Evolving the Enterprise

Mastering Digital Transformation with Dentsu's Mark Clare

• SnapLogic • Season 3 • Episode 12

Share episode

Join us on this episode of Evolving the Enterprise as we dive deep into the world of digital transformation with Mark Clare, Chief Architect at Dentsu.

Mark shares his journey from financial services to leading technology strategy at one of the world's largest advertising and media companies. Discover how Dentsu is leveraging AI, automation, and innovative data management to enhance operational efficiency and drive business value.

Mark's insights on creating a unified enterprise architecture and fostering a culture of innovation are a must-listen for anyone involved in enterprise technology and business transformation.

Sponsor
The Evolving the Enterprise podcast is brought to you by SnapLogic, the world’s first generative integration platform for faster, easier digital transformation. Whether you are automating business processes, democratizing data, or delivering digital products and services, SnapLogic enables you to simplify your technology stack and take your enterprise further. Join the Generative Integration movement at snaplogic.com.

Additional Resources

Mastering Digital Transformation with Dentsu's Mark Clare

Welcome to Evolving the Enterprise, a podcast that brings together thought leaders from the worlds of data, automation, AI, integration, and more. Join SnapLogic’s chief marketing officer, Dayle Hall, as we dive into the captivating stories of enterprise technology successes and failures through lively discussions with industry-leading executives and experts. Together, we'll explore real-world challenges and opportunities that companies face as they reshape the future of work.

Dayle Hall:
Hi, welcome to the latest episode of Evolving the Enterprise, the podcast where we get to dive deep into the strategies and technologies that are driving business transformation. I'm your host, Dayle Hall, the CMO at SnapLogic. Actually, I have to say these are probably my favorite podcasts to do. No offense to anybody else that we've already had on this podcast. But when I get to talk to customers, when I get to talk to people that are out there and using the technology every day, it actually reminds me of what is going on in the world, how capable we are of making change. Our next guest is no stranger to that. Mark Clare is the chief architect at Dentsu. He leads all the architecture and technology strategy. He's got a ton of experience in digital transformation, cloud infrastructure, automation, and the growing field of AI. He's an expert in steering complex technology and business ecosystems towards innovation. Mark, welcome to the podcast.

Mark Clare:
Thank you very much for meeting me today.

Dayle Hall: 
Yeah, we're glad to have you. What I usually like to do, Mark, is to give the audience just a little bit of insight on your background, how you came to this role, and maybe even tell us what chief architect actually means and how you got to where you are at Dentsu.

Mark Clare: 
Yeah, okay. I've worked across a number of industries, mainly financial services and investment banking. I've had a very traditional route to becoming a technology architect. I've been building trading systems and financial services systems and working for software houses as a developer, engineering manager. About 15 years ago, I moved more into becoming head of architecture across large organizations. What I've been doing over the last 15 years is, as a chief architect, you have a number of key roles. One, you're there to help the organization steer its own internal strategy and align that with a technology strategy that enables growth, agility in the organization. During that, you've got the right level of governance and control over what's happening on projects to make sure the right investment decisions are being made. Also another key one is that, and more and more important, is how you can actually drive innovation in a business where technology is offering new avenues of revenue within an organization.

Before I joined Dentsu, I was working at London Stock Exchange Group. It was a very product-centric organization. We were building out new data products that were being sold to investment banks, government organizations. That's a multi-billion-pound business. When I was working there, we were having a lot of joint ventures with people like Microsoft, FINBOURNE, who's a wealth management data company, a number of AI companies. We had a really brilliant product with a partnership with a company called ModuleQ, which provided automated alerts, which were targeted to the particular users, diary the customers they deal with, the industries they work in. Every morning they'd come into work and there'd be a whole set of market data and user information, price information about the customers that they're working with on a day-to-day basis.

I moved to Dentsu because an area that I wanted to get more involved in as an architect was more at the enterprise technology and organizational side, so move away from product development to actually help a large global organization like Dentsu become more agile to reduce the cost base, because margins in the business that we operate in, in media advertising, professional services are getting tougher and tougher. And then also looking at how we can innovate within the company, how we can provide new services to our customers and, internally, how we can use AI and automation to really improve the overall operational efficiency, but importantly, the agility of the business to go out and win new customers, to provide better customer service, and to be more agile to deal with the competitive pressures that run in every part of Dentsu’s business.

