Transcript of a podcast, John Seddon, October 2020

In July this year McKinsey published a report entitled Digital Public Services: How to achieve fast transformation at scale.

These are the arguments in the report:

  • The private sector has raised the bar on customer experience,
  • Better public services will increase trust in governments,
  • Digital services will improve the customer experience, decrease costs and boost productivity,
  • The means to going digital are Agile and smart programme management,
  • And people see no reason why public services should be more complicated than shopping online.

I imagine the report is aimed at politicians, for they hold the purse strings for what will be a multi-million, even multi-billion, investment. The arguments are plausible, but what do politicians know about service-organisation design?

I am eternally grateful to the late Nick Georgiades, the man who taught me during my MSc in occupational psychology back in the Seventies. He installed in me what he called a ‘crap detector’, a preparedness to question assumptions and seek evidence that either confirms or denies propositions. When I read the McKinsey report my crap detector went off the scale. So in this podcast I’m going to discuss the McKinsey arguments and describe what it takes to design truly effective public services.

Let’s start with the private sector has raised the bar on customer experience.

Oh no they haven’t! While it is true that the private sector has invested heavily in digital services, the impact on customer experience has been pretty dreadful. In our recent book Beyond Command and Control, we devoted four chapters to explaining how digital services have worsened the customer experience and raised costs. In an earlier podcast, I described how insurance claims organisations adopting McKinsey’s ideas about ‘best practice’ in digital services digitised non-value activity – the activities you’d design out if you knew what you were doing. If you think adopting new roles and rituals, dreaming things up – in other words doing Agile – and forcing customers down new digital channels through smart programme management is the route to effective services you need your head examined. Of course the big selling point with digital services is lower transaction costs but what actually happens is the number of transactions increases, provided, of course, customers stick around to get what they want. Many don’t. The reason for this failure is, simply put, that computers work on rules and rules prevent the system from absorbing the variety of customer demand. So while digital protagonists may crow about the increase in digital transactions, costs rise and the customer experience worsens.

As an aside, in this report McKinsey repeats the mantra of standardising services. Wrong, wrong, wrong. If you standardise services you inhibit the system’s ability to absorb variety so costs can only rise.

But let’s move on to the evidence that McKinsey offers on how digital services will improve the customer experience, decrease costs and boost productivity.

The McKinsey report cites examples of governments that, to the reader, would appear to already be gaining the benefits of going down the digital track. According to the report Australia has created a new agency, Services Australia, to “improve the digital user experience for many federal government services”. So I got in touch with a colleague in Australia.

He replied, saying Services Australia is a rebrand of the old Department of Human Services. Amongst their remit are services delivered via Centrelink, essentially benefit payments. This service introduced a digital means – using an algorithm – to claw back overpayments of benefits – they called it Robodebt – with disastrous consequences. In short, he described Robodebt as a great example of a digital cluster f***. Apologies for the language, they speak plainly down under.

Citizens identified as over claiming benefits were bombarded with computer-generated demands. The government hired debt collectors, made appeals difficult, imposed travel restrictions for citizens alleged to be in debt and changed the statute of limitations on debt collection. Bombarded with calls from desperate citizens who couldn’t get through, Centrelink outsourced this failure demand to the private sector. These are, by definition, vulnerable people; some were so distressed they took their own lives. Despite many complaints of false debts, government refused to change course. The chief executive of the digital agency resigned in disgust and opposition politicians called for a royal commission to investigate this shocking debacle.

So, no evidence of improving the customer experience, decreasing costs and boosting productivity there then. According to one commentator the public money spent on Robodebt almost matches the amount government originally claimed it would claw back. Given the government has now committed to reimburse half a million claimants we can only assume this early foray into digital services was an extremely expensive mistake.

Despite this disaster the Australian government is committed to plough on with more digitisation in Services Australia, with McKinsey’s help.

All of the players would have been better advised to learn from the debacle. What we know from working with benefits services in various countries is that overpayments are frequently due to errors in the benefits service rather than citizens being fraudulent. If they’d studied what was going on they’d have learned how rules – set in an algorithm – fail to absorb the variety of demand. And the only way to get the service working effectively would be to start the process with a human interaction.

Let’s move on to Denmark. According to the McKinsey report Denmark has national portals for resident and business services. It is mandatory to use a service called Digital Post to communicate with government. So I called a colleague in Denmark. He confirmed that it is mandatory to use digital means to communicate with government. Danes have to sign in to see if they have any mail but, naturally, a lot of people don’t bother or forget to sign in and so miss communications from government (himself included – and he’s  young, computer-literate person). As he puts it:

“The problem is that you get a text notification saying ‘you’ve got mail in the portal’ but then you have to sign in, and you can’t always do that, and so you forget”.  

