18 London End,
Beaconsfield, HP9 2JH
If you are interested in finding out how we can support your Customer Experience; drive Brand improvement; want to find out more about what we do or are interested in working for us; please fill out the form below, hit send and we will do the rest:
aura has launched a new tier 1 Text Analytics Module as part of their successful Supersite Platform. This new software is designed to identify key areas of customer concern from customer text feedback and provide analysis down to retail level. Available as a separate or integrated module within Supersite, when linked to online questionnaires it provides immediate feedback to companies and their retail networks empowering their staff to resolve customer issues and develop improvement plans.
The need for Text Analytics is growing with more customer feedback across more media channels than ever before.
This report from aura’s Director of Insight:
“We are currently receiving a lot of interest in applying Text Mining/Text Analytics to Customer Experience research. Although Text Analytics simply refers to categorising or extracting specific information in order to add value to a research project, it is a very broad and detailed subject and therefore isn’t always fully understood. Ultimately, this results in some poor applications/deployments in research.
The continued trend of reducing questionnaire length as a result of response rate attrition and the acceleration of mobile surveys is driving the renewed interest, allied to the growing use of predictive analytics and data visualisation to bring the information to life and provide meaning.
It is very easy for text analysis providers to market to their customers that their solutions remove the need for closed questions (e.g. asked on a 5/10-point scale) because they can fill the gap. However, that is based on the big assumption that unprompted customer responses will illicit the same feedback as prompted questions. Our belief is that it is not the case and a very short questionnaire has its drawbacks. Trying to understand why a particular customer group is less satisfied or less willing to recommend becomes harder when there’s just to open comments to pin-point the issue. This is particularly the problem when assessing smaller subsets in the data e.g. by location, store, branch, dealer.
Whilst Text Analytics is not a new approach in customer research, it is not fully matured. Some solutions are still “black box” in respect that they can’t be calibrated to meet the idiosyncrasies of a specific service or even sector.
The success of any Text Analytics programme/project will ultimately depend on how well the system understands the text. The accuracy of any categorisation or concept/entity extraction needs to be reviewed initially and on an ongoing basis so that it can be refined. Accuracy shouldn’t just be measured based on the matches that are made (e.g. positive matches vs false positive matches) but also those that weren’t made.
Topic categorisation tends to be either right or wrong but categorising sentiment is more difficult to assess. Providers have different approaches to sentiment and it needs to be carefully deployed to obtain good results. Purchasers of Text Analytics solutions need to understand that qualitative Text Analysis will never be 100% accurate. Expectations need to be managed for Text Analysis because if humans can’t agree on the categorisation of a specific record what chance does an automated approach have?! We’re often asked about the impact of sarcasm in customer comments but it usually isn’t a major problem. It is very difficult to determine if a comment is sarcastic just looking at it in isolation but by evaluating it in conjunction with other responses (e.g. overall experience/recommend rating) we can go some way towards validating the comment sentiment.
The next couple of years are going to be very interesting for the field of Text Analytics. The current surge in popularity will lead to even greater advancements in capabilities along with the trend for integrating internal and external textual data sources, including: CRM notes; Call Centre Voice Recordings (captured as text), email correspondence and complaints records. How all of this data is processed, analysed and reported in order to provide a single coherent view which helps to drive business improvement will determine if Text Analytics really is the game changer that evangelists have been promoting for a very long time.”
Director of Insight at aura