Working In Uncertainty

Systems studies that improved housing repairs


A particular form of wishful thinking that I came across years ago when working as an auditor is our tendency to think that exceptions to our usual or intended pattern of work are less common than in fact they are.

For example, when I was a young auditor I did a special type of review at a solicitors' in London, checking for problems handling clients' money. The rules for this are very strict and I found a small breach of the rules in one of my samples. It was clearly an accident, but I asked the book-keeper about it and she was surprised it had happened at all. I asked if it could have happened at other times and she was doubtful. An hour later I had checked the entire year and found around 50 breaches of the same type. She was horrified.

Over years of auditing I learned that this sort of thing is typical.

Watching John Seddon of Vanguard Consulting talking about changes to housing repairs services in Portsmouth (pictured) and Tees Valley reminded me strongly of this phenomenon. Although John used the example of housing repairs to illustrate a number of the ideas in his approach to systems thinking, I will focus on the value of finding out how things are really working. The key to the huge improvements in customer satisfaction and efficiency brought about in these cases seemed to me to be understanding what was really happening instead of assuming that what was happening was what was supposed to happen.

The housing repairs process before changes were made

The way work was done before improvements were made was fairly similar in Portsmouth and the Tees Valley Housing Group. The tenant (e.g. in a council owned flat) notices a problem that needs fixing, like a broken window or dripping tap. The tenant makes a telephone call to a call centre where someone who is not a skilled repairer takes the call, identifies the nature of the work needed, and passes instructions to one or more skilled tradesmen to do the work. The call centre person also makes the appointments with the tenant using appointment windows (e.g. AM or PM) rather than an exact appointment time. If more than one skill is needed then more than one tradesman is called for and each tradesman's task is called a 'job'.

The tradesmen arrive in the time window of their appointment, do the repair, and report that the job has been done.

The effectiveness and efficiency of this whole operation is measured and monitored monthly. Everyone is encouraged to ensure that each job is done quickly and the cost per visit is minimised. The tradesmen often work for another organization, so a lot of attention goes on making sure that organization is giving a good service.

What could go wrong?

In both cases the managers in charge were reassured by the performance statistics showing high customer satisfaction and good performance in getting jobs done according to the time targets for each type of job (emergency, urgent, normal). At the same time they were troubled by complaints about their services, coming direct from tenants or indirectly through local councillors. How could their repair service be less than satisfactory when the performance measures were saying everything was fine?

When they left their offices and went to see work being done, and when they began to collect numbers about the reality of work, they soon realised what was was really going on.

Those statistics showing jobs completed within the required number of days were not showing the end-to-end time from the tenant's first call to the eventual, satisfactory completion of the repair, which might involve a number of jobs. Furthermore, if a job was blocked because a customer was not at home when the tradesman called, or the tradesman couldn't get required parts on the day of the appointment, or could not get to the appointment on time, then the job was rebooked and the clock restarted. An emergency repair was dealt with by making the property safe then booking another job to effect the repair.

While the performance statistics seemed to be saying repairs were completed within the allowed 3 or 10 days this was just jobs with no problems. In fact the end-to-end repair time was more like a month or more. Between 40% and 60% of calls to the call centre were not new repair requests but in fact complaints and queries about when things would happen.

Some repairs were ineffective. A dripping tap had its rubber washer replaced 13 times within a year. It would have been cheaper to replace the tap but the incentives, reporting, and standard expectations prevented it.

On too many occasions tenants were not at home when the tradesman called. This was partly driven by the fact that they had been offered a wide time window and a limited choice of appointments.

Because jobs were not well understood from the first call it was common for the tradesmen to make a special visit to diagnose the problem and identify any materials they would need. Assuming the tenant was at home, which as I have just mentioned was not always the case, the next problem was likely to be lack of materials at the local supplier. If the materials were available then finally it was time to arrive at the tenant's home and do the job - if they were at home, and only if there was enough time left to do the job before the next appointment window.

The customer satisfaction surveys could be deceptive because they were questionnaires given to tenants who perhaps had not used the repairs service for months or even at all.

Starting improvements

As John Seddon puts it, 'the only plan is to get knowledge', and that knowledge had a huge impact on the managers involved. With reality finally shining through it was possible to start re-designing the processes.

This meant expecting the call centre to do no more than log the tenant's problem and pass the details on to the tradesmen, who were then responsible for calling the tenant to better understand the problem, for making an appointment (at a specific time, not in a window), and for managing the repair through to completion. That included letting the tenant know how things were getting on and trying to make the right repair, correctly, first time.

The maintenance teams began to hold local stocks of the parts they often needed, having studied the actual demand for various types of repair and found it to be surprisingly predictable.

Once the repair was done the call centre would call tenants to ask about their level of satisfaction with the repair and overall service.

The impact

The impact of these changes for Portsmouth and Tees Valley was similar. Customer satisfaction rose dramatically, with many more tenants awarding 10 out of 10. Calls to complain or chase reduced from 40 - 60% of total calls received to less than 25%, with scope for further improvement. Employees were happier too, even the contractors.

Instead of taking a month or so on average to complete a repair they quickly moved to taking about 5 days on average, with over 80% of repairs completed within 10 days. That's repairs remember, not just jobs.

Money would have been saved and perhaps was, but tenants began to call more often for repairs, rather than ignoring problems to avoid having to deal with the repair organization.

All in all, these have been important transformations from which everyone involved has gained.

Working In Uncertainty observations

Although finding out what was really happening, instead of relying on received wisdom, was not enough in itself to improve performance, it was the start. Interviews with the people involved make it clear that this early stage was the most surprising and emotional part of the whole process. It was for some a revelation – the discovery that they had deluded themselves for years – and involved a change in management habits that was difficult for some.

However, they did it and benefited. Our tendency to think we know more than we really do is natural but not insurmountable. Far from it. An interesting detail of this case study is that, as John Seddon says, demand is more predictable than most people think. So, although we think our predictions are better than they really are, if we make the effort to study the numbers effectively we can often make predictions that are better even than we thought we were making!

(Sources of information: A presentation by John Seddon in June 2011 at Tom Gilb's conference in London, and the following documents.

A systematic approach to service improvement, September 2005 from the Office of the Deputy Prime Minister, published May 2006.

A systematic approach to service improvement - an update, October 2006 published by the Northern Housing Consortium.

Systems Thinking repairs contractor heralds Revolution in service provision by Howard Clark, Vanguard Consulting.

Portsmouth's Owen Buckwell: A Very Local Hero by Howard Clark, Vanguard Consulting.)

Made in England


Company: The Ridgeway Expertise Company Ltd, registered in England, no. 04931400.

Registered office: 29 Ridgeway, KT19 8LD, United Kingdom.

Words © 2011 Matthew Leitch