Using behavioural insights to manage demand for public services

This is about an award-winning project carried out by the Housing team at the London Borough of Ealing, supported by With The Grain (that’s me). We think we achieved a 70% reduction in the number of families being placed in temporary accommodation by the Council, and the main principles are now being scaled into the Council’s ‘business as usual’ approach. We think it has the potential to save millions and improve the wellbeing of hundreds of families. So, if you’re working out how to do Demand Management in practice, or wondering how to make the most of behavioural insights in local public services, you need to read this.

The approach we took was co-designed. Council staff were involved in developing ideas, drafting scripts and specifying content. Get in touch if you want to know more about how we did it.

Helping people take control

We did this by helping families who are likely to become homeless (typically due to eviction or a breakdown of an existing household), to take control of their situation: to look for a new home for themselves if they are able to, rather than being funnelled into Bed & Breakfast (B&B) and Temporary Accommodation (TA) to wait for the Council to find them a home. This matters, because the reality of living in TA and B&B is not good for families’ wellbeing. It is also expensive to provide at a time of unprecedented cuts in local government budgets.

Our approach

Our ‘Reframing First Contact’ pilot consists of a conversation. We call it an advice session, with potential follow up phone calls and sometimes meetings.

We use a number of materials:

  • a script for officers to use
  • a leaflet shown by Customer Services Advisors to callers, to help residents frame the conversation. This helps identify people who are eligible for the pilot
  • a tablet computer with a front page of hyperlinks to the most useful sites/pages when searching for a home
  • an action plan for residents to take away.

I’ll explain the process, so you can see how the materials are used:

  • Resident arrives at Customer Services front door and is given a ticket for housing advice
  • When called, the housing Advisor makes eye contact and shows them the framing leaflet, to immediately establish whether they meet the criteria set out:
  • they say they risk becoming homeless
  • they have dependent children
  • they indicate that they do – or might – need to find a new home
  • If the Advisor judges (usually within a minute) that someone is eligible, the Advisor calls a pilot officer and asks them to help the residents as part of “our new service”
  • The officer collects the resident, and takes them to a room where they sit alongside the resident. When they can, they give the resident the best chair, to help them feel ‘in control’

advice session

Photo posed by officers

  • They then have a conversation based on the agreed script, with a tablet computer available – so they can search for homes and other information
  • We don’t collect any personal data, except contact information. We found collecting personal data tended to steer conversations away from residents’ capabilities, and also enforced an unequal power balance between the expert/gatekeeper officer and inexpert resident.

Behavioural Effects we used

Throughout, our intention has always been to present it as normal for people to look for their own home – one that they can afford – and then to make doing so as easy as we can. To achieve this, we used around twenty identifiable heuristics, including the ones listed below. We also stopped the inadvertent use of effects that were having an adverse impact on behaviour.

  • People are primed to frame the conversation. The What Do You Want To Do? framing leaflet tests that the resident is comfortable saying they are someone who needs to find a home, as distinct from being given one. (The business as usual – BAU – approach is to assume that someone wants to be a ‘homeless applicant’ – and therefore a customer.)

Framing leaflet

  • Scarcity effect – when the Customer Services Advisor calls the pilot officer, she says: “I know the new service is really busy, but it would be great if you could squeeze in Mr A right now”.
  • Talking about looking for a home, we set the default as ‘looking for yourself’
  • We increase salience by referring to a time-limit. “This is about finding the home where you’ll be tucking up the kids at the end of next month”
  • We have an emotional ‘reward’ in mind – settling down and being happy – and we talk about a ‘home’ (whereas the BAU approach is to refer to a ‘property’)
  • We make social proof available – to demonstrate that others like have done this and are happy
  • We reduce cognitive load – avoiding jargon and unnecessary concepts (of which there are many in the BAU approach)
  • We avoid endowment, like “duty” and “entitlement” (which anchor the conversation unhelpfully in the BAU approach)
  • We avoid scarcity effect when it’s unhelpful, like telling people how tough it is to get a council home. (In the BAU approach, this was seen as “managing people’s expectations”; however, Prospect Theory predicts that this encourages risk-taking behaviour).
  • We have a commitment device – an Action Plan – so that residents can note the websites, agents, etc they will contact

Action Plan

  • We increase the salience of, and of plans and information by asking people to write them down themselves
  • We help people visualise their plans – asking them to explain where and when they are going to search – so they’re more likely to do them
  • Reciprocation – “when you find somewhere, we will be able to help you with the deposit”

What did we find out?

