics where you can just ask, “How many customers have bought this product in the last 12 months,” without understanding why you’re asking, what you’re going to do with the answer, or how you’ll measure whether it’s had a positive impact.

Amit: Do you think organisations are spending enough time training people up? There’s obviously data owned by a particular person, team or department, but where is training happening, or falling over?

Sanchia: A lot of data teams are trained, but it’s often very technical training: you don’t always get insight training. The “so what” of it. I used to have a director who, every time I pulled information for him, would just keep saying “so what, so what” until I gave him the insight. So for data teams specifically, there’s a lot of technical training and not much insight training.

For wider teams, people don’t feel comfortable asking the right questions. They don’t know what data can do. There’s also that Rumsfeld thing: the unknown unknowns. They don’t know what they don’t know about data.

Who Actually Owns the Data?

Amit: A lot of our focus is financial services and professional services: we’re helping consultancies take on bigger, more complex challenges with their clients, and helping them rewire their approach, hence the name of this podcast. One thing I’ve noticed is a lack of awareness of who owns the data, often on the client side. Who do I need to speak to? Who’s going to help shape the data offering, the strategy, the sourcing? What have you seen in terms of the role of data on a stakeholder map: who owns it, how are people made aware, how is that governed?

Sanchia: It depends on the size of the organisation. Some of my bigger clients, Samsung, Monzo, Enterprise, have that infrastructure and governance in place. Smaller clients often don’t have those systems.

But even where the systems exist, they’re not necessarily connected, and that’s not just about the systems, it’s about the teams too. You’ll have a finance team doing commercial reporting, a data and tech team doing web analytics, and a customer team holding all of that rich natural-language data, and none of it is necessarily joined together. I’ve never seen anywhere, and that’s a big statement, with a perfect connection across all of their data systems.

Amit: Do you think companies are doing a good enough job of saying, “this is who owns it, this is who’s responsible for it”?

Sanchia: When it comes to privacy, people are much better at assigning ownership, probably because of the consequences. But everywhere I’ve seen, there’s more work to be done. It comes back to having multidisciplinary teams: designers who understand AI, who understand data. It’s not about checking other people’s homework, it’s about being able to go to the data team, or whoever looks after your data, even if it’s just one person with Google Analytics, and say, “can we sit together and work on this?” Everybody needs at least the fluency and the language to talk about data.

Why Data Feels So Scary

Amit: That’s a nice segue into fluency. What do you think makes data so scary?

Sanchia: I think a lot of it goes back to school: a lot of people are genuinely afraid of maths. I feel this quite deeply, I still do a bit of maths tutoring, and my husband’s a maths teacher. We’re lucky, we found maths easy, so we built those foundations early. But talk to most people and they’ll say, “oh no, maths isn’t my thing, I’m scared of numbers, don’t give me numbers, I’m a creative.” There’s a perception that data is hard-core maths, that it has to live in some massive cloud database. And people perpetuate that, because in some cases it is true, data science is hard, but in other cases it can be a lot more accessible.

When we came in and talked to one of your cohorts at Experience Haus, we talked about the little crumbs of data needed to power your journey from home to getting here today. How does Google Maps or Citymapper get that data flowing? GPS data, TfL data, traffic data. If you break it down into something people use every day, they understand it’s not a massive, scary database that you need Python to code: it’s a conceptual thing. You don’t have to be good at maths to be good at data.

Amit: That’s a really nice way of breaking it down. One thing I love is the multidisciplinary point: it’s not about being a generalist, it’s about knowing quite a bit about each of the pillars that power an AI-powered journey or digital product.

In terms of learning data: say you’re on a team designing a new workflow, and you’ve sensed where the data is coming from, how much does someone who isn’t working directly with the data need to know, to communicate properly with the data team that takes it forward?

The Three Pillars: Powered, Captured, Measured

Sanchia: We like to break it down into three things.

What’s the data that’s powering your design? In the Citymapper example, that’s your location, the TfL data.

What data are you trying to capture? What do you need from the customer or user to power the experience? So, “I need to get to Worship Street,” that’s the data you’re capturing from the user to make their experience better.

What’s the measurement and impact? This is arguably the most important piece. We talk a lot about breaking down your impact story: by making this change, we’ve improved this variable from X to Y, which means happy days, we’ve done a great job.

If you come to the conversation having addressed those three pillars, it helps you communicate much better with the data team taking it forward.

Amit: So they’re things you can start to pinpoint exactly?

Sanchia: Exactly. If you ask, “I want the most seamless journey to the studio possible: how am I going to measure that, and how do I measure it for hundreds of thousands, for millions of people?” If you start with the outcome you want, and the impact on the user and on the business, because we’re not working for free, it helps you get laser-focused on the right questions, because everything cascades from there. People talk about KPIs, but I’d argue not everyone uses the right ones.

Unlearning: Why “We’ve Always Done It This Way” Is the Enemy

Amit: Let’s talk about unlearning. This is fundamental to how we approach redesigning workflows, because so often you’ve got SMEs with a decade or two of knowledge who ask “so what,” or say “we’ve always done it this way, why change?” What does unlearning mean to you?

