BBC Data Internship
In the middle of the summer break I participated in an amazing opportunity at the BBC. I was placed within a centralised analytics team. Being centralised, they take on requests and projects from all across the BBC, whether it be News, Sport, iPlayer or Sounds. All the work has a focus on understanding audience behaviour, finding out how product or editorial changes affect audiences and constantly trying to make the audience experience better.
My first day got to take place in-house! I arrived at 10 O’clock at BBC Broadcasting Centre in White city — actually let’s be honest, I of course arrived at the wrong building in the huge complex and had to be directed a further 10 mins away. My first day was filled with intros, laptop issues and getting lost on multiple floors. It was nice to have met a few colleagues in person as I would not see them for the remainder of the internship.
I worked under the Analytics manager of the team, with help from the other analysts depending on the project. Each member has their own specialty, whether it be data manipulation, dashboarding and visualisation or reporting, but they all tend to work on similar projects.
Day to day I would work on any open requests, my own internship project, team duties such as manning the email inbox and working through learning and development goals. Often this involved attending BBC wide webinars, using Linkedin Learning and the online BBC Academy platform — whilst also grappling with the pitfalls of using a 10 year old operating system.
I worked on 4 main projects during my 6 week stint. My first job was assigned to me early on to help out a colleague. I will have to be a little vague as I have signed an NDA, so bear with me! This involved using an analytics platform called AT Internet to query data and present to a stakeholder from another team. This job had originally been assigned to another junior colleague but as it was pretty entry level I joined the job. We met with the stakeholder to understand her needs and requirements, the deadline (pretty loose) and to also gain some background as to what it will be used for. It seemed they would be making some product changes and needed to get data to understand where to implement the changes first. I found this part key for me as it led to a greater understanding of the kinds of questions the analysts are seeking to answer; how analysis can be used to grow and change historic organisations such as the BBC. I soon set to work on getting to grips with the platform and began downloading data in CSV format. The stakeholder wanted some aggregation to be done to the data so I utilised my knowledge of Excel logic formulae to create an easy to understand chart. The stakeholder was not an analyst so it was quite key to ensure my report was clear. I worked on this project for a couple of days, quite soon we realised that there was an error in the data from AT Internet, this meant we had to access the data stored in AWS Redshift tables instead. Unfortunately I was not able to gain access to AWS Redshift as the onboarding process would have taken too long, so my colleague wrote a SQL script to query the data there. He shared the script with me and gave me a crash course in Standard SQL. By this point I had also completed a SQL training course on Linkedin Learning, on request from my line manager, so it wasn’t too hard to get my head around it. My helpful colleague pulled the data for me and left me to use pivot tables and functions to present my chart to the stakeholder. At the end of the day she was happy with the data, although she did have a couple of questions and a few demands along the way. My colleague mainly left the communication to me so I had to manage the relationship and expectations as best as I saw fit. This was tough at times as I was tempted to say yes to everything and make minute changes at each email but my colleague gave me the confidence to voice my own ideas and at times disagree with her. I’d say I definitely got the chance to lead and persuade senior staff.
My next project gave me the opportunity to exercise my newly honed SQL skills. Again working on a request from another team I was to create a query in SQL, pull data from AWS Athena and then present using Excel (I did complete a Tableau Linkedin Learning course and played around with this visualisation tool but Excel proved effective enough). On this one I worked directly with my line manager. In the grand scheme I think they wanted to understand whether a very common video player feature (which shall not be named) was effective at increasing user engagement. This project took up a lot of my time, mostly because of my struggles with SQL. I would spend hours trying to figure out how to do aspects of the query so this slowed me down tremendously. I was able to ask for help but sometimes the support was not as strong as I would have liked and I felt a bit alone (especially during remote working) but it meant most things I eventually figured out on my own, which is definitely something to be proud of. I worked on this project right till the end and presented an excel spreadsheet and simple charts. There were a number of hiccups with the data format so I got a bit of practice at data manipulation too.
Another mini project involved organising the team to update documentation on the organisation’s internal Confluence site. I created the directory and added pages in the appropriate area, liaised with all members of the team and ensured everyone did their part. I also did some critical analysis to make sure that the documentation was accessible to even non-analysts.
Lastly I took the opportunity to conduct my own project for the duration of my internship. I worked on a piece of editorial content and tried to provide some insights for marketing teams. The work involved a lot of project planning and management; I scoped out the project, wrote ideas, goals and drafted a timeline to share with my manager. I made use of the array of tools Dropbox Paper has to offer. Full of ideas I started to build the query and again SQL got the better of me. It took weeks to finalise the query and I definitely thought it was not going to get finished. In the eleventh hour I managed to pull my data, load into R, transform the data into binary and run a logit regression. I created a presentation using the great tools of Dropbox Paper and presented to the rest of the team who were all impressed with my work. This is the piece I’m most proud of by far. Many times I thought it would never get finished and I would have nothing to show but in the end I managed to do it.
Throughout the placement I was able to use my time to learn about things I was personally interested in. Through email chains I was able to find webinars on machine learning and artificial intelligence, sit in on BBC Strategy meetings, book in one-to-ones with staff from data science and had plenty of opportunity to see showcases from my team and other teams. At least once a week there was some kind of show and tell where analysts presented dashboards, charts and presentations used to make editorial or platform decisions. It was great to learn about topics I was more interested in and also see some amazing projects across analytical teams. There was definitely a lot of opportunity to learn.
Despite my projects and the great learning opportunities the internship could have gone better. At the start I had to go through a lot of onboarding strife: laptops, access issues, requesting IT; this all really slowed down the start of my placement and took up a lot of time during my short experience. And probably heavily related to the virtual working, I never really felt a part of the team and felt the support I needed to do my best work was not always there. This is a shame as I was incredibly thrilled to do this internship but was often held back by a variety of factors. The work I was given mostly involved pulling data and handing it to another team, which objectively is a little dull, I would have liked to have gotten more involved in the analysis and visualisation to really test my skills. I think my issues with the internship were related to technical constraints, the short timespan of the placement and the fact that it was remote — none of these were really in anyone’s control, I realise that now. However, I did learn a lot from the experience, namely that I feel more comfortable with data science and analysis as opposed to analytics. Seeing presentations from data scientists explaining their A/B tests, segmentation, how they build recommendation systems and show causality just connected with me more than the day-to-day data querying and Excel use.
Overall the experience was immensely useful and I really did get to see a lot and use a variety of tools that will most certainly help in future data science applications. It was not perfect but maybe my expectations were too high. I would still be interested in seeing what kind of roles the BBC has to offer in future.