If people fail to respond (to something they often haven’t seen) a reminder is sent and a financial penalty is applied. Apparently quite a few young Danes who don’t see their call-up letter instructing them to start military service get picked up by the police. He tells me many older people and vulnerable families have to get help from carers or friends to use the digital system. Also some local municipalities have created teams to help people use the digital portal.

No doubt the digital protagonists argue this will work out eventually, with a bit of nudging and coercion and one day every Danish citizen will conform and comply. Even if that is the case (and I doubt it) the next hurdle is: does this digital means actually work as part of a service?

It turns out my Danish colleague was able to answer that question for one service – planning applications. And the answer was a resounding no. If Danes want planning permission the digital portal requires them to fill out a standard form. The consequence is a bucket-load of failure demand. He said that in summary the digital platform did NOT facilitate a swift and easy to use process, did NOT secure relevant information to the application and did NOT create  better resource utilisation – all of which were claims made at the outset. Once again people who don’t give up because they want a service have to get help, so they have to pay agents who are used to navigating the system.

He went on to describe the results of studying a series of planning applications. Inevitably applications had missing information, even missing attachments (lost in the digital ether). In short the digital solution resulted in too much, too little or wrong information. Working with a local municipality they set up a conversation with applicants, designed to ensure the municipality got the right, relevant information. This more effective way of working puts the digital platform in a different place. In essence it becomes no more than a first point of contact that is unable to do any value work. Danish municipalities are obliged by central government to pay through the nose for this. It reminds me of an example in our early work in housing. A housing organisation in Leeds was obliged (by central government) to use a central call centre, run by the city council, for reporting repairs. When the leaders studied what was going on they learned that, similarly, no value work could be done – all the call centre could do was pass a record of the call on – so they were paying a quarter of a million pounds a year for nothing.

We’ve seen much the same with planning services in the UK. One of the first services to be digitised, central government mandated the installation of an IT system called ‘One-app’. To command and control thinkers it was a wet dream. Adherence to activity targets could be monitored, the responsibility for making a correct application was pushed on to residents, planning staff could game the system and, quite naturally, ordinary folk who wanted to build an extension or whatever were driven to use agents, as in Denmark, at a cost.

I should say that there are planning services that have followed our method, putting the IT solution to one side, and they start as our colleague did in Denmark: having a conversation with the applicant so that from there the service can just focus on doing what matters. These services are loved by residents and developers – in fact developers, who interact with many planning departments, couldn’t believe how good the service could be. One app has the same problems as I just described in Denmark. I remember speaking to John Swinney, when he was finance minister in Scotland. He told me Scotland was thinking of introducing a software platform for planning services. I asked if it was the same as one app. He confirmed it was, I suggested he should not go down that road as we had seen what a mistake it was in England. But he did, even though a Scottish council on the western isles had applied our method with extraordinary results. It illustrates the power of a plausible narrative. Evidence doesn’t trump narrative.

Another country cited in the McKinsey report is Germany. In this case, the report includes a series of numbers as evidencing the potential for improving the customer experience, decreasing costs and boosting productivity. So I asked my head of research to look into it. I said my bet was these so-called results were speculative rather than delivered. This is what he had to say:

“Yes, you were right: it’s all extrapolated from projected, not actual, figures; lots of ‘could save’ or ‘would save’ statements; nothing that’s been delivered. The source that McKinsey have extracted these figures from is referenced as being published by the German National Regulatory Control Council in October 2017.”

He went on to say:

“Having looked at the German language version, guess who the report was actually originally written by? Yep, although commissioned by the German National Regulatory Control Council, the report itself was created by McKinsey, who have their logo on the front cover. I think that’s quite cheeky, don’t you?”.

No, I don’t think it’s cheeky, I think it’s outrageous.

It’s time to turn to the problem and the solution. I’m going to do this by reference to a book entitled A New City Operating System. The authors have, without doubt, drunk the digital Kool-Aid. Like McKinsey, they base their argument on the need to restore trust in governments. They also assert that this lack of trust has only been amplified by the new social media – people use social media to make a lot of noise about governments’ failures. And they assert that the technological changes that contribute to heightened cynicism can also power the very transformation of government that will turn that cynicism into trust. Really?

This so-called new operating system is tech-heavy. Mobile and cloud computing, GPS, Big Data, data mining, data analysis, digital platforms, artificial intelligence, internet of things, omni-channel services and so on. It makes me wonder how much money will be spent building this operating system before you learn whether it works.

