Weeknotes: Sprint 15

Our latest sprint has been focused around our publishing platform, Florence.

During the website build we spent a lot of time testing Florence with our publishers, using the same approaches to give them a product that meets their needs as we used on the site itself. Whilst we have made a lot of changes to the system since we went live, to enhance functionality or fix issues, we have not reviewed its usability more generally now Florence has been in production for a while.

With this in mind I felt it was important to take a sprint out to look at some of the issues that, whilst they maybe don’t affect our ability to publish, made life more difficult or frustrating for our publishers. It was also a chance to refresh ourselves about how it all fits together and start to look at how things might need to change as the system develops to support changes planned in the roadmap.

We worked closely with the publishing team to identify some of the pain points of the system with a view to fixing as many as we could in a single sprint. As part of this we reviewed all previous requests and had a mini workshop to discuss additional ideas and possible solutions. This was then prioritised by the publishers to give an idea of what caused the most problems.

The changes we have made are varied and range from incorporating the manually clearing of the homepage cache into our publishing process, to bug fixes, and streamlining of navigation round the system.

We were obviously unable to fix or change everything asked for, but were able to cross a lot of things off the list and also have a much better idea of where some of the difficulties are likely to occur as we make improvements to the main website.

If you have any thoughts or comments on any part of the website please do let us know by email, the comments on this blogpost or on Twitter to @ONSdigital.

Social Media Week 2016

Last week (12 to 16 September 2016) was Social Media Week in London so I went to find out the latest trends and to get ideas.

Two of the talks I went to were about data and were, by far, my favourite sessions. They  got me thinking about how we do monitoring and analytics here at the Office for National Statistics (ONS). I’ve just finished a paper about this so it’s interesting to discover the similarities and differences in approaches.


First up was a talk from Anthony Fradet, Chief Operating Officer for Linkfluence. The talk discussed the definition of social media influence. The measurement of influence is usually linked to a number: how many followers an account has, how many shares or retweets, how many engagements or mentions, or how many impressions.

It was refreshing to hear Anthony share stories about how numbers are often linked to success but are usually completely irrelevant … have you ever watched the number of likes a post gets and then used that as a measure of success?

Anthony talked about going beyond meaningless numbers and looking at social influence in context. This is where it got really interesting!

The talk discussed the difference between interest and influence. Just because you are interested in something, doesn’t make you an influencer on this topic. This means that bios or content you post on social media demonstrate your interests, not your influence.


To map influence, you need to map the audience through links. Links to and from accounts inside a relevant network creates a better measure of influence in a topic area. An example could be links on websites, through retweets or mentions on Twitter.

To finish Anthony’s talk, he made the statement that influence is not scalable. Some influencers are massively influential about one topic in a small community, others are influential in broader subject areas with a larger online audience. When reaching out to influencers, one is not more or less important than another. To engage with the right influencers for you depends on your objectives.  

In summary, it’s all about context!

Building data teams

The other data talk I attended during Social Media Week was given by Hollie Lubbock, Associate User Experience Director from Code and Theory. Helpfully, Hollie has done a write up of her talk so I will skip straight to what I found most interesting. This was the concept of using varied data sources to build a data powerhouse! This is something we want to achieve at ONS.

Hollie’s talk focused on the value different backgrounds and expertise can bring to decision making. When she sets up data teams, she puts the User Experience (UX) or User Interface (UI) designers, social media people and web analytics people together so the whole user experience is thought about and has an input into the overall decisions being made. This multi skilled team has the ability to deliver the whole picture. Joining up data “touch-points” in this way adds value to teams, clients, designers and strategists that make decisions within an organisation.


I’ll finish this blog post with Hollie’s “10 commandments of data”. These have been developed from her experiences of different backgrounds often resulting in conflicting opinions.


If you’ve got any questions or want to get in touch about social media at ONS, please tweet me or email me.

Weeknotes: Sprint 14

Our latest deployment to our live environments has just taken place so a quick update on what we have been up to.

Delete content functionality

As mentioned by Andy in a previous post we had enhanced Florence to include the ability to remove content from the website through our CMS rather than a manual task. This has now been made available to our publishers in the live environment and we spent some time this sprint making sure this worked for them. We had held off deploying this straight away to give our publishing team the chance to test and familiarise themselves with it, but also to think about the scenarios where deleting content is appropriate.

Embed the chart

We have received a number of requests from users wanting to know how they can reuse our charts and this has been something we have always wanted to offer. Each chart is stored separately as its own object dating back to our thinking around this when we created them.