We think the approach we took, and the way it worked, showed three main things:

  1. Co-production works for behavioural techniques. Drawing from a wide range of behavioural effects, council officers worked alongside a behavioural practitioner to create a new approach. It’s their project.
  2. Using behavioural insights, we can increase demand for an ‘upstream’ service that supports independence and self-sufficiency, and so reduce demand ‘downstream’ for services that are expensive to provide, may not improve wellbeing and may increase dependency.
  3. Local services can use behavioural insights at an operational level. It doesn’t depend on developing or changing local policy. This work has been commissioned and sponsored by a Director and service heads.

What was the impact?

We didn’t have enough control over the front door of Customer Services to set up a randomised control trial. However, we think the two main measures we do have are pretty conclusive:

  • First, a qualitative measure at the end of the advice session, asking the resident if they plan to look for a home themselves (and whether they will look further afield if they cannot find somewhere local they can afford). Over a four month period from November 2014 to February 2015, the vast majority of residents who took part agreed to look for a home themselves (31 out of 34), including 21 who explicitly agreed to look for somewhere they could afford even if it wasn’t in the area they were living.
  • Second, a hard Demand Management measure. Officers checked whether residents who take part in the advice session went on to become a ‘homeless applicant’ by cross-checking with application records. Just 2 of the 34 became homeless applicants, far fewer than would normally be the case.

How does this compare with Business As Usual? During the same period as the pilot, Ealing Council accepted 234 other families as homeless . These families did not benefit from the behavioural pilot service. This table compares the outcomes.

November 2014 to February 2014 Behavioural Pilot Normal (BAU) method
Number of approaches about potential homelessness 34 1127
Number of homeless applications taken and accepted (leading to B&B/TA) 2 234
Acceptances as %age of approaches 6%* 21%

*NB small base

These results suggest that our behavioural approach has the potential to reduce demand – in the form of provision of B&B & TA to families designated as homeless – by up to 70%.

Why this matters

When families are helped to find their own homes, in areas of their choice, that they can afford, they are able to settle down and begin to re-establish the family life that households in B&B/TA often find difficult to sustain. For more on the impact on families of homelessness and living in B&B and Temporary Accommodation, see Shelter and this report on the impact of temporary accommodation on health.

There is another driver of course. Like most of my work, this is about Demand Management. The London Borough of Ealing has faced severe cuts of £96m over a four-year period. However, demand for homelessness services is rising – due to a vanishingly small supply of social housing, and rising evictions in the private rented sector. So reducing the number of families in temporary accommodation is vital to reducing the cost of this multi-million pound service.

Our behavioural pilot points to significant savings, more practical to measure than an increase in wellbeing. Accommodating a family of 1 adult and 3 children in London for a year typically costs the Council between £18,000 and £27,000 plus officer time. So the potential saving to the Council, if it is able to assist at least 100 families to find their own homes, is £1,800,000. No wonder the Council is working out how to scale up the approach.

Award-winning

We’re proud that our project won the Grand Prix at the inaugural Nudge Awards, ahead of projects from the world of advertising and finance. My thanks to everyone at Ealing who played such a big part, and to Professor Richard Thaler for choosing a local government project as the Grand Prix winner. Those of us who work in and with #localgov know that it’s rarely seen as glamorous, but it makes a positive difference to the lives of millions. And, right now, we need to ramp up the use of behavioural analysis and insights to deal with the reality of major budget cuts.

Deep blue water between policy and behavioural insights

Some stark news today illustrates a point I’ve been making for a while. Sadly, this is life and death stuff.

Last October, it was reported that Britain would “not support any future search and rescue operations to prevent migrants and refugees drowning in the Mediterranean, claiming they simply encourage more people to attempt the dangerous sea crossing”

This seemed like the Rational Actor Model gone mad to me. Even without the benefit of any ethnographic research (without which it’s often best to reserver opinion), was it even remotely plausible that demand for these crossings would be reduced by this decision? I thought not, so I tweeted:

The flow of refugees has of course continued since this decision, with tragic consequences. A lead item on the news, amid much hand-wringing.

