Rethinking retail - Picnic delivers on a data-driven approach

Picnic started its retail business in 2015 in one Dutch town, where it decided against the need for physical stores and took an ‘app-only’ approach to food delivery. The retailer has grown significantly since then and now operates in over 60 cities in the Netherlands,  Germany and France, delivering groceries via electric vehicle, and claims that the secret to its success is “an end-to-end business and just-in-time supply chain”. 

With a team of over 200 people, Picnic has taken a digital-first approach to operating its business, with some noteworthy characteristics. Taking an Amazon-like approach to product development, the organization works in cross-functional teams where the entire supply chain and all software is run in-house. 

Everyone working at Picnic owns their part of the business, but the focus is ‘products’ - whether that be the warehouse management, purchase order, or customer facing systems. And each of these products is handled by a complementary team of front and back-end engineers, all focused on delivering the best possible customer experience. 

Central to all of this is how the retailer uses data to guide development and deliver on customer needs. Picnic sees it as essential that everyone throughout the business has access to the data that they need, which is held in its home-grown operational systems, to understand where problems may lie and how things can be improved. To deliver on this, Picnic has created a joined-up data platform using Snowflake’s Data Cloud, which Frank Gorte, People Development Lead at the retailer, says has enabled a strong data-driven culture. 

Gorte says  that every analyst – regardless of whether they’re working in the supply chain, fulfilment or the commercial team – has access to the data. The platform also creates full transparency around which data is being used for what purpose. Gorke gives an example and says: 

The commercial team might make business decisions based on supply-chain impact. They might see that this is a product where customers have made complaints because there’s a break in the supply chain somewhere due to the packaging being too fancy. Then they can contact the supplier and ask them to change the packaging.

This joined-up approach means that data flows across the entire company and people across Picnic are aware of the impact of development changes they’re making, thanks to the regular data insights. For example, Picnic has a variety of data sources being ingested into the platform and those working on the supply chain and products can see how the groceries are being impacted on the ground in real time. Gorte explains:

For example, we have a supply chain with frozen, chilled and ambient products. You need to be careful with temperature monitoring. Snowflake allows us to build the pipeline and get the data flowing in real life immediately.

Delivering data-led change

As noted above, Picnic develops most of its software in-house. And the retailer sees this as a key benefit, particularly as it advances its AI and Machine Learning ambitions. Gorte notes, for instance, that because the entire supply chain is within the company’s control, it can do interesting things like use Machine Learning models to reduce waste, passing on insights to suppliers for them to take action. 

Data is central to all of this, of course, and another predictive feature the company is working on is called ‘Driver Coach’, which is aimed at improving driver safety. Picnic’s delivery specialists, known as runners, have rugged Android smartphones that they can use to scan items during their work, but the retailer has developed applications to nudge the drivers into better behaviours. Gorte explains:

The device also measures speed, braking and can also detect unsafe driving conditions. It acts as a coach and our runners get a safety score at the end. This is personal feedback. If they’ve gone through a corner at a speed that is less safe, it will tell them. 

We can then aggregate all these views and create maps of cities where speeds are generally a little lower, which means it's more likely to be an unsafe point. And we can use that data to create a safer driving environment.

Looking forward

Picnic has utilized automation and automated tooling for years, given its focus on just-in-time supply chains and providing customers with a positive digital experience. Gorte has played a key role in this over the years, and helped design the company’s approach to automation. 

Picnic today has two automation concepts – one is fully automated and uses robots and conveyor belts; the other is a hybrid solution that mixes automation and human pickers. One of these concepts is selected and deployed when Picnic moves into a new market. 

The Snowflake platform will also play a pivotal role in future automation capabilities, as it provides a capability for teams throughout the business to build machine learning models and use the data first-hand. Gorte says: 

That would mean they don’t have to go through a data scientist first, unless it’s for the really big models. That kind of wider access will help us to make sure that business teams can sandbox and innovate themselves, and then leverage the knowledge of the platform teams to use machine learning.

One example of how the retailer is planning to use AI/ML is to measure employee sentiment. Picnic already uses Slack to ask employees how they're feeling every week, but Gorte says the survey uses a set of pre-defined questions, but that it would be  more interesting to use Large Language Models to refine the questions subtly and test hypotheses on how employees are feeling. He’s also keen to use AI and ML to identify trends in performance reviews: 

That could be super-exciting. We do performance reviews every six months with the employees. So, that's a super-rich data source. To be able to leverage that information to be directly on top of the trends is going to be a game-changer.

In terms of advice for other business leaders and organizations thinking through their digital development, Gorte says that it is key to get senior executive buy-in and to build a strong grassroots movement as quickly as possible. He argues that once people see the benefits that data access can bring, everyone will get excited about the features. Gorte concludes: 

If you can identify a part of your company where you can find a broader set of people who are willing to invest a little bit of time in getting to know the technology, and who immediately reap the benefits of the accessibility of data, then Snowflake becomes an unstoppable force.