We have spent time this sprint working to allow this, initially for our visual.ons site, but with a view that what we really want to do is extend this functionality to users. We will be looking to get this live shortly and then the next steps will be look at how we can  make this functionality clearly available to users, without breaking the flow of our content.


For a while now we have wanted to look at a more elegant solution for gathering feedback on the site. The current approach has worked well but the orange button we have feels disjointed from the page and we want something that feels more in keeping with the page itself.

We also want to try and gather more information that, ultimately, we can use to measure the performance of the website and user satisfaction.

There are a couple of processes we need to change with how we manage feedback before this goes live but this should not be to far away.

Data discovery spike

The team have also spent some time this sprint looking out the outcomes from recent exploratory work into multivariate and subnational data to get a better understanding of the changes we will need to make to the site to support this.

If you have any thoughts or comments on these changes or any other part of the site please do let us know by email, the comments on this blogpost or on Twitter to @ONSdigital.

The ONS web roadmap

aka – the one where I talk about road maps

Roadmaps are good. Public roadmaps are better and public roadmaps where we talk about how they evolve as we go along are the very best of all.

This should be the permanent link to our roadmap. We don’t have an internal version of this, so (and this is a little scary for me) we are continuing to do all of this in the open.

The themes are at a very high level with more detail where we have it, but I think they begin to show the scale of the challenges we are looking to address.

Essentially, we are looking to improve the performance, security and scalability of our platform whilst iterating the audience-facing web experience and adding in the support for local area and multivariate data.

This last point is key. It is our top priority for the year and it means we hope to quickly move from offering the static serving of bulletins and compendia to a geographically personal offering that means the information provided can be put in a much more meaningful context for each user (it also means our caching model will break, our file storage system needs to be swapped to a database and the way we deploy code all needs to change before we can really start any of it).

Alongside this, we are looking to iterate key components of the service experience and try and be even more open in what we do.

Some highlights that we are looking at in the short term.

Meta, Metadata

As part of the GDS Service Standard (which we choose to follow) we need to provide public performance data. One of the challenges we have faced with this, is that this performance data is easier to articulate for a transactional website (how many people signed up for X, how many people downloaded Y), which is not how we are set up. We are exploring some options around this and I hope the team will be showcasing some really interesting thinking around this (and a working dashboard) very soon.


Our audience loves a time series chart. This much we know. We are also choosing to invest time in significantly improving our chart experience. Key areas of focus are around better support across device and the simple need to support more chart types.

Work on our API

Our website is our API, but unless you built our website, our API can be a little tricky to understand. To try and help with this, we are going to develop some much more substantial support around the API and include sample applications and code libraries to help ensure people are more able to use our information in the ways they wish.

As you can see, the roadmap is pretty full. This is an essential component of the work we have in defining a service. Continuing to iterate at pace to continually seek to better meet the needs of our users.

the challenge is service transformation GDS poster
the challenge is service transformation

We have bold ambitions to update the way we present content, inform the users we are aiming that content at and consolidate the overall experience of interacting with our outputs online. It is going to be a busy few months, but please do ask me questions and let us know your feedback.

Weeknotes: Sprint 12/13

I have been back in the office for a few days so here is a quick update on what we have been up.

We had stacked up quite a few changes locally whilst I was off so part of the focus has been getting all these live and out to users. There probably a couple of item the team have been working on that I have missed but here are the highlights from the last couple of weeks.


All of our work around updating PDFs has now been completed so you should see a lot more consistency with these across the site. This includes a complete compendia PDF and the ability for our statistician to append additional tables to the back to help users now across all our content types where previously this had been limited.


We have implemented a solution using MathJax to allow us to render equations on the site. This is something we have been keen to get working for a while but needed a rethink in the way we planned to implement it.

Performance platform

A significant chuck of the past couple of sprints has been looking at getting our performance dashboard up and running. We are not quite in a position to share, partly due to a couple of last minute bugs but also while we wait for some live production data to populate the charts. We will be doing a separate post about that on here in the next week or so.

screencapture-performance-ons-gov-uk-1473066493678 (1)

User research

Our researchers have been busy looking at the next round of lab testing, reviewing our personas and extending these to some of our internal users.

We will be back on the road in London next Tuesday, 13 September (we still have slots at 11.15 and 15.45 – if you are interested and available to help, please contact Alison) and will be focused on  testing our recent changes to time series as well as validating some of the wider data journeys.

We will also be doing our annual user satisfaction survey soon and have a shorter, 2 question, survey coming out this week which is supporting some changes we are currently working on around how we ask for feedback on the site.

