Talking CRAP On the Beach
Plenty of pizza and a well-stocked drinks cabinet – what better way to kick off the CRAP? Hosted for the second time by On the Beach in their brand new office in central Manchester, the conversation flowed freely as we relaxed in the beach-themed bar before the presentations began.
When like-minded people come together, it creates a buzzing atmosphere that’s perfect for an event like this. Looking around the room, it was clear that everyone felt comfortable enough to get talking with the person next to them. Nobody was left out or stood on their own; a credit to the inclusivity and welcoming nature of CRAP Talks.
The event space itself was decked out with deck chairs and brightly-coloured cushions. Once we were all comfortably seated, Becky Lacock – our host, and Analytics Manager at On the Beach – welcomed us warmly and introduced the first speaker…
Ceri – Music Magpie
Ceri’s topic was Data Visualisation and Presentation, with Tips and Practical Examples. His passion was infectious and his points were concise and clear – exactly what CRAP is all about. According to Ceri, data analysis is all well and good, but you have to be able to effectively communicate your findings to the mercy of the (often less technically-minded) HIPPO, or HIghest Paid Person’s Opinion. And that’s where problems can arise.
Armed with examples of good and bad graphs, Ceri showed how less is often more when it comes to explaining your bottom line findings. It seems that graphic design principles and simplicity can go a long way towards making sure your data is understood. With an engaging mixture of audience participation and practical examples, he showed how to streamline the presentation of data into easily digestible graphics and reports. Incidentally, CRAP Talks might be the only place where a speaker can ask: ‘What’s your favourite part of this graph?’, and get a genuinely enthusiastic response from the crowd!
Ceri’s advice is to ‘tell a story with your data’ – to make it coherent and clear by removing irrelevant information, keeping key messages in the foreground and avoiding ambiguity. When communicating data via email, he warned against ‘data essays’ that may cause the recipient’s eyes to instantly glaze over. He advocated a good BLUF (bottom line up front), otherwise known as making sure that you lead with the headline. Got something to shout about? Start there, and then get technical once you have their attention. Ceri ended his talk with a KISS – the advice to Keep It Simple Stupid.
Joe – Code Computer Love
Next up, Joe posed the question: Is Bandit Testing the Future of Experimentation? Named after casino slot-machines, or ‘one-armed bandits’, bandit testing is a variation of AB testing that uses an algorithm to push traffic towards the winning variant in a live environment. These split testing methods are often dubbed ‘multi-armed bandits’. Less than half of the audience had heard of bandit testing, and even fewer knew what it involves, and Joe did a great job of bringing us up to speed.
With a laid-back but switched on style, Joe cited the benefits of bandit testing. Firstly, you can earn while you learn – setting a test live means you can immediately start benefitting from the algorithm when it tips in favour of a successful variant. He also celebrated automation as another plus point – the testing effectively manages itself. Thirdly, he explained that stepping from exploration to exploitation is much quicker with this method of split testing because the winning variant already handles most of the traffic.
As far as drawbacks are concerned, Joe brought up the fact that bandit testing is often currently managed through third party products and therefore the infrastructure can be poor. He also advised that tests can take longer to reach significance and the results can create more questions in an optimisation programme. So when should multi-armed bandit testing be used? Joe suggests that it’s best suited to headline tests and short-term campaigns to maximise the effects of the campaigns as soon as possible, but bandit testing may well be the future of experimentation… watch this space!
The third presentation of the night – after a brief but delicious pizza intermission – was on Efficiencies of Learning – High Fidelity Prototypes. Dan has a PPC/web analytics background, which informed his views on prototyping in a really useful way. He explained that prototypes allow users to experience a chosen aspect (of a website, for example) in a way that’s as close as possible to a true reflection of the original user interface.
The pros of prototypes according to Dan include stakeholder excitement, the ability to gain intricate feedback and insight, and cultural buy-in for large projects, as well as timescale and budget efficiencies. He cautioned that the cons centre around the prototype not being a real-life environment. He also stated that users shouldn’t know they are using a prototype rather than the live website or service.
Usage of prototypes could be appropriate for riskier projects, where the potentially negative effects of live changes to a website can be avoided. Dan suggested that prototypes could also be useful for working with stakeholders who are new to experimentation, and might be reluctant to implement changes. Ultimately, prototypes are a great fit for big, bold ideas that would otherwise be deemed not ‘on brand’ enough to go live.
Daniel Liam Glyn –
And now for something completely different. Daniel’s presentation was all about how he crowdfunded, composed, recorded and marketed an album. Changing Stations is his 11-track record, inspired by the London Underground. Daniel has synesthesia, which means he associates memories and sounds with colours. His unusual perspective on the world gave him the inspiration to combine his colourful inner soundtrack with one of the most iconic pieces of design ever produced – the Tube Map and the corresponding colours and ‘moods’ of each station.
Using Kickstarter as his crowdfunding platform, Daniel described how he successfully pitched and funded his creative project, which gave him the money and time he needed to start composing. 11 piano pieces later, he had captured the feeling of each London Underground Line using a different key signature for each station. Now he just had to get his album out there.
As he was new to marketing and sales with this project, Daniel offered an interesting insight into these practices. He created his own record label and limited company, put his work onto online streaming services such as iTunes and Spotify and drummed up interest with social media. He says one of his proudest moments was seeing his album for sale in Piccadilly Records – one of Manchester’s most beloved indie record stores.
Until next time!
Thank you to all four of our speakers, who started interesting discussions with each of their presentations. How many events can say their topics cover everything from the future of technical experimentation to crowdfunded creative projects? CRAP Talks certainly can. Fancy joining us for the next one in Manchester? Sign up here.