What were traditional competitors like Omnicom or WPP or Publicis, now every single consulting house is eaten into the different parts of the Dentsu business. The competition is becoming wider, and we need to compete with that. I love working as a chief architect because half of what I do is around business transformation, and then half was around how technology can support that. It keeps my interest from a technical perspective, but it also enables me to work on what are the best business models that we should be adopting, have all of our operational units more efficient, and working directly with the markets and the regions in terms of how they can actually become more productive and compete in their local marketplace better.

Dayle Hall: 
Yeah, it's interesting because sometimes when you see this type of title, chief architect, probably to the layman, it would feel like a very technical role. Maybe a little bit more on the strategy of the technology, but I think what you're telling me is it also has a lot of focus on business value and how you can drive the business, even from the technology side. So how much are you involved in business decisions as well as obviously the technology side?

Mark Clare: 
I'd say it's becoming almost 70% plus is really working with our internal practice areas and business units rather than technology teams. Most of the activities I'm involved in are really looking at what's the best operational blueprint for the company. Dentsu itself is going through a very major transformation activity. Dentsu, like many media and advertising and professional services companies, has grown through acquisition. It was almost like 600 companies within an organization with literally hundreds of CEOs and CFOs at the market level and the practice area level.

I'm heavily involved in as we're moving to what we call a One Dentsu operating model, which is to say, rather than have all of these little fiefdoms in your organization, now we're going to be heavily client centric, everything is targeted at the client. So Dentsu becomes almost a large resource pool that you can use to service the client needs, whereas previously, it was every single practice area and every single brand and every single market really owned its own P&L, it operated almost as a little silo. We've got a huge task now how we bring that together into a more integrated company. That's a really challenging activity, especially as we do not have the luxury of just saying, right, we're going to rip apart our enterprise architecture and our technology estate and put everybody on a single set of ERPs.

There was a previous strategy for Dentsu to move to global ERPs. Through a number of reasons, and we've learned a lot of lessons, that did not really give the returns that people were looking for, and I've only been there for 15 months. The approach we're taking right now is to look at how we can federate that more autonomy down at the customer-facing level we’re at, at the market and regional level. So moving away from a centralized IT-type organization to pushing that out to individual markets, clusters, and regions.

That's a really interesting aspect of my current role, is to explore how do we do that in a cost-effective way. You want to allow the autonomy, but it needs to be within a set of guardrails, especially data guardrails, because to run a large enterprise, the whole enterprise is driven by up-to-date insights and analytics on what's going on about customers and all the project work from our financial position. We've already got over 500 enterprise systems. If we continue allowing the markets just to do what they want without any guardrails, it will just become worse. So how do we shape that? We’ve got the right framework internally that allows the markets and the regions to optimize their operating model but still have some interoperability across the organization.

Dayle Hall:
One of the things that we hear a lot, obviously, is you mentioned the term data guardrails. With so many entities and different systems using 500 enterprise systems, what kind of guardrails are you talking about? How can you manage digital transformation and try and bring all this together?

Mark Clare: 
That’s a good question. The key aspect is that we've really had a very strong focus on master data. This is all reasonably common sense. But the key master data items in any large client-centric organization are client and customer management, supplier management, financial dimensions, company setup, their service catalog. All of these things, they sound simple, but if you have every single part of the organization creating their own client record and using different financial dimensions and cost centers and project codes and chargeability codes, it becomes impossible to bring all that together to provide the right data and insights at not just that senior level, but even at the local level.

We've got a program at the moment, and we have a very effective chief data officer. That was another initiative that Dentsu took on last year, was to make sure we have a chief data officer. And it's really a case of, one, identifying those entities that are important, making sure that we've got the right data stewards in the organization, so we have somebody who owns client data, who owns our supplier data, and then implement the right technology and process to capture that information and master it and then distribute it down to all of the relevant systems across the organization.