If you have any thoughts or comments on these changes or any other part of the site please do let us know by email, the comments on this blogpost or on Twitter to @ONSdigital.

SlideShare at ONS

ONS has had a SlideShare account since 2013 with most of the content uploaded until recently including made up of event presentations, documents or infographics.

In March 2016, we decided to try new content types to see what does and doesn’t work in SlideShare. Since then, we’ve published 8 presentations onto the platform with varying themes and formats. This blog shares what we’ve learnt so far…

Why we tried SlideShare

SlideShare was founded in 2006 and joined the LinkedIn family in 2012. since then, the platform has grown into a platform for publishing professional content. It has 40 content categories and one of the top 100 most-visited websites in the world.

There’s an average of 70 million people browsing the SlideShare website monthly with a predominately business audience including engaged business owners. Similarly to other social media platforms, SlideShare has a varied audience but has up to 5 times more traffic from business owners than Twitter, Facebook, YouTube, and LinkedIn.

In the context of our user personas, SlideShare works best for the Information Forager:


What we did

Since March 2016, we’ve posted slides from the Public Policy Forum, Economic Forum, reformatted an article originally published on our Visual website and created bespoke pieces of content.

The highest reach of any of our SlideShares was the’Young People in the UK‘ post, with 18,576 views.

This chart shows the total views of all SlideShare posts to our profile over the past 6 months.

Total SlideShare views
Total views of all SlideShare posts
The most interesting comparison for me came from our SlideShare on e-cigarettes. This had 3,816 unique views but only 855 unique views on the visual article. This demonstrates the reach of this platform may be wider than ourwebsite for specific types of content.
Total views for the 'smoking' SlideShare
Total views for the ‘smoking’ SlideShare

Looking at other SlideShare presentations from the past few months, it is interesting to learn how people interact with the content. Below you can see how people, out of a total of 6,506 views,  engage with the content:

Screen Shot 2016-08-23 at 15.31.21
Engagements with the skin cancer SlideShare post
The analytics available to monitor our performance within the SlideShare platform are limited but provide a ‘high-level view’ of reach and engagement, allowing us to track and compare basic metrics.

What works

We’ve come up with three basic recommendations for the types of content we will share on SlideShare. These have been developed from a mix of our recent findings and from best practice guidelines for the platform. As always, we will continue to review and test new types of content.

1: Event slides

SlideShare expert, Julius Solaris said “a presentation that shines on SlideShare is not a presentation you made at an event, it is made for SlideShare”. With this in mind, the main focus for content on SlideShare should not be event presentations or slides from conferences.

We should restrict event slides to only large events that are relevant to a wider audience, not just those who attended the event; for example, presentation slides from the Economic Forum or our Public Policy Forum. Slides from events posted on our SlideShare profile should appeal to the Information Forager user persona.

2: Storytelling

Storytelling content works very well on SlideShare. Content written in a storytelling format should have short, bite-sized chunks of information knitted together to tell a story about a particular topic. There should be a clear theme throughout the SlideShare deck and each slide should build on the last to give a complete story by the end of the presentation. 

This content format works well if you are trying to address a problem or issue; attempting to answer a question; or tackling a specific subject to teach.

Our Skin Cancer in England and Young People in the UK SlideShare posts follow this format.

3: Listicles

SlideShare can be used to share a short list of key facts or “top statistics” about one, specific topic. 

An example of a SlideShare we have posted in a listicle format is our 8 Facts about the Environment post.

What next?

We’ll continue to test the platform with various types of content to grow our knowledge in how to optimise content for this audience.

The recent purchase of LinkedIn and SlideShare by Microsoft is a potential threat to our future use of the platform; however, following the buy-out of Yammer by Microsoft a few years ago, we’re not worried… time will tell!

We’ll continue to monitor our social media accounts and optimise, prioritise and develop our social media presence based on feedback from our users.

To wrap up, here are some top tips for using SlideShare:

  • Slides should be 1024 pixels wide by 768 pixels tall
  • All presentations need a unique title, description and a minimum of three tags
  • Content should be unique to ONS (or your organisation) so there isn’t competition for ‘views’ inside the platform
  • File sizes are optimised and reduced to make the presentations suitable for sharing on other social media platforms (such as Twitter)
  • All uploads should be PDF
  • Each slide should have only one specific focus. You should avoid having more than one fact on a slide.
  • Content should be kept to a minimum and each slide should have powerful imagery or charts. Readers on SlideShare respond better to graphics so they should feel as if they’re reading a visual article or storybook as they consume the deck.
  • Hyperlinks should be available on every slide after slide 3 to give context to the reader and easily allow navigation to more information.
  • Include a call to action at the end of the deck to allow the user journey to continue.
  • It’s important to hook your audience as fast as possible. Ensure that the headline on your first slide is compelling and has the ability to lure your audience into wanting to read more
  • The text on every slide should use large fonts
  • The cover slide should not be plain white, black or a solid colour but as bright and eye-catching as possible.