Sanchia: Big question! For me, it’s having the courage, the inquisitiveness, to ask: why has it always been done this way? What am I trying to achieve? What’s the purpose of this piece of work, this report, this analysis?

I’m not saying throw everything out, that’s not what I mean by unlearning. It’s more about questioning. Intellectual curiosity is really important, it’s what I try to teach my children, when it works. Manny and I talk about this a lot: it’s like being a kid who keeps asking why, why, why. That can be annoying, but it’s really important. As adults and professionals, we need to ask why something is the way it is. If the answer is satisfactory, great, it’s not broken. If not, could we do this better? A lot of that comes down to culture.

Amit: Taking that framing of unlearning and culture: say you’re redesigning an entire workflow, an e-commerce journey or a new account set-up. What are your recommendations for getting a team to start doing things differently?

Sanchia: I’ve seen a lot of interesting things. The first thing is: draw a picture. It’s easy to get dragged into the weeds without stepping back, and I don’t mean a giant bird’s-eye view, I mean literally: what is happening with the user? “I need a new phone. What do I care about?” I always go on about this because I want a small phone and no one makes them any more, you can find them on Vinted, though. Think about the amount of data they must have on people’s searching habits at midnight. The drunk purchase.

So you think: this is what I want, this is who I am as a user, these are my needs. Really map out who you’re designing for, and it’s not just a basic pen portrait, you can use real data. Vinted will have a profile on me, and on people similar to me, and you can use that to say: this is what she cares about, this is what she buys at midnight.

To bring it back to unlearning: it’s almost like stepping outside your own box and putting yourself in someone else’s shoes. I’m not saying everyone needs to be some kind of empath, but think: what would be nice for this person? It’s impossible to have lived experience of everybody else, and that’s where data and evidence come in: this is what would be good for a parent, this is what would be good for a business traveller.

When we mapped out an Airbnb journey, Manny was a segment, I was a segment, three very different people with very different needs, especially around booking a holiday. So: be inquisitive, be curious, ask those questions, put yourself in somebody else’s shoes. As soon as you do that, it forces you out of the weeds: “oh, I never thought about the safety element when you’re travelling with children,” or “I never thought about the fact that a business traveller who’s always on the road is going to miss home, how do we make this place feel more homely, and less like a sterile hotel?”

Data as the Connective Tissue Between Teams

Amit: You’ve talked a lot about stepping into the shoes of the consumer you’re designing for. Listening to you, I started thinking about internal workflows too, because a lot of the empathy isn’t just about the client, it’s about who’s doing the role before me, or the step after me. In large organisations, a Samsung, or one of the banks we work with, you’ve got SMEs who own one slice of a journey but are never really concerned with what happened before or after. What’s fascinating about data is: what do I need from the step before me, and what do I need to hand off for the step after me? Now that we understand we’re designing for different people, and what happens before and after, what changes in terms of the data flow from step to step? How do you approach that?

Sanchia: There’s a team-culture piece, and there’s a systems piece. As you were describing that, I was picturing a relay race: if you’re not even looking at the person passing you the baton, it’s not going to work.

Teams need to understand each other’s requirements and roles. At Digitas we used to do a lot of lunch-and-learns, or you’d drop into someone’s team meeting, I remember the head of UX asked me to talk to his team in their meeting, because you might work with some people on a client or a pitch, but never really talk to the whole team. Firstly: go in and explain what your team does, because they may have only touched one small part of your capability and never seen the whole thing.

We also used to run something called the Unicorn Academy at Digitas, because the logo is a unicorn, where you’d take a few people from each team and really show them how the business works, across each of the different teams. Bringing together that multidisciplinary piece: data really is the connective tissue.

One thing we did a long time ago for Samsung still sits with me: we had the insight that people were more likely to look at consumer reviews on a particular site than on the actual e-commerce site, or a tech-advisor type site. AI search has flipped that now, but back then, we built a targeted audience to test it. Being inquisitive means asking: we think this insight is telling us this, but is it true for our audience? We see it at a macro level, but is it the reality for our Samsung audience? Turns out it was, and we could prove it with a test, which was powerful because we could show the end-to-end journey. The insight work was mine, I passed it to Manny, who was the audience person at the time and built the audience, and our colleague Tucker was the measurement person who tested it, so you had that closed loop. That’s within a data team, and even within a data team, you have silos. That’s how we were able to break it down.

Amit: You came back to the word “silo,” and to systems thinking. What’s fascinating in some of the teams we’ve worked with is the data that gets left behind. When we run our rewiring work with a team, we’ll often get them to think about something unrelated to their own industry: if we’re working with a financial services organisation, we’ll say, let’s design a hotel, or an airport experience, take them completely away from what they normally think about, so they have the freedom to learn new approaches they can bring back into their own way of working. We’ll get them doing systems thinking, systems maps, but data gets left behind. Why do you think that is?