In parallel with that, we've then got to look at really effective data governance. Obviously, we're subject to many different privacy and data retention laws across markets. That is a huge initiative that we're going through right now. But from a guardrail perspective, it's really on any project or all of our systems, we're making sure that those master data items are being utilized. I'll give you some simple examples. We've got a common set of guardrails now on every client project that’s set up. There's a set of project chargeability codes that are there so that we can even understand through our time, the time that our professional services and creative teams are spending on client projects, can we analyze that from a utilization perspective? Can we push that into P&L reporting directly?

Up until recently, even producing a global P&L took probably a quarter to produce. You cannot have an organization that cannot provide their data and insights even down at the client profitability level. We’re resolving that now because we've got common data guardrails. We’re using products like SnapLogic to pull all the data into a number of different data platforms, and we can provide that more real-time analytics and data insights into our internal stakeholders. [inaudible] because we don't have the luxury of moving everybody onto one ERP. We've got 70 finance systems. So the key now is to make sure that we're consistent at least at the data level across those 70 finance systems, which is a massive undertaking by itself, but we don't have the luxury of pushing everybody onto SAP, or Microsoft Dynamics, 365, or something like that.

Dayle Hall: 
I want to go back to something you said about it was an initiative to bring in a chief data officer, because I think I'd be interested to know, I think the listeners would be too because what we usually hear from the people that listen to the podcast is they have some similar challenges. And listening to people like you talk about these helps them to channel themselves, know what to do, know who to go talk to. With this level of systems, and you said that they had an initiative to have a chief data officer, how do you work with a chief data officer? And what are the other stakeholders that are involved as you try and pull all these systems together and create the guardrails? How hard is that to actually do, Mark?

Mark Clare: 
Well, we're still in the process of doing it. Never underestimate how hard it is to do it because, first of all, you have to win hearts and minds and persuade people why it's needed in the first place. Because people say, why do I need a master data management program? What value does it provide? We've got teams that are happily creating their client records and pushing them in their local systems. First of all, you have to put a valid problem statement or give a business case on why it's needed. What we did in that business case, we just highlighted horror stories within the organization where poor data management and poor process guardrails has led to quite a high- there were some significant operational write-offs because of poor data management and poor process, where there were teams that weren't invoicing correctly, or we were putting revenue on the balance sheet because their whole process for managing that wasn't handled properly. Invoice for it, we had to write it off.

So we pulled together all of these examples and highlighted the true cost of not having high-quality data. We also used benchmarking. So we highlighted where some of our competitors, people like Publicis, have a very strong data management ecosystem in their organization. Publicis are probably one of the most profitable competitors compared to Dentsu. We highlighted how they had a very strong data governance ethos in the organization. They had also the luxury that they consolidated down to a smaller number of systems in Dentsu, but that, together with proper data governance, has made them a very effective organization. There's some other reasons. When we did the benchmarking, we looked at how our competitors operate. The ones that are performing highly have a very strong approach to data governance and management. It's critical because if you don't have the luxury of moving to a common set of systems, you have to ensure that they can talk together. And the language of doing that is those data dimensions, gotta get those right.

Dayle Hall:
Yeah. Well, I like the concept of you create the problem statement. I'm sure horror stories really do highlight to everyone what couldn't happen. Talk to me a little bit about, without detail, without specific details, but what about things like then getting funding for maybe technology you want to bring in, for maybe resources that you need to maybe bring in from other parts or maybe hire? If you're winning the hearts and minds for the initiative, how do you go about, okay, but this is what it's going to cost? Do the executives immediately look for, okay, so what's the ROI, when we're going to get a return? How do you go through that?

Mark Clare:
Over the last 15 months since I've joined, previously, as there were a lot of big programs going on in Dentsu with multi-million-pound, multi-year programs, we've pivoted away from that, and the approach that we've been taking around digital transformation and the funding for that is we've been partnering with specific key markets or clusters in Dentsu. We have this motto, it's called locally led, globally connected. Whereas in the past, it was driven from the top and it was viewed as a top-down initiative, we're now working with a cluster of markets in our EMEA region. And they went through a very bad patch with quite a number of operational problems.

