The impact of AI on the pricing of professional services

Welcome to the future of the firm podcast. I'm Emma Carroll, head of content here at Source. And today we're going to be looking at one of the biggest challenges that firms are wrestling with today, which is the impact of AI on pricing of professional services. And to do that, I'm really delighted to be joined by Source's CEO Fiona Cheney Afska. Welcome Fiona. Thank you very much. So as I said, it's really one of those topics right at the top of our firm's agendas, but it's a massive topic. So can you try and set the scene for us by how will you sum up what's worrying firms about the use of AI, particularly in their own operating models? So I think we have to start from the point of there's a lot of things worrying firms at the moment from geopolitical stuff going on, but also pressure on firms margins, our clients happy with what they're buying from firms, all of those types of things. There's a lot of nervousness around the model of professional services. AI just drops into that world and then makes it a lot more complicated. And it makes it more complicated because unlike almost every other part of business, professional services hasn't been massively automated or we've not replaced people with technology, it's craft work, it's almost a pre-industrial in some respects. And that's great and true and it's been a source of strength because people want expertise, they want people, it's all about people, issues that they're dealing with. So as everybody I'm sure listening to this podcast knows, professional services have long and always charged for their time. So the building block isn't a assembly line with cars popping out with inputs of raw materials coming in, it's people doing work and that gets counted and money is attached to what the hours that are spent. And if you change that, if you have fewer people or you have the same number of people but doing less work on the particular project, then you end up with a project which is simply smaller. And the obvious conclusion, both the clients and the professional service firms are jumping to is that that means you get less money for it, you can charge less, you want to pay less. So in theory we could go from a world in which the professional service sector is worth $2 trillion to one that's worth considerably less because there are fewer people doing work and the work's not being charged out in the same way. Same volume of activity in a sense but done by by machines and we don't charge machines at the moment. So through a few people it's a smaller industry. So shrinking industry, is it going to be a catastrophe for firms? So we don't think so. What we do think, and we're going to go on and talk about this I know, is that clients see two sides of this, they see certainly a world in which the work that they ask for will require a few people delivering it and they will expect that to some degree to be reflected in price but they don't simply want a cheaper service, they also want a better service. So they see the possibilities of there being uses of uses of AI which are making the service better, perhaps making it faster, perhaps putting more depth of analysis in it. So clients certainly don't jump to the conclusion that this is all about the decimation or more of the of the professional services industry. Okay and you've put some figures on it but to an extent it still sounds a little bit theoretical but I know we've got a model that actually measures the impact of all of this and it measures it on individual firms, on individual services. Fiona, what exactly does that model do? Well we do love a really good model here and so we spent a lot of time thinking about this and what we have is a model that looks at both the efficiency gains but also the quality of service gains and we have to look at both of those because our hypothesis which I think has been proved in some of the data we've got so far is that they offset each other to some degree. So the first dimension is efficiency and efficiency is clearly how many people do you need to do the work. The second dimension is around the quality and value of the work that's being delivered and that can be in the enhanced by AI it can be enhanced by many other things but to get a picture of the future we need to look at those two dimensions not just look at the efficiency part. So for each of those two dimensions we then look at two things. We look at the process the firm uses and we look at the output the firm delivers to the client. So that gives us because we're consultants a very nice two by two matrix along one dimension we've got process imagine that if you like across the horizontal axis on the vertical axis we've got the output and for each of those two axes we've got two parts we've got the human part and we've got the AI part. So imagine in our two by two the top right hand corner is where AI does almost all of the work so it's an AI delivered process but the solution is also AI so maybe rather than getting a porch that takes hundreds of pages you've got a dashboard that's AI enabled that you can interrogate. So both the process and the output are very heavily dependent on AI bottom left hand corner is the opposite. So the opposite would say we can use AI a bit but probably not too much in the process and actually the output is still going to be the output that it pretty much has been all the way through. Change management. So change management is about speaking to people and understanding our causes are that restrict change and slow things down and you could find some AI tools that might help that a bit but primarily you're going to be talking to people and the output is going to be people changing. So there's very little there in terms of the AI part and clearly we've then got two other boxes of the matrix top left where the solution is going to be AI enabled but actually the process hasn't changed that much there's a lot of humans involved in that. Bottom right you've got lots of AI in the process but actually there's still a lot of human intervention in the result. Wonderful so you've given me that lovely matrix which I can picture in my head there which is part of the module. It's very good that you can picture it. Thank you and then we go about calibrating that don't we by actually talking to people. We do so we've just been experimenting with various markets so far so the most recent one we've done has been to look at tax. So tax has high potential to be very very AI enabled both in terms of the process and the output so potentially top right and it's a huge industry so over 40 billion dollars spent in the US alone on tax services so it's a very very big industry and therefore efficiency gains here could really bring down the amount that people earn from tax services and indeed therefore the amount that people pay for them but we're also interested in whether clients think there is more value to be added by AI in tax if the AI is used to help support the output or the process in a way that clients would really find value from it. So for example we might look at something like transfer pricing and say there's some help here that AI can do in terms of assembling lots of data from different countries and looking at the tax regulations and so on but there's also part which we've done by people interpreting that and coming up with a plan for it. The output could be could be some AI enables with dashboards helping them people pay people do things but it's probably also a report. So what we've done is to do surveys of clients to ask them what their hypothesis is their assumptions are about how AI will impact the tax services side of things and what we found was that a lot of clients recognise the opportunity to use AI in the process so if you like it on the horizontal axis they could see that but they also recognise they needed experts so there was a limit to how far AI could deliver tax services it was about 60% of the work could be done by AI the remaining 40% needs to be done by people and then when we look at the output parts that's the