Automating consultancy

You’ve written a lot about the concept of ‘automated consultancy’. Can you expand on what you mean by that? How much of a typical consulting engagement do you think could be automated? 

If we're talking about the extent to which consultancy can be automated, we need to distinguish between different types of consulting work. Where you have a big strategic problem that requires a creative solution, there's—so far—probably not much that automated tools can do to help. But a lot of other consulting questions—‘Has our strategy landed correctly?’ ‘How innovative is our organisation?’ ‘Are we hitting our transformation goals?’—are really just data questions. And AI can play a big role in helping clients solve these sorts of ‘soft’ issues, where you cannot find the answer in the corporate administration.

Right now, we're seeing more and more of what I call ‘activity-based consulting’, where clients want someone to come in and tell them how to digitise their processes or how to become more innovative. And I think a lot of that work could potentially be automated; despite what the client might think, their issues probably aren't wildly different from the ones that their competitors have.

Consultancy, as it currently exists, is an elite product for the boardroom: More people buy Chanel than buy consulting services! It's rare that a middle manager can bring in consulting support—and if your company is under 1,000 people, it's unlikely that you're using any consultants at all. In the Western world, 99% of consultants serve 1% of companies—and globally it might be closer to 0.001%. But if you can automate part of the consulting process, then you can enormously expand the range of companies that you can sell to—and you can be much more granular in how you solve their problems.

What would you say to people who think consulting is, by its very nature, an activity that requires a large amount of creativity and human thinking—and so will be difficult to automate?

There's a perception that automating consulting activity is harder than it actually is. Everyone thinks their own organisation is unique and requires completely bespoke advice and support. But research suggests that's not the case: For example, it turns out there are roughly six flavours of team effectiveness. Companies are not so much unique as they are unique combinations of known ingredients. If you can properly categorise and understand those ingredients, a lot of the surface-level complexity starts to fall away, and you can start to automate the process of solving that company's problems.

A lot of consulting projects, to be blunt, consist of the same kinds of activities: You carry out some interviews, you do some calculations, and then you present your findings in a PowerPoint. If you start to automate some of those activities, you don't just create new efficiencies—you fundamentally transform the project structure. For example, if an AI is doing all of your reporting, then the client doesn't have to wait until everyone has been interviewed; they can have direct access to live data. And if you collect that interview data through automated surveys, then you can actually sample the whole of the organisation instead of just a select few stakeholders: It costs the same amount of effort to get 10,000 responses as it does to get 100.

The wisdom of the crowd will usually outperform the smartest individual in that crowd. If I'm a consultant leading a project, I might have done this type of work five times, maybe 10 times at most—and I'll draw on that experience to help make the new project successful. But what if you could have an automated alternative that looks at the last 1,000 similar projects? That information can be used to create a list of strategic options, which can then be put in front of 1,000 or 10,000 people in the organisation. When you start automating the consulting process in this way, you end up with a totally different dynamic: You start being able to give your clients real-time answers to their questions, and you add in a whole new level of granularity.

In ‘Consulting 1.0’, you spend all your time doing interviews or creating Excel and PowerPoint files; 95% of the time you're busy with production. So you do three all-nighters in a row and present your findings to the board. And if the board says, ‘Great! Can you do this for all of our 327 teams?’, you can't, because you just don't have the capacity. But with automation, you can transition from 95% production to 5% production and 95% analysis. So you're not just reducing the cost of your work: You're transforming the scale you can operate at.

So in your view, automation in the consulting industry is really about unlocking a deeper layer of insights?

A big part of what we've been trying to do with Transparency Lab and with our research is to use automation to ask questions in a different way—so that we can see not just the surface-level trends but the dynamic undercurrents as well. Our approach is based on one simple principle: Don't ask people for opinions, ask for facts. For example, in my PhD research we studied organisational alignment in 3,500 teams—looking at 1,000 different organisations spanning 12 industries in 32 different countries. And what we found is that even when you have positive responses to management's attempt to create organisational alignment—people saying, ‘You did a fantastic town hall!’—that often doesn't translate to lasting change. In 41% of the 3,500 teams we studied, team members themselves were aligned in their choice of strategic priorities but not aligned with the strategic priorities the board had in mind: confrontation or compromise.

In the old days, consultants had to do a lot of manual data gathering across different parts of a client's organisation. But AI changes that. Say you have 300 teams: You can give each of those teams their own interactive dashboard. And those dashboards collect input on the actual situation of the strategic topic at hand, but also provide these teams in return with made-to-measure targets, step-by-step improvement suggestions, and knowledge sharing. Then you can go even further and provide personalised dashboards to everyone in the organisation—and actively use those dashboards to improve operational efficiency. For example, you can identify what's working in some teams and roll that out to other parts of the organisation. Or you can promote networking through those dashboards—‘I can help you with A and you can help me with B, we have to connect!’

Over the last few years, we’ve seen relatively healthy growth in the consulting sector in most markets that we track. So what’s the big incentive for consultants to invest in automation? Clearly, lack of investment hasn’t been an obstacle to growth so far.