We've come out of that, and what we're doing now, we're working really closely with that market, getting funding from that region, so we're getting regional funding, to look at what's the optimum blueprint for operating across their practice areas and their business lines, putting in place a much more integrated technology or estate that supports that cluster. Because we're letting them drive the project, from the center, myself and the chief data officer and the head of process, we're guiding them in that process, but they're leading it. That's looking like a successful model because it means we've got a large pilot in your organization that we can demonstrate the return on investment quickly, and then we can say that model and that blueprint, we can roll it out to other markets. So that's that approach. That approach is much more successful than trying to have any form of top-down design.

I mentioned this concept of the One Dentsu operating model. What our CRO has said is that he wants some guardrails on how that gets implemented, but he's allowing the regions and specific markets to define their own implementation of it within the guardrails. In the past, Dentsu would have said, here's the process, go and implement it. And that just wouldn't work globally. Now it's more of a collaborative bottom-up approach, get it working in that environment, show that it’s working, show the return on investment. And then what's happening is that other markets are saying, actually, I want to adopt that approach. I want to use that process. I want to use that end-to-end automated stack of technology. I know I've got some autonomy, I can go and do it myself. But actually, you've shown it's working, let's go ahead and use it.

Dayle Hall:
And then each of those local areas, are they responsible for their own ROI? Or do you roll everything up to the-

Mark Clare:
Yes, they're responsible for their own benefits, realization, and justification. They're being measured on a set of OKRs, key performance indicators now. The practice areas are being measured on boost in their utilization. The client leads in that environment are responsible now of showing how it's improving their profit and loss and improving revenues in that area. It's down to them. They have to get the funding. It didn't come from the center. They had to get it from the EMEA budget. So they've got a stake in the game, and now they've got to justify that funding.

Dayle Hall:
But again, given these guardrails in place but let them execute locally, you feel like they would prioritize part of their budget to do it this way? Because you can show the value, but you give them the opportunity to execute it themselves and not just force it down their throat.

Mark Clare:
That's correct. They liked that approach, but they also liked what we call the globally connected aspects. They don't want to be left to their own devices. What we've been doing from the center, from the architecture team and our process and data teams, is really helping them reach out to other parts of the business where they got part of that process working really well. So we have collaborative blueprints and exercises with, say, the US and EMEA teams. And also, we're just helping them with the technology choices. It's working very well as a collaborative exercise where it's driven by the region or the local team who are responsible for gaining the benefits, but we're just helping them in that along with that process, but not telling them what to do.

Dayle Hall:
Yeah, I like that. On the same vein is ROI and return, could you talk a little bit or elaborate on things like how you're using or being able to do better process orchestration or automations without necessarily a ton of increase in overhead? How are you managing that either, maybe it's an example locally, given they're executing it that way, but how do you address that in Dentsu?

Mark Clare:
Yeah. That's a very good success story within Dentsu. I'm not taking credit for it. It was created over the last couple of years. Internally, an internal team started looking at how we can use a range of automation tools, things like UiPath, SnapLogic, Power Apps, but they set up initially a centralized automation Center of Excellence to really drive out identifying use cases for automation and then providing the solution to that in the different marketplace. But as quickly as possible after that started, they started up what was called a citizen developer network. What we did there is that we basically provided training, tooling, a really robust support network where people at the market level are trained in identified automation opportunities, and they're trained with the tooling that we've got internally to provide that process automation.

Now I know that UiPath and things like that are often used as sticking plaster mechanisms. But we don't have the luxury of replacing all of the systems, so you have to use products like that. And then we’re also using SnapLogic as well. So we trained all of the local IT personnel. We have a support structure, so they don't need to rely on this central team anymore. At the market, cluster, and regional level, they've got a huge backlog of automation requests, but they're fulfilling those themselves. And over the last 18 months, there's probably been about 3,000 person days of savings, at least from just introducing that type of automation.

Dayle Hall: 
Yeah, that's interesting. I have a question specifically about citizen integration, because it's a term that's been used heavily across the market. Analysts talk about it all the time. Just on a call with Gartner this morning, and they're constantly bringing up how many people get access to our own platform here at SnapLogic. Explain to me a little bit about what citizen integration means at Dentsu. I understand you have this central team. You enable the local IT organizations, but then who within Dentsu are the citizen integrators, and what is it they're actually doing?

