How the Guinness Partnership identified damp and mould risks and prioritised red-flag complaints

Transcript

I'm Chris Haynes. I'm head of customer insight for the Guinness Housing Partnership. I've been working there for about six or seven years. Prior to that, I was in a similar role at Eurostar, the train company. Before that, NatWest in the financial sector.

What problem were you looking to solve by bringing in Wordnerds?

We had large amounts of unquantified comments from customers, both in our surveys and also coming in through things like complaints through our contact centre. And I guess the fact they were unquantified was a problem for us. So without any way of grouping them, scoring them, the only real option is to read them through, which is obviously a useful thing to do each month, but it doesn't really give you objectivity.

You'll inevitably remember comments that correspond with whatever you were interested in that day. And I think that just chucking a few comments at the end of a report you're doing, calling that the voice of the customer is a bit superficial. And similarly, counting the words and creating a word cloud that shows people mention "Guinness", "repair", and "flat" a lot doesn't really tell us anything.

So it's really a way of trying to process all of those comments that we had in an objective way that maybe we could trend it, dig into things in more detail and get to topics that we might not have considered before.

What is the main benefit you’ve seen from using Wordnerds?

I guess I can give you some example of the practical things we've done with it. One of those is pushing the description of outstanding complaints through the tool, picking out sentiments and any red flag topics, which could be things such as vulnerability, health and safety issues. And my purpose of pushing those case descriptions through, was to allow us to have a sort of prioritised list for the appropriate team to work through. So they had all these unresolved cases and they needed to know which ones to get through, get to first, rather than having to make a sort of manual call on that, we could use Wordnerds as a sort of objective way to prioritise that list.

Another example would be some work I did last week for our assets optimization team, which looked at the frequency of mentioning different components of the building. So that could be like brickwork, windows, doors, etc. And the associated sentiment and the volume with which customers mentioned those things. And that was allowed to allow that team, which is very sort of asset building based, to expose them to some customer data to complement the sort of asset based data they already had.

I think another one as well is to (I guess this is the most interesting to me really) is to identify sort of micro issues that we'd never thought of before. So as an example of that, we had a couple of comments I noticed from customers when they felt that ivy and creeping plants were blocking ventilation on the exterior of their property. And it was their belief that that was causing damp problems.

So I was able to use Wordnerds to identify every comment which mentioned those sort of plants. So as you can imagine, it was to create quite a sort of botanical topic to capture all the possible things that could go into that. So we created that topic, whether damp was mentioned or not, and then cross-reference that against our damp and mould data. And we were able to demonstrate that there was a higher incidence of damp and mould if that topic had been triggered by the customer in one of their comments or some contact they'd had with us.

So that was really useful and it allowed me to turn a sort of loose hypothesis into something which was a bit more data-led in terms of proving it or disproving it.

How does Wordnerds fit into your strategic vision, past and future?

I suppose my strategy working insights, I'm dealing with survey data and other data all the time. I'm trying to sort of move the organisation from backward facing reporting where we produce a monthly report and say what happened in the last four weeks, which is fine because we need to do that. But what I'm interested in as well is sort of more predictive approaches to data.

So we've built models looking at the likelihood of tenancy failure, the likelihood of unreported damp and mould, those sort of things. For Wordnerds, that can be part of that sort of strategic shift, I think.

So like all providers, we're doing the regulator's TSM surveys, which is a questionnaire which has been designed by the regulator, which everyone must use. And as part of that we obviously want to understand what drives satisfaction and we've built a driver model for overall satisfaction but at the moment it essentially just tells us which other metrics in the survey have most connection with satisfaction so if it's not something that the survey asks about then it won't be included in that model.

So at the moment that our driver model may not be the whole story. And there's a whole ton of other stuff which it wouldn't be practical to ask customers about in the code of question, which appears in verbatims. So ultimately, what I want to use Wordnerds for is to take each of those themes of which there are tens, if not hundreds, and treat them as a sort of binary on or off metric and see the connection between those themes and whether the customer is satisfied or not.

So basically to understand if there are drivers which are very specific and outside the remit of a coverage of a TSM questionnaire, I'd actually have a big connection with satisfaction because we need to obviously work and improve those things as well.

Anything else to add?

Only that I'm impressed with the website that we use where we upload the data and where we can surface the results. And since I subscribed, which I think is just over a year, you've added lots of new features. So the ability to train your own themes, which we use, and then test the score for them to see how good a fit they are. We use that.

It came out last week, but we hope to use the drill down facility as well. So that will give us, I think, hopefully a more natural way of using it because we tend to upload the data with the aim of answering a specific question. And then the drill down will allow us to follow our train of thought when we go through the answers to that question, rather than having to worry about, all right, what type of chart do I need to construct to best show the data?

I think it will be a more natural sort of flow as we follow up our investigations through the website.

‍