Sentiment and Personas – social analytics around #ONSalpha

Our recent purchase of Brandwatch has opened the door to new opportunities for us in the word of social analytics. After getting to grips with the platform, the ONS Alpha project provided the perfect opportunity for some perception and persona analysis.

We’ve seen many a negative tweet about our current website over the last couple of years, but who are these users? What do they really want and does social data reinforce the feedback already obtained in recent user testing?  Let’s see what Brandwatch can tell us.

Having bought data back to August 2011 – when the current website launched – the challenge was to use Brandwatch’s vast array of operators to make sure the right data is available at the click of a button.  No matter how good search queries are, there are always some sites or mentions that will slip through but it was time well spent sifting through! (Given the sample size we’re looking at, I picked through each mention to ensure the positive / neutral / negitive category was correct.)

The initial result was a pattern with a number of significant spikes over the last 3 years.

Activity Aug11-Dec14

Switching this chart to show sentiment and events that triggered spikes:

ONS Website Timeline-page-0

The picture has been a predominantly negative one since the launch of the current site. However the pattern of negativity has fallen steadily since January’s issues forcing the site to be taken down. In June there was a positive highlight in the launch of the Economic Time Series page – well received by a number of data experts.

The above timeline suggests that sentiment has shifted somewhat – from negative to more neutral, but the first week of the Alpha launch paints a very different picture if we zoom in.

Sentiment Bar Chart post alpha

Almost 250 of 400 tweets (63%) have been positive in the first week of launch.  Contrasting this with the first week of the launch of the existing site…

Sentiment Bar Chart Current Website

…where initial reactions were mixed for the first couple of days, but at the end of the first week, only 5 of 96 tweets were of positive sentiment (5%).

It’s interesting to see this pattern, but digging a little deeper, who are the users behind these sentiments?


Over time, the trend in Data Experts/Information Foragers/Inquiring Citizens vocalising their views on Twitter seems to have shifted somewhat.  (See this earlier blog exploring the personas we’re focusing on). At the launch of the current website those classed as Information Foragers were the most vocal in their views – a majority of 62%.


Upon the launch of the Alpha, they were the least vocal group.


Initial findings suggest that each of the persona types are finding positives in the Alpha, a massive improvement from the launch of the current website when all groups again vocalised negative experiences.

Here’s a few examples of feedback from users…

Though it’s perhaps a little difficult to draw any major conclusions at this stage, signs are good that the work is moving in the right direction.

One comment on “Sentiment and Personas – social analytics around #ONSalpha”

Comments are closed.