Mark Clare: 
The way that technology operates within Dentsu is that we have a core central IT team. At the regional level, you have a regional IT team, but each market has its own developers. Because there, we still have a lot of local development around media systems and data and insights-type work that's needed for data management, Power BI development. That resource pool at the market level was quite straightforward to actually train to use tools like Power Apps, UiPath, and to a certain degree, SnapLogic because that requires quite an extra level of training. What we've done is that we rolled out a very extensive training program to those IT personnel at the market level.

So let's take Turkey. Turkey, we've got a team of about seven people at the local level. Out of that team of seven, three of them are now doing our progressing automation development across the systems they use in that marketplace. They don't even go to the regional EMEA team or the [inaudible] team. If they have a problem, they've got a really good support structure. There is a Teams channel that you can go on to and you get almost an immediate response with somebody telling you how to solve the problem. There's the central automation team that you can reach out to and they'll be happy to look at the problem and help you solve it.

Initially, there was a central automation team of about 30 people. That's gone down to about 10. But we're progressing probably 500 new automations every six months at the moment, at the last count. Now that whole model is now being pushed into the AI space. We have a central AI team, but it's very small. Again, what we've done there internally is that we've set up a whole set of sandbox environments so people can go onto an Azure, OpenAI sandbox, AWS, Google Cloud. They have access to a whole range of LLM or small language models, which are approved. We have an approval process to say, yes, you can do that.

We have a suite of data management tools, and SnapLogic is part of that process, where we're creating anonymized data and enabling that to be used in the AI pilot. People have the opportunity to get trained in the use of that technology. So a lot of developers are going through a training program on that. They have a sandbox environment, and they can play in that environment. They can come up with new pilots that they want to test out in that space. If it proves that that pilot has got value, then the team will help them productionize that application and get it out into- [inaudible] to our customers or our internal teams.

At the last count, our sandboxes were being used by about 750 internal developers. A lot of them are doing it in their own time. There's not so much part of their day job. They've got an idea working with their local customers or their local teams, and they're going into that environment and building out new use cases. Some of them have become really successful. We have one recently where they were creating an automated process to create pitch books in our creative business. That application that was built was reducing the amount of time to produce pitch books by 75%. It was a really successful exercise.

We're using AI tools for things like our internal staffing and resource management. When you've got 70,000 people and 40,000 consultants, getting the right people on the right projects is a difficult process. I suppose in most professional services organizations, it's done often on spreadsheets and Jira ticket systems. We've been spending a lot of time trying to build out a way of capturing all of the skills and experience of all of our personnel, either from the manually key in this data, or we capture project briefs that they were working on on behalf of the customer, interpreting those project briefs. And then we can pick out the experience that was applied by our consultants on those projects and then keep that knowledge base updated. That way, when we're searching for somebody, we can use natural language to say I need this type of person who's worked in luxury goods with these types of clients, with these skills, speak these languages and so on. And we can find the right people across the organization.

Dayle Hall: 
When you set up the sandbox environment, particularly, obviously, with the advent of Gen AI, was there a reticence, or were people really eager? You talked about 750 internal developers using the sandbox. Were people very eager? Was the management chain, people like yourself, the chief data officer, was there some concern around people creating these?

Mark Clare:  
There was some concern- not concern, but there were some prerequisites that we put in place. Our code of conduct in the organization that you have to sign, a part of that is the use of Gen AI tools and the data that you’re using in itself. You basically have to ensure that you've read through our internal guardrails and guidelines on using AI models and the data that goes into that. Then in parallel with that, we've got monitoring mechanisms to see what type of data is being pushed into these models, just to make sure that- and if we're using customer data, that we have tooling to anonymize the data so that we can prove that we're not using any customer personal data in the models. So there are some foundational aspects that we had to put in place.

But once you have the sandbox environments, there's a huge demand for using them, you don't have to go through- in most organizations, to get hold of IT and cloud resources, often it's an onerous process about having a budget code and everything else. Dentsu, probably like many other companies, have taken the view, we're going to foot the bill for these environments because the value that comes out of it is enormous. And the great thing is, again, it's the community that's been set up around it. They've done a really good job of setting up an AI developer community that meets twice a week. People come to those sessions with ideas, problems, questions. Because you've got 60 people on the call all doing similar things, they're helping each other and swapping ideas and solving a problem. And we've got the internal stack overflow-type environments where people can put in their questions, and then people will answer those questions. It's been really well set up. We've got it from automation and now AI.