The book poses examples: A housebuilder who uses the same plans for the houses he builds; who has a family that needs their home built within 6 months. But he faces long waits for planning permissions. A mother of two with a part-time job who qualifies for benefits but has to take time off to apply and has to provide the same information to different departments as the benefits services are specialised. A woman who is opening a restaurant with funding from friends who finds herself paying rent on the premises while waiting for approvals and inspections in order to open for business. And so on.

The authors, to put it bluntly, speculate that information technology will speed up solutions to these problems by getting the right information to the right people at the right time, will also ensure that routine work will be done by machines (for example chatbots answering routine questions), will speed up analysis, will eliminate repetitive form-filling and so on.


In essence the problem we have is that services are specialised, functionalised and fragmented. Their argument – which I should remind you is speculative – is that IT will bring it together. It reminds me of the conversation I had with Jeremy Hunt when he was our minister for health. I said the problem we have with our health service is that it is fragmented and provides no continuity for patients, so people find it hard to get what they need; and what we should do is redesign it to provide continuity and ensure that services provided match people’s needs exactly. His response was that he agreed with my analysis but disagreed with the solution. In his view integration would be better achieved by an IT system. Well, we’re still waiting.

The better solution – and a solution that costs beans compared to massive IT investments – is to firstly gain knowledge of demand in citizen terms. This knowledge is then used to create the right expertise at the first point of transaction, whose job it is to firstly establish with absolute clarity what we call the citizen’s nominal value – the thing that matters to them. From there you focus on doing only the value work – delivering the things that actually matter and when you’re building such a design you build the information technology last because it is at that stage that you know exactly what you need. That’s how you design services to absorb the variety of customer demand. The consequences are reductions in end-to-end time, dramatic reductions in costs, astonishing improvements in service and much less expenditure on information technology. And these improvements are real, delivered, not speculative.

And so to McKinsey’s argument that people see no reason why public services should be more complicated than shopping online

We are not stupid. We don’t imagine that getting planning permission or claiming on insurance should be a simple one-click process, like buying something from a retailer. Of course we do expect things like renewing a driving licence or paying a toll charge should be so, but these are simple, predictable and repeatable demands and IT is good at those. But as demands increase in variety – think benefits – IT, because it is based on rules, fails to absorb variety. Just think about the multi-millions wasted on Universal Credit in the UK, a service mandated to be digital by default, which still doesn’t work, has escalated in cost and will never work via digital means.

What actually matters to citizens is that public services work, not that they can be accessed on their smart phone.

And so we turn to the problem of trust. Why McKinsey and the authors of A New City Operating System think improvements in public services will increase trust in government is beyond me. We certainly do have a problem with trust on governments. Rutger Bregman, author of Human Kind puts it this way:

“Democracies around the world are afflicted by at least seven plagues. Parties eroding. Citizens who no longer trust one another. Minorities being excluded. Voters losing interest. Politicians who turn out to be corrupt. The rich getting out of paying taxes. And the growing realisation that our modern democracy is steeped in inequality.”

Note: no plague called distrust of public services.

We citizens are perfectly able to distinguish, for example, between our views on the National Health Service and our politicians’ management of the Covid 19 crisis. Similarly, the fact that many politicians have a pecuniary interest in privatising health services may lead us to distrust them but not distrust the service. Likewise we can hold quite different views of our teachers and schools compared to our views of our politicians’ use of an algorithm to assess student grades.

In fact the path that governments are travelling on – persisting in digitising all public services – will only serve to increase distrust. As people become frustrated with trying to access services, as they become aware of the public money being wasted, the damage to people’s lives and the revolving doors where consultants move to senior roles in the civil service and politicians and senior civil servants move to consultancies, people are bound to become more cynical.

McKinsey claims its report to be a tried and true formula. I don’t think so.

There are three things we all need to know:

  • Digital services only work when demand is simple, predictable and repeatable.
  • As the variety of demand rises, if we use digital channels, we will see a rise in failure demand, so costs rise and capacity is reduced.
  • When we learn to design public services against citizen demands and make IT the last thing to do the cost of transformation falls, services improve, and the cost of services falls – and each of these results is dramatic.

I hope you’ve noticed a pattern. In the past we drove services into call centres, today we drive them into what we call digital channels. The motivation for both is to achieve lower transaction costs. When these services face high-variety demands they don’t work, driving up failure demand. The remedy is to provide a human interaction as the first step in designing an effective service, then we realise the first point of contact isn’t necessary, then we ponder… how much money did we waste creating it?

To access the McKinsey report:

For more on the follies of digital services in the private sector, listen to this podcast on digital services.

For more on the follies of digital services, read chapters 8 to 11 In Beyond Command and Control.