One of the biggest reasons for consultants to automate their delivery models is that, by doing so, they can broaden the range of clients they work with. But it also puts them on a much firmer footing when it comes to advising their clients. For years, consultants have been telling their clients that they need to digitally transform themselves—but clients are realising that consultancy itself is one of the most under-automated industries there is. It's like a doctor with a beer belly and a cigar telling you that you need to do something about your lifestyle!

Consultants still cling to their analogue tools because they think it creates a sense of mystique. But that's gradually changing. We do a lot of work with consulting firms—two-thirds of our revenue comes from renting our SaaS automated consultancy platform to them—and we have a service where we will go to their office, pick up their flip charts and post-it notes, and dispose of them in an environmentally sustainable way. At the end of the day, clients don't care about mystique—they want transparency, they want results, and they want partners who are easy to work with. We live in an economy that is all about reducing friction—so consultants need to thoroughly examine their processes and ask themselves whether their existing ways of doing things actually do that.

Adopting automation doesn't just make firms more attractive to clients—it also makes it much easier for them to attract talent. Millennials entering the industry don't want to spend all their time making spreadsheets—they want to work for the firms that have embraced new technologies and where they feel they will be able to spend more time doing complex analysis and detailed thinking.

Do you think the role of a consultant—and the skill profile they need—is going to change as a result of automation?

I don't see products like ours as something that replaces consultants. Rather, it's about giving them more tools. It's like a doctor who only has a stethoscope; if you give him a 3D scanner, suddenly he's going to be able to serve many more patients and give them much better diagnoses. A webshop is open 24/7, so why isn't a consultancy firm? And I do think that these new tools will change the role of the consultant. As the capacity of each consultant increases, the industry becomes more demand-driven. And that means that buyers will have more options to choose from, and consultants will have to become hyper-specialists to compete. It also means that their job becomes less about repair and crisis response and more about assurance and longer-term client relationships.

The American futurologist George Gilder said that the best business models are those that waste cheap resources and are careful with expensive resources. People are expensive and computer processing time is cheap. If she spends two weeks with a client, a consultant can do 20 interviews and make an Excel file and a PowerPoint—or, in that same two weeks she can get 5,000 people using an automated dashboard, she can analyse click behaviour, and she can give the client a roadmap for the next four months. Which do you think the client is going to choose? And what would it mean for the amount she's billing to the client?

What sort of impact do you expect automation to have on the prices firms are able to charge?

As consultants embrace automation, the business model will have to change. It doesn't make sense to bill by the hour; you'll have to charged based, for example, on how many dashboards the client is using or how many insights you've generated for them. Consultant partners may be a bit scared of that, but they should recognise that it will give them a lot more freedom, provided that they're able to build a strong business case around the insights they're uncovering for their clients. For example, we calculated the value of knowledge sharing enabled by our dashboards (the monetised amount of time employees no longer have to spend reinventing the wheel) and it worked out at 500 to 1,000 euros of savings per person per year. In a company with 5,000 people, that's a few million euros’ additional productivity per year. So even if you charge the client just a small fraction of that, it's still a very attractive value proposition on both sides.

Obviously, pricing can be a very sensitive topic. But I don't think it's out of the realm of possibility that the price of a consultant's time drops by a factor of 100. That sounds scary, but if that happens you can expect the market for consulting services to become 200 times bigger. Assume that 75% of firms aren't able to make the transition to that new model: 200 divided by 100 and multiplied by four means that those who remain will be eight times bigger.

We’ve talked a lot about ways in which automation is already transforming what consultants can offer their clients. But what about the future? What do you see as the next frontier for automation in the consulting delivery model?

Already, we offer online do-it-yourself workshops through our platform. What often happens is that the client has, say, 300 teams and the consultants identify that half of those teams need some sort of support. But they can't feasibly deliver that support to all those teams, so they prioritise and pick the bottom 15 or bottom 20. If you automate those workshops, and let teams complete them themselves through an online portal, then you can give everybody the support they need. You will use click data to track how people complete those workshops, and use that information to refine how you run them in the future.

We're also exploring how the process of creating client questionnaires can itself be automated. Clients always want the questionnaires you use to collect data to be highly bespoke; right now, we have a list of 160 pre-made questionnaires that work with our dashboards, and a database of 15,000 questions in our format that we can use to customise them. But we're now using that database of questions to train an AI—so that eventually you can just feed in, say, a consultancy firm's white paper and output a complete questionnaire on that topic in the correct format. And we have developed algorithms that enable us to further tune the questionnaire based on the first set of responses.

I think that in the future, AIs will be able to do a lot more legwork when it comes to creating maturity models for clients. I see it as a way of giving clients ‘traction control’ for their digital, organisational, and technological transformations; the client can decide how fast they want to go, and whether they want to be in ‘sports mode’ or in ‘comfort mode’. It all comes back to this question of control and precision: How do you give clients the ability to find patterns and use that information in a productive way? For me, AI is the answer to that.