If you look at how our enterprise architecture is going to evolve, we're definitely moving away from going out and buying off-the-shelf ERP systems. The tooling that's available now enables a lot of our internal processes- we can build applications in-house to do it. So it’s not an application development shop. I don't have 50 dot net developers or ReactJS developers for internal systems. But the tooling now enables you to build new applications, really powerful applications for decision support, analytics, process automation really easily. And that's going to become more and more prevalent.

From an automation and integration perspective, and this is where we're using SnapLogic extensively, in the past, integration was quite dumb. It was effectively, this chunk of data has to go from here to here. Then it became a bit more advanced in that it became more event based. So you became more real time. An event happens here that triggers an API to another part of the ERP. Again, that's a reasonably done process, and events happen and we trigger something afterwards. Now we're extending that so that when the event happens, you can actually use AI tools to determine what should happen now. It's not necessarily just calling another API. We can now hook into decision engines that say, ah, these three events have happened, this is the input data. Based on that, I actually now need to route this information to this person or this other team.

That now starts to enable us to have a diverse tech infrastructure, which we do have, unfortunately, in many respects, but we can now start integrating it in a much more sophisticated way. So I'll give you an example. We joined a mover lever process in a company with 70,000 people with the huge turnover that you get in an advertising company. All media companies have very high attrition rates. That process, with all the different systems, before took an army of people to manage it. And that sounds trivial, but it was- but we've been using products like SnapLogic and some of the automated decision paths to work out, this person's leaving, I don't just want to ensure that we’re stopping their account, and we need to ensure their laptop is sent back. That person would have been working on multiple client projects. They would have had timesheets that weren't approved. We just will ensure that when these events happen, we’re ensuring that the whole process has been cleaned up automatically. So ensure that the project manager is advised, all the activities they were on have been shifted to another person. That is saving manyears of effort and making sure we're not messing up client projects, so it’s important. The integration and automation now enables us to become much more sophisticated at doing that.

Dayle Hall:  
I think it's a really good example. I think one of the use cases that we talk about all the time here is things like hire to retire. But there's very basic use cases like provisioning a laptop and making sure it shipped out. But I think your point of what if that person is on multiple projects, in fly, they haven't approved things, again, multiple people would have to go through the process of saying, what was this person doing? What are the critical paths? Generally, when people leave, they don't stick around and make sure everything's wrapped up nice and neatly before they leave.

Mark Clare: 
Definitely, especially if they're contingent workers. Like in any big professional services company, we have a high number of contingent workers all the time. That fills in the peaks and troughs of demand, and they leave quite quickly. So the key thing with that is that we really focused on what's the end-to-end process map when people leave in your organization. Identify all the touchpoints that could impact. What we've done using products like SnapLogic now is to pick up the event that they've left or they've been terminated, and have a pipeline that basically makes sure that all those events are handled. And that's created a really powerful infrastructure.

We still got some way to go on it. It's not perfect. I could use of more intelligent integration and automation that we're looking at from an ERP perspective. I have built large-scale systems with a use mapping and microservices-type infrastructure, which is heavily based around APIs. My current role is more of an enterprise tech, making sure that hire-to-retire records and reports [leads cache], all that is as automated as possible where we've got 500 enterprise systems. That's a big challenge.

Dayle Hall: 
I want to go back to something you said earlier about the sandbox, the developers testing things out with things like Gen AI. One of the things that we're seeing here is whenever we talk to a new customer or a prospect, there's a new use case. There's something else that we can look at, that we can potentially use some Gen AI capabilities. How does Dentsu go through- because you said sometimes these things then get rolled out. What's the process for assessing whether this is something that you can let the local people decide to roll out and something that could span and have benefit more globally? How do you decide who [inaudible] investment?