My point is not: I told you so. My point is about how immature the use of behavioural insights is in government and policy making. If government were systematically applying insights from behavioural sciences to policy-making, this policy (not rescuing people, in order to influence the behaviour of potential refugees) wouldn’t have got off the drawing-board. Not because it’s immoral, but because – in behavioural terms – it’s illiterate.

Yet we’re told how influential behavioural practitioners are these days (the Independent, for example, claiming a “profound effect on Whitehall” for the BIT). That’s not how it looks to me. I’d say that we’ve influence when commissioned, in a strict supplier-client relationship. This is not real influence.

Some of us in the behavioural trade have a chuckle from time to time at attempts to influence behaviour that don’t cut the mustard. Now, it’s about time we started calling out policy that is clearly flawed.

This is What We’re Like: how the most counter-intuitive thing you’ll hear all week will help you manage demand

(Originally published by the i-Network following my talk at their recent Shaping Demand conference)

You know how you sometimes lie awake at night planning to do something the next day – and then fail to do it? Well, you’re not the only one. We all do it.

Me, too. I’ll resolve not to have butter on my toast in the morning. Or walk to work. Or decide that, tomorrow night, I’ll wind down for sleep by doing the cryptic crossword instead of flicking through TV channels, because I know that screen time before bed makes sleep harder. And what’s more, I really, really, really mean it. Every time. But I almost never follow through on these decisions. Why not? It’s not because I change my mind and decide that I was wrong.

Here’s the counter-intuitive thing. The reason we don’t follow through on these good intentions of ours is that one of the main ways we understand human nature is wrong.

We’ve all grown up with a clear sense of self, believing that we’re in control of our actions. We take account of facts and opinions, make up our mind, then enact the decision we’ve made. And because I think I’m in control, I think everyone else is, too. So, when I want to influence someone’s behaviour, my instinct is to persuade them. If they’re doing something different, I assume that they’re doing it for a reason. So I try to change their mind.

Though we grow up with this understanding, there is a mountain of evidence that this isn’t what we humans are really like. Our evolution to be arguably* the most successful species on the planet has many quirks we’re beginning to understand better, but which are not yet in the mainstream and which overwhelmingly haven’t been reflected in how we have designed public services and shaped localities.

So what are we really like? Well, we have loads of mental shortcuts. Why? Because it gave our ancestors an evolutionary advantage to leave much of their brain’s cognitive capacity available to spot and instantly deal with predators. This is why you find it harder to apply your willpower if you’re already doing an ‘effortful’ mental task. It’s also why you’re really good at pattern recognition; your ancestors who realised that the leaves rustling nearby might be due to a predator were the ones who got to influence the human gene pool by running for cover and living long enough to pass on their genes.

Life expectancy has typically been under 20-30 years for most of human existence, so it’s no surprise that we also have mental shortcuts that help us grab the resources that will help us through the night, or the winter, and hold on to what we’ve got.

We also have plenty of short-cuts that help us be social, since human co-operation has also give us a real advantage. These biases have a much bigger influence on us all than we’d like to admit; in short, no-one likes to be the odd one out.

As I said a few times in my session at iNetwork’s recent Shaping Demand conference: this is what we’re like. And I think these insights are really helpful in taking new approaches to managing demand.

Freed from believing that people always have a reason for their actions, we can try to systematically examine what people do that creates demand for public services, and understand how it happens (not why).Freed from the need to change people’s minds when we want to influence their behaviour, we can concentrate instead on making self-sufficient behaviours easier and seem more normal.

And we can make this evidence-based understanding of human nature accessible, in order to involve staff, citizens and others in co-producing new designs and approaches.

The question for each of us is: how are we going to make sure this happens, rather than going the same way as my late-night commitment to walk to work tomorrow?

[*ants might beg to differ]

When @futuregov asked the questions

I did a talk for local government heroes Future Gov recently, on the back of which they interviewed me. This is the result:

http://wearefuturegov.com/2014/04/behavioural-effects-demand-management-and-local-government-an-interview-with-warren-hatter/

Doing demand management better, with behavioural insights

There may still be local authority chief execs and directors in local government who think they can steer a course through the financial challenges to come (the handy headline figure is a forecast £14.4bn shortfall in services by 2020) by reducing staff numbers and service levels, raising efficiency and changing eligibility criteria, but I don’t know them and I’m pretty sure they don’t exist. And, encouragingly, neither do the authors of Managing Demand: Building Future Public Services, Anna Randle and Henry Kippin. As they point out, these measures realise immediate savings but don’t shape services for the long-term.

