How Huel leveraged self-service analytics to upskill its business users - and free its data team from dashboard hell

Huel, a privately held UK food brand that's all about sustainability, says democratizing access to data internally has not just improved overall data literacy across the business but also opened the door to faster, evidence-based decision-making in a highly competitive market.

Specific business benefits the company can directly attribute to the move, according to its Director of Data, Bhav Patel, include savings of 100 manual hours per month on analysis - with 75% of that from self-service search - and reducing the time it takes to get to useful insights from hours to minutes.

The move is also enabling 150 lines of business colleagues - who are all, of course, proud 'Hueligans' - to deliver high-value, customer-built reports back into the business.

These business users, Patel adds, are now much more data-literate, as they are now familiar with data joins, aggregation, and key business metrics like Lifetime Value (LTV) - the total revenue a customer will generate for your business over the entire time they stay with you - Customer Acquisition Cost (CAC), and tracking retention, meaning the percentage of customers who stay with you over time by continuing to subscribe, buy, or use your product.

While training and vendor support have for sure been part of that evolution, in practical terms they can answer both useful operational questions and perform exploratory analysis without the need for back-end Information Technology (IT) support.

The company's Head of Performance Marketing, for example, used to request multiple custom dashboards. Now, Huel claims, they ask for more data to work with on their own.

That's a revolution, Patel says, that very much helps the company:

Being able to have that information available instantly means that, as a data team, we're no longer a bottleneck. So, we as a company can basically work at pace. As one of Huel's core values is around hustling hard and working fast, access to this data effectively allows us to live and breathe that.

Moving from Tableau to agentic analytics

Patel says this has been largely achieved through moving from Tableau to working with agentic analytics from ThoughtSpot - whose Spotter agentic Artificial Intelligence (AI) system, which diginomica recently report on,  and which is also on Huel's radar, as we shall shortly see.

He explains:

With our data tools back then, people would have to go to meetings, get questions, then have to pull the data, analyze it, and bring results back a week later. Now, analysts spend less time building dashboards and more time enriching datasets and doing deeper analysis - and so can self-serve and easily get the data they need.

'No single source of truth’

Patel works for the business-to-consumer (B2C) company set up in 2015 to offer "nutritionally complete" plant-based meals, snacks, drinks, and food supplements.

Striving for minimal impact on animals and the environment - it achieved B Corp status in 2023 - its name is a contraction for 'Human + Fuel.'

The company claims over 250 million of its meals have now been sold in over 90 countries, and that it reached £100 million revenue last year.

The road to data analytics self-service, Patel says, started four years back, with ThoughtSpot coming on-stream in May 2022.

At that time, the company was trying to analyze business problems out of data silos and had no single source of truth. Solving problems was slow. He says:

Right from the word go, data was very much a part of our growth and success, but because of a bunch of different reasons the platform we were previously using just wasn't fit for purpose. So, we went out to the market to try to find something that would meet the growing needs of us as a company.

A particular bottleneck was the fact that everyone across the business had access to its then BI tool. Multiple dashboards supported decision-making, and adoption was strong.

But before long, analysts were spending huge amounts of time building and formatting dashboards, pulling data from different sources, and dealing with more and more requests for data help.

A clean, modern data stack

That was taking a lot of effort on both the company's analytical and data engineering side to build, while remote working could be a problem due to Virtual Private Network (VPN) issues.

In response, the team Patel is now part of, commenced building what he characterizes as a 'modern' data stack.

This comprised of, at the 'bottom' of the process, Fivetran and Clean as the primary new Extract, Transform, Load (ETL) tools.

Depending on the use case, both are used and run over sales, marketing, customer experience, finance, and other data, with the output then loaded into Snowflake as the main Huel data warehouse.

That data warehouse is now the 'single source of truth' for the company, with dbt then acting as a transformation layer, letting the Huel data team clean, join, and produce clear datasets for end users.

On top, the company uses DataRobot for Automated Machine Learning (AutoML) predictive analytics, and the whole company now relies on ThoughtSpot for analytics and self-service - or as Patel sees it, with it acting as a layer over that clean data that allows the Hueligans to "go in and get that data and access it and make decisions off the back of it."

Once the new data stack was in place, the value of not just self-service of data but data as a business asset was rapid, he says.

To use just one metric, in February 2019, out of 70 total staff precisely one had a role as a data analyst. As of late 2025, the organization has a full-time Data Director, three analysts, and three Data Engineers out of a headcount of nearly 350.

Next steps: unleashing creative flexibility with AI

It's clear Huel has started to leverage this new way of working with data very effectively - but what's next on the horizon?

AI is definitely going to be part of the picture, Patel says - and the vendor partner's Spotter could be a big part of that.

He explains:

We're not going to see a huge change in our underlying data stack. What the data team is trying to advocate is the use of more automations and AI here. At the risk of sounding 'AI, AI, AI', if everyone is able to self-serve from a dashboard's perspective, that allows you to get insights in a way that is quite rigid and structured - but we want to explore the idea that by using Spotter as an AI element built into ThoughtSpot, we could unleash that creative flexibility of stakeholders who might want to ask more nuanced questions that we can't templatize in the form of a dashboard.

He adds:

After all, when you combine the dashboarding element of the platform with the ability to talk to that dashboard, you start to create a conversation from a data perspective. And with AI, by not having to constantly build, build, build, we could increase the amount of output we can deliver. All we have to do is maintain an underlying dataset and allow people to just ask questions of that dataset - and I think that's really exciting.

Summing up his experience so far, Patel believes that ThoughtSpot has enabled self-service, improved speed to insight, raised data literacy - and freed his analyst colleagues to deliver higher-value insights without any need for his team's constant involvement.