Hear about what a data scientist gets up to in their first few months at Deliveroo
Kia Ora! I’m Florence, a Senior Data Scientist at Deliveroo. Originally from New Zealand, I moved to London eight years ago and have been working in data ever since. I joined Deliveroo at the end of last year and want to share what I’ve learnt and been up to since.
Why I joined Deliveroo
My interest in Deliveroo can be summarised into three parts; my enjoyment of eating and getting to think about food all day, the complexity of the business model, and the advanced approach to data needed to remain competitive as a company.
The first reason is pretty self explanatory, I am yet to prove to myself that I can cook but I certainly enjoy eating good quality and varied meals throughout the week. Starting my career in the food tech and delivery tech industry at Ocado showed me what a fascinating industry it is, it also confirmed that nothing makes you more hungry than thinking about food all day!
I really enjoy being able to see different business models from the inside and learn the subtleties of how they actually work, so the three sided marketplace of Deliveroo really fascinated me. The three sides are the customers (which I was a bit too familiar with from my ordering habits), the riders and delivery, and finally the partners (restaurants and grocers). How all three of these interact adds plenty of additional complexity to any simple analysis. For example if you wanted to run an experiment where a customer could choose to have their bread substituted for a similar loaf you’d need to consider all sorts of things that you’d never think about as a customer: 1. How the customer interacts and chooses this within the app, 2. How the grocery partner would choose what a suitable substitute would be based on stock availability, 3. What would happen if the customer rejected the substitute to the rider on delivery. I couldn’t wait to get stuck in and understand the levers to drive growth at each of these stages.
Here’s a taste of some more questions and discussions that I have enjoyed being a part of throughout my career working in this industry. What items are most frequently bought with pasta? Cheese or pasta sauce are obvious answers, but how about red wine or basil? How does having eco-friendly delivery options affect the customers likelihood to choose certain delivery slots? How should premium brands, such as Le Creuset and Dishoom, interact with the changing landscape of online delivery and startup brands? These types of relatable questions intrigue me and made me very keen to be back analysing.
All of the above require very complex and nuanced data points to understand what truly happens. Having spent the last few years focusing on analytics engineering, I was excited for where I could add value and contribute. I had actually interviewed at Deliveroo in 2017 when it was much smaller and split across many small offices, but at that point I didn’t have the specific data science background that they were looking for, and so after specialising and learning another business model I thought I’d come see what it was all about once again.
My first few weeks
There are many highlights to pick from - joining a new and ever-growing team, presenting to Directors, joining Deliveroo’s first hybrid hackathon, as well as teaching and writing a doc on testing clickstream data in Kibana. However the two examples I really want to share in detail are the Deliveroo Tech Academy and my visit to the Islington Deliveroo Editions site.
The first two weeks of joining are spent in the (currently virtual) Deliveroo Tech Academy along with any other engineering, product and design new starters. The first day kicked off with computer setup, a general introduction to the organisation and a discussion on DEI (diversity, equity and inclusion) from Busi Sizani, our Head of Global DEI. The following days covered more technical topics, such as how to run cross-geographical experiments, deploying releases with feature flags, details into mobile app development and incident management. Having two weeks to absorb information with no expectations of contributing to project work is a conscious decision at Deliveroo and felt very person-centered, where the focus was on individual development and getting a good grounding in the company. It also meant that at the end of the two weeks everyone was super keen and raring to go. Following the tech onboarding I officially joined my team, which is the team focusing on Deliveroo’s grocery efforts, and I was given my first project! The grocery team is a fairly new area for Deliveroo compared to the more mature restaurant side of the business, which means that there is a huge amount yet to be built and understood, so being comfortable with the unknown is essential - luckily that is exactly how I like to work.
As mentioned above, I am no chef but that doesn’t stop me from watching cooking YouTube videos and being super keen to sign up to visit a Deliveroo Editions site. Editions are hubs where we host collections of restaurants, all specially designed for delivery. This meant for a few hours I got to don my chef’s hat, mask, apron and be given a grand tour of how numerous restaurants run out of a creatively built neighbourhood warehouse. My two favourite learnings from my visit were that to be an Editions restaurant you need to have 5 stars on the food hygiene rating scheme, and that the site managers use analyses and reports made by data scientists to ensure all the restaurants under their site are performing at the expected level across all our key metrics - it was great to hear the experiences of other colleagues using data science outputs outside of the main product teams. To top it all off, at the end of the day I was able to choose a takeaway meal from any restaurant within that site - delish.
What was surprising
Historically, I have been guilty of looking jealously at a larger tech companies and thinking that all the data and workflows would be fully polished and pristine, but I quickly found out that to cope with its enormous growth Deliveroo has been in “scale and survive mode” where everything being implemented tidily and with best practice wasn’t the highest priority. This has meant that people from all ranges of backgrounds, whether that be a startup or an established business have a huge amount to contribute and bring learnings from elsewhere.
Another thing that hasn’t necessarily surprised me, but has been reassuring is that despite growing the data science team from 45 to 150 people in one year the “scaling with kindness at front of mind” pursuit has certainly been achieved. The team has been so curious, kind and welcoming. I think having the majority of people join remotely has also helped create the “we’re all in the same boat, and know that a friendly face at the end of the screen really helps”.
Autonomy and trust is really important to me
When I am interviewing for a new job, a question I try to ask each new person I meet is “can you share a time you had an idea that you believed in and were given the freedom to explore?” Something that has always been very important to me is to have trust and autonomy to dream big and suggest ideas instead of purely relying on senior management to dictate the path forward. Those interviewing me provided an assortment of examples, and in my first few weeks there was an area of opportunity to improve my team’s understanding of clickstream data between Segment into Kibana and then on into Snowflake in a more efficient and realtime way. This was an area I have a lot of experience in and so I jumped at the opportunity to help upskilling with the team and creating documentation to share around the wider business. Another example that has been supported is adding additional latency type metrics as a monitoring metric across variants in an experiment, this is an area I have previously found incredibly valuable and results in large business value.During various discussions I understood the opportunity in more detail and connected dots of people interested in this across the business and am looking forward to having a look at it this year.
My favourite data insight so far
As I’m sure many people do, I like to tell myself that I make unique choices and am individual in my food tastes, however when I joined I was able to look at the most popular items from my favourite local Greek restaurant. Little did I know, despite having thousands of meal options available, my favourite lamb and tzatziki wrap wasn’t so unique afterall. In fact, it was one of their most ordered items! Not so original, but that doesn’t change how yummy it is.
One thing I want someone thinking of applying to know
It can be hard within the data and tech world to find organisations which value the individual contributor (IC) career path equally to the manager career path, let alone at all. This was a huge factor in my decision to join - Deliveroo actively encourages and supports the Data Science IC career progression all the way up to director level. This means that there are very senior ICs leading the way with complex analyses and providing upskilling opportunities for more junior colleagues. I really feel that my desire to “code more than sit in meetings” and keep up to date with my analytical abilities will be possible and supported without question. If this is the kind of career path you are looking for, I would definitely encourage you to apply. Come join us!