This report is essential reading if you’re interested in the future of local public services. To me, it marks out the territory that should be occupied by anyone working on developing local public policy and practice. It should be, of necessity, the new black.

DMCover

Anna and Henry cover a lot of territory – perhaps too much. There are dozens of aspects worth writing about, but I’m focused here on the topic closest to my heart, as its my job: using behavioural insights.

Managing Demand and the emerging practice it reveals, is particularly interesting to me because of the coincidence of this stark challenge to public finances and the rapid growth of evidence from behavioural sciences. As I’ve explained to many people, I’d be promoting the use of behavioural insights to policy makers and practitioners right now, even if public finances were in rude health. The fact that they are not, though, provides even more of an opportunity for us to design services that reflect what people are really like: demand management done well means reframing public services and places – and we can knit into this redesign our rapidly improving understanding of just how context-dependent human behaviour is, and of how much less control we have over our actions than we like to think. If local government budgets were ticking along nicely, and the demographics were in our favour, the rationale for change simply wouldn’t exist.

If anything, Managing Demand understates the extent to which public services’ approaches are behaviourally naïve. Here are some examples I’ve come across recently:

  • Experienced Housing officers will often explain to cititzens likely to become homeless what the limitations are to what the Council can do for them, in effect anchoring the conversation at the level of service that used to be possible when supply and demand were less unevenly matched. When resources are scarce, we want them more, for good evolutionary reasons, so officers may be inadvertently encouraging demand that is costly to manage and damaging for residents’ wellbeing.
  • Waste and recycling communications are often framed around the needs of the service, describing the benefits in terms that give the citizen no emotional reward, and using terms (like ‘food waste’) that are at best ambiguous to most people.
  • Letters to a tenant in the process of reducing their arrears often don’t provide feedback in a way that acknowledges their success; they may receive an identically-worded letter to that sent to tenants with increasing arrears. So we miss the chance to embed the key behaviour: paying off the arrears.
  • Councils involved in projects to give away home efficiency measures often have limited success. One reason is endowment effect: our inclination to see something free as valueless. But it is also because they tend to introduce cognitive barriers into the conversation immediately, referring to ‘retrofitting’, not the emotional reward to the resident: a cosy, draught-free home.

There are plenty of demand management approaches to issues like these which don’t utilise the insights from behavioural economics and psychology that are needed to reframe what’s on offer in a way that works with the grain of human nature. So what are the emerging approaches that do explore these, and which look like becoming part of the toolkit of the behaviour-smart local authority?

One is behavioural auditing. This is a way of examining the way a service works to identify where demand is created, track this back to the need that the demand expresses, and use this as a starting point. For many services, apparently identical need will lead to differing levels of demand from different residents; analysing the behaviours of those who create little or no demand helps re-design in a behaviourally-smart way, as it give us insight into which behaviours to encourage.

Another important aspect (not new, of course) is ethnographic research and observation. When we focus in on the behaviours that create service demand, understanding what it is that people do and how they do it is much more important to re-design than the reasons people give to explain them – and the rationale that officers apply. There is plenty of evidence that we adjust our views to accord with what we do, much as we would like to believe the opposite to be true!

A third is co-production. Guided by a behavioural practitioner, officers, managers, service users, citizens, community representatives, etc (I’ve worked with teachers and nurses, for example) are able to understand and work with real insights from behavioural sciences and generate many of the approaches that they go on to put into practice.

A word of caution, though. We are realising how difficult a behaviourally-smart approach can be to put into practice – for good reason.

Why shouldn’t we be surprised at this? Individually and collectively, our default model for influencing people is to try to change their mind and persuade them. It’s based on the rational actor model. My experience on a wide range of demand management projects with local government and public health is that policy-makers and managers can accept the overwhelming evidence that behaviour is context-dependent and so target behaviours need to be made easier and appear more normal … but implementing this, when it challenges the mental models we have worked with since we were toddlers, is not straightforward.