Sanchia: There are a lot of reasons: it’s too hard, they don’t know where it is, but the simplest reason is that people don’t even think about it. A lot of that is because data has traditionally been a back-office function, a bit mysterious, a bit Wizard-of-Oz. People forget about it. What used to happen when we were working on something for a client, or for a pitch, is people would say, “this is my theory, can you just find some data to back it up?” It was always an afterthought.

It takes a long time to break that and unlearn it, and part of it goes back to those three pillars: showing people that if you had this tiny crumb of data, you could do something significant, and it’s not that hard.

Another thing is that a lot of people are waiting for the data to be perfect. And I’m sorry to burst anyone’s bubble, but it’s never going to be perfect.

Amit: Explain that a bit more: what are they waiting for?

Sanchia: A perfectly clean database, with a perfect workflow, everything seamlessly joined together. I used to work at the White Company, and we had such great data, because it started as a direct-mail business, pre-GDPR, so people had to put in their correct information or they wouldn’t get their brochure posted to them. You had these really loyal customers. But when it’s up to the user to put in their own information, it’s not always going to be beautiful: people put in a throwaway email address they use for all their marketing, or something funny “because wouldn’t it be hilarious,” or they’re buying a present for their partner and put in the other person’s details.

So when you get into the nitty-gritty, you’re looking at lines of data where five records might be the same customer: mapping a customer to all of their orders, cutting it by category (did they buy bedding, did they buy swimwear), by date, by year, by marketing channel, connecting it to customer service, connecting it to segmentation, oh no, not that segmentation, a different one, oh actually they’ve migrated segments. Something as supposedly simple as one customer record explodes into a mad mess. It’s the job of the data team, the data infrastructure and the data governance to wrangle that back, but there will always be points where people put garbage into their personal details, where a customer service rep can’t update the latest record, where there’s a problem in the warehouse fulfilling an order. Even somewhere like Amazon, with amazing data, how many times have you ordered something and forgotten to update the address, and it’s gone to someone you sent a present to, or your wish list?

If you think about the actual volume of data people are working with, it’s not surprising it’s a mess, and it’s never going to be perfect, but it’ll be good enough in a lot of situations.

Getting Data Ready for AI Workflows

Amit: So if you’ve got all of that data, Amazon, or the White Company, and a team comes in, or there’s an internal request to say, “we need to get this into an AI-powered workflow, we need agents doing stuff behind the scenes,” what needs to happen to get that data ready for that next level? For some people that would be very scary. What needs to happen to get the data into a place where you can start building an agent workflow?

Sanchia: It’s all about the foundations. I don’t know if you’ve seen that picture on LinkedIn with the bridge and the train, I’ll send it to you after. There’s a train going across a bridge, and it’s like: this is an AI workflow with perfect data foundations. Then there’s the other one, where the bridge is a pile of rubble and the train’s falling into the canyon.

That’s how I think about it. You need the structure and the foundations, because if you want AI to work properly and not hallucinate (well, obviously you set the temperature to zero), but really, you need to tell it where to look.

Historically there are different ways of structuring data, but you’ll typically have separate tables: at the White Company you’d have a customer table linked to an order table, linked to a payment table. All of these sit in separate places but are linked by a unique identifier: your customer number links to the order table, the order number links to the payment. You’ve got this mad spider-web of things, but if you’ve got the structure and the rules, it’s much easier to put AI on top of it. You’ve got your foundations, and then the AI can say, “right, you’ve asked me this question, I’m going to go and find Sanchia’s order from 2024, where she bought this bedding,” and return it.

But you also need proper scrutiny: that’s where data fluency comes in. If it comes back and says “she spent £5,” you need people who know that’s not right: a full bedding set costs a lot more than that. Being able to do those spot checks, to have a good number sense: we’ve got X million customers on our database, it doesn’t make sense that only 1% would have bought our most popular category.

Amit: I think we’re going to have to do a part two at some point, you’ve touched on some things I want to come back to. We’ve got a couple of minutes left, so I want to ask one last question.

Sanchia’s Mission With Data Everyday

Amit: You started off talking about what you’re doing now with Data Everyday. I’d love for you to tell us what your mission is with data fluency.

Sanchia: At Data Everyday, we’re on a mission to make data accessible and understandable to everybody. With AI, making it accessible to everybody doesn’t necessarily make it understandable to everybody. If you’ve got the power to access data, and I’m not trying to be that old Spider-Man’s-uncle quote, but it’s true, there’s a great responsibility that comes with that power. So always think about what you’re trying to achieve, what your outcome is, and how you’re going to ask the right questions. That’s why data fluency is really important, and it’s only going to get more important in the age of AI.

Amit: That completely aligns with what we’re seeing across all the teams we work with. You’ve made some really good points about everyone needing to understand it, and the foundations and pillars you’ve given us today align closely with the approach we’re trying to take. Thank you so much for joining me today.

Sanchia: Thank you for having me.

Sanchia Neilson is Co-Founder of Data Everyday, helping organisations build confidence and fluency with their own data. Follow Re:Wire for more conversations on unlearning, workflow design and the future of work.