Mark Clare:
That's a really good question. The way we operate, what happens is every single use case that people are working on, we log that use case. We've got a really good portal that says, I've got an automation use case or I've got an AI use case, this is the description of the problem, here are my expected benefits that I want to get from that. And then as that pilot progresses, we make sure that that feedback goes into the portal to say, yeah, it did provide the benefits that we needed, and then more we can do. We've got effectively a dashboard on all in-flight pilots and test cases that are ongoing. When these requests come in, as an example, we've got hundreds of requests coming in across the organization, we can match them up and say, oh, actually, that request, somebody is already building this use case. So it automatically gives us a better view on the [pickability] of an automation or AI toolset that we can expand. So I give you an example.

One of our markets in EMEA, they've been working on some of the resource management automation. One guy in the team just says, oh, I started playing with copilot against some resource data in Excel. Actually, it looks quite useful. It expanded. And he's now going to be rolling it out across most of the markets in EMEA. He spent literally, probably a couple of months building out the application. Up till now, we've been using off-the-shelf products like Adobe Workfront for project and resource management. We know these products are fine. They don't have very powerful AI or natural language support. And also they've got a high cost when you add it up for all the people in the organization.

Being able to produce these in-house tools quickly is a game changer. But this is a big problem that a lot of enterprises like Dentsu are facing at the moment in that, obviously, all companies are trying to improve their revenue streams. And so what we're seeing, I and a lot of my colleagues and other companies are seeing the same, SaaS solutions are going up in price quite extensively at the moment. When you start to add this up across a 70,000-person organization, you have to start the question, how far do we go with some of these solutions? Can we find in-house options? Because it's now becoming more cost effective to build something in-house, whereas previously, it isn’t there to build enterprise software. That never happened. It was, oh, we have to buy this product.

Dayle Hall: 
That's a shift, right? Because we all thought SaaS is going to take over everything. But again, in an enterprise like Dentsu with hundreds and hundreds of applications, at some point, you have to rationalize, do we need all these? Build versus buy, I think, is- particularly, Gen AI feels like if you've got some core systems, maybe you don't need all the other SaaS applications.

Mark Clare: 
No, I agree, and especially with some of the smaller markets. Dentsu, like most big advertising, professional services companies, 80% of the revenue comes from the top 20 markets. That's fine. Those markets, we do invest there with the big SaaS solutions. But when it comes down to the other 70-plus markets where their revenues are in the tens of million a year, your expending a high cost on ERP is just not justified. And also, there are much smaller markets. This is where some of the local applications that these local market teams are building are really powerful. Some of our markets, they just didn't want to use our global expense system. They could just build their own expense system themselves with some intelligent routing of the approvals. It costs somebody probably a couple of months’ development in their spare time to actually build it. It's becoming that easy now to do it.

And that's the interesting point about Gen AI in terms of software development. Previously, you had to be a skilled programmer. Now, you can build powerful applications supported by a lot of the AI-supported developer tools, especially with cloud infrastructure for building web applications and hosting them. We've had teams building their own pricing tools, whereas in the past, it was all done on Excel spreadsheets. They're now building integrated enterprise pricing tools from scratch, hosting it themselves, with a team of three people doing it. We can build it now.

But this is where tools like SnapLogic become really important because you use those types of tools as the glue to link these individual components together. If you look at the plethora of small language models that are coming out that do very specific jobs that do not need huge cloud infrastructures to run on, you need a mechanism of gluing them together. We’re using SnapLogic with some of the Gen AI snaps that are there to enable us to build end-to-end pipelines that might use four or five models rather than just one. We have a model that's doing text analysis to start pulling out key parts of a legal contract and then putting that in a vector database. Then we have another model that's doing a comparison between the previous contract and the new contract to see whether the clauses are correct. That can be built internally and just evolved.

Dayle Hall: 
That specific use case that you talked about particularly around contracting, we're rolling out in the next week or so. We're rolling the same thing out here for our own internal teams and finance. They defined the use case. That was one of the things that they identified, the difference between contract terms that we change that sometimes you don't see. It feels that someone should be able to spot those kinds of things. But look at contracts these days, 20, 50, 100 pages, this just enables people to move faster. I think we're going to see a lot more of that.