My view? Let’s keep the focus on demand management; let’s make our approaches as behaviourally-smart as possible; and let’s build the capacity in local government and communities to use behavioural insights.

Doing it better

Below is a post I wrote for the Behavioural Design Lab a short time ago. The aspect that has attracted the most interest is the use of proposed use of defaults to move from a project management-centric view (where residents make a fresh decision at each stage of the process that has been mapped out) to a citizen-centric approach (where, once someone has chosen to have a warm home, the default is that they are taking part.

Two quick points, drawing on conversations that I’ve had with people since writing this.

First, this is a really good litmus test for whether we are comfortable using behavioural insights. Personally, my view is that our choices always have a context, and that framing the choice in this way makes it clear and easier to make a decision that is salient to the person making it. But I know that some people not used to working with behavioural insights are a little uncomfortable with it; because it seems a little, well, sneaky. In a world where there are numerous websites on which I can click a ‘buy now’ button, before verifying a host of details, I’m comfortable with it, and that’s why I see it as a litmus test: it only seems sneaky if you compare it with a context in which people have to repeatedly decide.

Second, reflecting that most of the feedback I’ve had has been of the “what a great idea!” variety, there is so much that we can do better in public services, and it is within our grasp. I hope that this helps demonstrate the value of what I’ve been trying to do with With The Grain over the past couple of years. There are so many behavioural effects that we know can affect our decisions; and there is an ever-growing body of evidence of how and when they work. So let’s generate new approaches drawing on this knowledge. This idea was generated in a workshop of stakeholders brainstorming ideas based on different effects, of which defaults was one. Most of them hadn’t been exposed to much behavioural science before; if they can do it, so can you. This is one of our best chances of meeting the demand management challenge.

“Did I just use behavioural science? But I’m not a designer!”

So said a dozen or so stakeholders of a project aimed at retrofitting 160 draughty homes in Crawley, West Sussex. With good reason. They had co-produced a wide-ranging set of design and communication ideas for the project.

The project offers work such as external wall cladding, funded by the Energy Companies Obligation, through which the Government is obliging utilities to fund energy efficiency work on Britain’s coldest, draughtiest and most energy inefficient homes. The problem many have found is that, in the absence of existing demand (that is, people who are aware that they want their home retrofitted but haven’t been able to do it yet), building demand for something free is tricky. Price perception tells us that if something is free, it doesn’t have value. Homo economimus might see free cladding as a no-brainer; real people don’t.

So, when introducing the opportunity to people, we decided to frame the choice as being between a cold home and a warm home – not as the chance to choose a named process or product. And we avoided terms (such as ‘retrofitting’), known by professionals but which may provide a barrier if not familiar to residents. Crawley-ECO-leaflet-section

Adding the use of behavioural insights to the team’s existing expertise in community engagement had a major impact immediately, speeding up recruitment 4-fold, compared with similar projects being undertaken elsewhere in the South East.

Another innovation is that, when people say ‘yes’ to a warmer home, this becomes the default setting. So, instead of initially agreeing to an “assessment” which leads to another choice once the surveyor has visited, householders make a single choice: the surveyor will make a recommendation of the measures to adopt “ … unless you drop out”. From a project management viewpoint, we’re moving from a process whose success depends on people saying ‘yes’ at several different stages to a process designed to support and prove people’s positive choice to have a warmer home.

There are a dozen other ways in which we are using – or plan to use – behavioural insights. Rob Bennett, who leads the community engagement team, says, “It is really important that we find ways to encourage communities as a whole to get behind these initiatives, So whether it’s the initial decision to participate in a scheme, or ensuring that residents communicate what works best by sharing good practice and experiences – we expect behavioural evidence to play a critical part in successful ”delivery”.

We think we’ve learned what the With The Grain tool has also demonstrated in other settings: that behavioural insights are accessible and usable; that these insights help make approaches more people-centric and therefore more efficient; and that it’s possible to get away from the default setting of trying to persuade people.

So we now have a platform for using behavioural insights in the future. And we have a group of stakeholders who are comfortable with knowingly using behavioural insights to affect the context within which people make decisions.

In the future, this won’t be unusual. Right now, it feels revolutionary.