Take a look at our SlideShare profile and email us your feedback!

Should tables be off the table?

We’re in an art gallery…… but there are no paintings on the walls. The paintings are just described to us. Also, the person describing them is really quite dull, and is just listing off colours and their locations on the canvas. You don’t get a feel for the mood and you’re not even sure what the painting is of! You’ve wasted your time on a terrible day out!

Most of the time, this is how I feel when I see tables of data. When we produce our publications, the temptation can often be to pour all of our data onto the page in a table. This is the quickest way to get the data to the user, right…..?

Not necessarily!

When you produce a table, consider what the data is saying. Presumably it has a story to tell, or you wouldn’t be publishing it? Think back to how we felt in the art gallery! A good picture can paint a thousand words, so let’s really try to SHOW the reader the messages in the data!

A recent tweet from data journalist Alberto Cairo perfectly illustrates this point. Would you see the Data-saurus by just looking at the summary statistics or observing the data in a table? Almost certainly not!


A more classic example of this challenge is illustrated by the Anscombe’s Quartet.

Visualising your data isn’t always simple, and there aren’t always right or wrong ways to show a single dataset. It’s all about what you feel the main messages should be. In a previous blog, Rob Fry spoke about how you can pick the best charting strategy for the message in your data. This could help your thinking if you’re wondering how to best get started with presenting a table of statistics.

Now let’s apply this thinking to a genuine ONS example. The House Price Statistics for Small Areas (HPSSA) release aims to report house prices by local areas. As data go, these would be reasonably easy to simply publish in a table. The reporting periods are simple and the data can be taken at face value, with no confidence intervals or mitigating methodological quirks.

Below is one such table with 2014 house prices by local authorities in Wales. I suppose the table does a reasonable job. It’s ranked by value and fairly easy to explore, but could it do more?


Presenting the data in a simple horizontal bar chart addresses a few of the table’s shortcomings. It is easier for a user to really see and compare the magnitude of the values, with any outliers such as the particularly low Blaenau Gwent standing out from the crowd at first glance.

I’ve also placed a bar with the Welsh average on the chart. This clearly stands out from the rest, and offers wider context. The reader can make meaningful observations about how areas compare with each other and the overall average without being drowned in figures.

Finally, but quite importantly, I’ve given some thought to the title. Rather than a dry statement of its content, we have a question that primes the reader to look for the story we are trying to illustrate!


We’ve now produced a simple, clean and insightful chart. This is far more effective at giving the user a feel for the numbers than our original table and we could happily publish in our article, bulletin or even on social media.

But like I said, there are no right or wrong answers!

I might have decided the most appropriate story to focus on was the disparity between areas of Wales. How does each area compare to the Wales average, what is the difference? This isn’t unreasonable, so we could take a different approach. Take the chart below for example. We’ve just painted the same subject, but from a different angle. Whereas before, our common “horizon” (the baseline from which we read values) was £0, it is now the Welsh average (£140,000).


Once again, we’ve produced an insightful chart that describes house price disparity in Wales far more effectively than our table ever could!

Lastly, let’s think about another story that the table is pretty hopeless at telling. How do prices vary in the north as opposed to the south? Between rural and urban areas? Tables are terrible at showing geographical patterns in data. Even if the user knew the exact shape and position of every local authority in Wales, it would take a special kind of genius to form a picture of the country’s house prices perfectly in their head!

A good way to show this would be to draw a map. Below I’ve produced a very simple map that broadly describes the spatial distribution of prices in Wales. The user can instantly see how the most expensive areas are in the south and closer to the east (English border). They will also see the cheapest areas very close by in the Welsh Valleys. Isn’t it interesting how all of the action is happening in a relatively small area in the south east of the country? It is not at all easy to see this by looking at our original data table!


Even in our very simple table of 22 house price figures, we could go on and on picking stories to visualise in different ways. If you need to produce more than one chart to represent different relationships in your data, go ahead and do it. The possibilities for educating and enlightening your readers should really encourage you to go that extra step when you are about to publish a table.

Let’s now imagine we’re in a better art gallery. All of the art is clearly displayed around you, and the paintings are beautifully lit and framed for maximum impact!

Why should your data be any different?