Mark Clare: 
The other thing that we use in these types of tooling form is because we operate in so many different markets, every market has its own data privacy laws and data retention laws. Now we're working with companies like Finders Keepers who are very valuable too because they do the interpretation of these contracts. But we're also using natural language processing to start to pull out key clauses from the country-specific data privacy laws to say, based on this contract- based on the regulations, not the contract, these are the key points that you need to adhere to. So it just summarizes it, makes it easier to navigate. That's becoming essential these days. We've had to deal with a raft of data regulations that we have to adhere to.

Dayle Hall: 
What's next for Dentsu in terms of big projects? What you've talked about is incredible work. I love the concept, the sandboxes and the community of AI developers, and you obviously have a pretty robust control in place, the guidelines you talked about that are excellent. What's big for you in Dentsu? What's the big project you're looking at? And my last question on top of that is, what are you really excited about in terms of potential of technology, whether it's Gen AI or something else? What are you really looking forward to, Mark?

Mark Clare: 
First of all, I’ll answer your first question. The big project that we're really looking at is not so much a tech-centric project. It's move into this One Dentsu operating model. So how do we move away from a holding company with federally 600 separate companies into an organization which is really focused on clients? That is a very large exercise in terms of organizational redesign, process reengineering, how can we achieve it without spending 50 million on a tech refresh program, which we can't, and how do we have a better use of automation and AI capabilities to actually support that process.

For this year, going into next year, that's the critical- data review, when they purchased companies like Sapient, how can we create Publicis as a client-centric organization that is really well integrated with strong data governance. They really put a lot of effort in that over the last five or six years. And now they’re streets ahead of the competition in terms of revenue, organic growth. You can look at them as a brilliant case study. Our ambition is to try and align with that approach as cost effectively as possible to enable us to provide a much better service to our customers.

In terms of technology, I love the fact that, indeed, I wish I was more involved in it, but we are a very large creative practice in Dentsu. I think the capabilities around video production, computer vision, the ability to create whole creative campaigns, enables our people to become more productive. Rather than get rid of people, they've now got another set of tools that- you can create some amazing creative artifacts now using these products. In some respects, I love to just give up my day job and work from home, but I can't do it. The use of AI to significantly improve the creative process I find really exciting. And I know there are risks. If I'm posing the acting profession, you're worried that you're going to lose your job. But in terms of the ability to create new campaigns for customers, speed up the whole production process so that you can get new ads to market quickly, that is so exciting. I just absolutely love it. And I really hope that that technology is used to make people more creative and more productive and not just used as a mechanism to wipe out the whole creative profession.

Dayle Hall: 
I’m not giving anything away, but I had a similar conversation with the guy that we have that there’s a video. He's worried. What I've tried to tell him is like, don't be worried about someone else doing a part of your job. Start using the technology itself so you can be more productive. Go and see what you can do to offer something else to clients because you can probably go faster, do different kinds of things. But there's always these challenges, right, with this kind of technology where people get nervous. But I think there's a massive opportunity.

Mark Clare: 
Dentsu, I think, is doing pretty well, as I mentioned earlier. I think the support they've given to internal staff to understand this technology on not just how to use it in their day job but actually how to build new applications and use cases, that's critical. And I think most companies are adopting a similar approach. If they're not, you really need to get that up and running because you just want to enable everybody to get excited about the technology, understand how they can use it to help them become more productive. And then they become a more valuable part of the organization. That's what I hope. I want to see this tooling make our internal teams much more productive and provide a better service to clients. If you get more business, then you're not putting people out of work.

Dayle Hall: 
Exactly. Well, Mark, I really appreciate it. I loved the discussion. I loved so many things that you talked about. We're so happy that you're one of our great customers, and we appreciate you've given me an hour of your time today.

Mark Clare: 
It’s been a really good conversation. Thanks very much for the opportunity.

Dayle Hall: 
Thanks a lot, Mark. And for everyone else out there, thank you for listening, and we'll see you on the next one.

As we wrap up this episode of Evolving the Enterprise, we want to extend our gratitude for joining us on this exploration of enterprise technology. Keep the conversation going by subscribing, rating, and sharing our podcast. Together, we'll shape the future of work. Until our next episode. Stay innovative, and stay tuned.