vertical axis we found that people can definitely see that there were some outputs from tax that could be enhanced by AI but broadly speaking it's about 50% say that actually it's still people talking to other people about this. So if we imagine our two by two tax does sit in the top right hand corner but to the bottom end of it so kind of not too far from the centre here and that means that when we start to think about how big the market is we're going to take those two things into account. So we've got our matrix that you've talked us through we've then spoken to people to really calibrate that and make sense of it and then also we've got our data everywhere we can pull our market sizing data into that. That's right so the whole the idea behind the new AI model is it really dovetails in with our existing model of market sizing. So when we take the data from the survey we can obviously we analyse that and we then come up with rules apply to the market sizing data that we have. So we already know that the tax market is worth about 42 billion in the US. If we look at the survey data survey data sell it gives us a number which suggests that about 12 billion will get knocked off that by efficiency gains that's the scary number it goes from 42 to about 30. However if you put it the other way around then clients also recognise that AI could be used to create more value and if firms do that and they charge for it and these are quite too big if we will come back to but that actually increases the size of the overall market. So we see the overall market going up to about 49 billion but only if firms think about where to use AI properly and also think about the values that they can use AI to create and think about the wider value they can add to tax services. So this isn't a simple picture it's not simply saying AI is going to reduce everything by x percent it will reduce some things some areas of tax more than others but it also creates opportunities to create new value that clients haven't yet imagined. That's fine if firms do that but firms therefore need to start looking at how they plan for this future they can't just wait and just make some assumptions about we'll just take 30% off there and have to live with it. It's more complex than that. So it doesn't have to be that catastrophe that we had lined at the beginning. It does have to be. It might be a firms don't take the right steps though. So in terms of this insight that we can get from this modelling and this three-stage process pull that out from me again why do firms really need to know this? Because they're making they're all making big decisions at the moment about where to invest in AI and how much to invest. So there is a nightmare scenario in which you decide that what clients really want for this part of tax if we continue on the tax theme. It needs to be done by AI with virtually no human input to it whatsoever. So you've got an efficiency saving of about 80% and that means your market shrinks by 80%. To avoid that firms need to be able to work out where could that happen because it won't happen everywhere but also where is that value that could be created that will offset what is lost. So it really means taking a quite granular approach to let's say tax services and going for each one what is the opportunity here for us to save money? Some of that money goes back to the client probably quite a lot of it they're not necessarily all of it but we're still left with a market that's smaller. So to offset that we need to think about value of this product and by doing that and articulating it effectively we can push the price back up so that clients get something that's more efficient but they get something that's better which is kind of the win-win in all of this. Of course there is a nightmare scenario in which firms spend a lot of money on something that clients really don't want to pay for it's busy and AI enabled but they're not actually that worried about it. It would have been fine just doing it quite slowly was junior person. So people could spend a lot in the wrong places and they could miss the opportunities to create more value and so we would need to speak to individual clients of those specific areas to really be able to use this model wouldn't we? So it's designed to give a sense of things at a generic level so we did tax and we can say okay transfer pricing looks as I would be affected in this way but because every firm does things a bit differently and firms position themselves differently the chances are it will be a bit different for each firm so it's a model also that tries to show what it is that how individual firms will be differently impacted by it and what we do there is around doing a survey specifically for that firm so rather than one of our generic surveys that says this is what people think about tax altogether it's what do you think about firm x's tax services because the results might be a bit different there be some firms that maybe they really do need to look at exporting the efficiency because clients don't have much faith in the idea that they could add a lot of value but there'll be others where they think there's lots of opportunity for this firm to add value and they're not so worried about the efficiency so why build lots of AI tooling to give you something that they don't want? And do you see it being used specifically at a services level rather than overall firm or a geography is best suited to services? So it's a model and one of my favourite quotes many people from home say as before is man called George Box who said all models are wrong but some are useful so the model will be wrong because it's trying to extrapolate from where we are today to what could hypothetically happen in the future but it does it in an extremely structured way but the aim here is not to give people the right answer it's not to give a black and white view it is going to be this much what we're saying is look if this happens and that happens this is what would happen to the market and therefore it's really a tool to help firms think it's not a rapport that says this is what will happen although clearly we can do that it's designed to be something that provokes thought but thought on a variety of different dimensions our worry at the moment is that firms are really thinking about one dimension which is efficiency and that becomes a self-fulfilling prophecy there'll be the risk to the bottom because everybody thinks that's what they have to do and clients always keen to get a discount so that's fine it all feeds off itself and it all get the market really does shrink dramatically it doesn't need to so the real thing the model is designed to do is to prompt thought and debate about what firms can do to improve the values that they add and use AI to increase the value that they add while also acknowledging yes it will have some impact on efficiency so final question I think about the insight that comes from our model what do you think that there are one big thing that firms just can't risk ignoring about the impact of AI on pricing I think it's this value point because I just said it's a vicious circle where everybody starts to look at the the the efficiency gains from this it's very it's much easier to think like that the real risk here is that because the value part is harder to think about firms decide that hard is actually impossible and they don't innovate enough and they're not thinking about it deeply enough to be able to push back and say no we've used AI in this way and that gives us a much much better model so value there because value Fiona that's the bit that increases the size of the market again that's exactly right and that's what people really really need to focus on in other words there is space to grow the market not just space to shrink it lovely thank you Fiona it's a pleasure if you found today's discussion interesting you can find more episodes on Spotify Apple podcasts or anywhere else you get your podcasts to find out more about how we're helping shape the firms of the future head to sourceglobalresearch.com