PegaWorld iNspire 2023: Electricite de France Increases Upsell of Consumer Home Services with Pega

Highly popular with Électricité de France’s (EDF) customer service representatives, EDF’s global solution – named “Conquête” (conquest or win) – comprises a complete marketing system based on one-to-one customer engagement. Learn how this easy-to-use, omni-channel pre-sales process with next best action allows EDF call center advisors and client prospects to obtain personalized offers and price estimates fast via self-service. The results: significantly increased sales of optional home services to gas and electric customers, such as warranties, insurance, and repair services.

Transcript:

- Thank you for coming in this last session for these two business days. I am Olivier, I am the CIO of EDF B2C in charge of the information system on the B2C market in France. And we'll talk about our experience in Pega with Emile Lam during this session. Just some information about EDF very quickly. EDF, Electricity de France in French, is the largest European utility company, mainly on electricity. It's mainly, it's the largest company in terms of production, mainly nuclear and on hydraulic power, and we manage about 40 million of customers worldwide, essentially in Europe. That means in France, UK, Italy, and Belgium. On the B2C, on the B2C, we have this example of A key figure that means the amount of transaction every two minutes. It's this way to tackle with the amount of transactions we have to deal with, and which is quite important is that we have about 20, 23 million of customers that have about 20-27 million of contracts, mainly electricity, but also gas and services, home services. And these consumers, these customers are managed by about 5,000 advisors based in the, located in the 20-25 call centers based in France. Only. In France. It's a choice. And on this, on this slide, there is some interesting numbers of visits on the, for example, on the EDF mobile app. 200,000 million, excuse me, 200 million of visits per year. That means that some of customers decide to come on this mobile app every week or twice a week. And it's due to the consumption analysis functionality we have in this app, and due to the electricity crisis during the last two years, we use a lot this application in fact. So this is a large element of interaction and we have to deal with that. And the question for the project we will talk about was on mainly the competition situation. EDF in France, the market opened in 90, in 27, but the competition began really in 2015, between eight and 10 years to achieve a complete competition. And in 2019, there was about more than 30 competitors on the market, on the B2C market, including big one like NG, not total energy at that time, it was direct energy, but there were some big competitors and a lot of little competitors. So we tackled with that. Between 2007 and 2015, it was quite long life, easy life without any competition. And suddenly in 2015, the competition began really, and some months we have lost more than 100,000 customers per month. That means more than 1 million per year. And at that time, we decided to do something. To do something, we have to, so we decided first to launch a counter-attack and to try to re-conquest the customers, to keep consumers, to search for new contracts on gas and electricity. And the second idea was to enlarge, to increase the average value of each customers by selling more products. That is gas, electricity, and services. So we need, for that, we need a tool to do that because we don't have any solution except our, as I explained just after, except the ACP solution we had at that moment. So we decided to build a tool first to convert the prospect into our customers in order, the good idea to us to produce a cure, a quote, a quotation based on some behavioral questions. And in order to be able to give a price and offer in comparison with the competition, we proposed a tool that can manage a bundle of services because it's, if you want to increase the numbers of services, the numbers of offers, the best way just to sell this offer altogether in a bundle. And it's easier with a dedicated tool. And last idea was to target prospects with large marketing campaign, personalized campaign, and to recontact and to, yes, to recontact the hesitant prospect keeping them inside the system in order to be able to call back, for example. We have another requirement, which was to build a fully omnichannel tool in order to reduce the cost to serve. It's evident, but if the consumer is, the prospect filled the form instead of the advisor, it's a keeper. And last idea was to be able to take into account a lot of specific legal requirements in Europe like GDPR about the data protection for the consumers, like regulatory rules. As we are the next monopoly in increment, we have to deal with that. And for example, we have to manage a strong, a strict separation between customers having, using regulatory offers and the customers that have open market offers. And it's not the same type of sale we can do with that because it's for forbidden by the law. So for the, if you see that, simple. So that was the concept in 2015. And the technical situation and the challenge we have to tackle with are the following: First, at that time, it's true, it's always true. In fact, we used quite old SAP solution. Our billing system is SAP ISU. We use the CRM 7. We have more than 100 applications to connect altogether. So it's a quite complex system. And the main problem that we don't have any prospect modernization, any prospect management. We are only able to manage customers. It's a good idea, but not in a competition. And the volume we expected at that time was based on those 3 million of customers moving each year from house to another house. Because the major sale we can do is when the consumer changes the contract when he moves. The technical challenge... The technical challenge about this project was the following: First, it's a mix of technical challenges. And in fact, it was on the previous one. It's a technical challenge based, due to the numbers of solutions. It's like a change management project because the idea was to change the way to sell the offers to the consumers. And the idea was to change the behavior of advisor. So we have both to build a new tool and to change the behavior on an advisor. It was the key idea of this project and that explains the following story. So we don't have time and money because of the competition. We don't have time or money to rebuild the whole system. More than one 100 applications, different applications. To do that, we have to couple the Pega system, the new tools, with about 30 different applications. So it's quite a large complex job. And we need to tackle with numbers of channels, like the web, like the phone channel, which is mainly used today, but the web at that time was just increasing, in particular, SMS, email and so on. We only have some social media network, social media interaction and so on, like everybody in fact. But we have to tackle with all this interaction and the process we decided to apply was quite interesting. We decided to launch a call for tenders between a couple of integrators with editors. We decided not to buy a solution, not to buy a product index to buy the integration, but to buy the couple, the bundle of solution and integration. And these calls from tenders integrates a working prototype in order to be able to validate the solution on the technical point of view. But even on the field, because we deploy, we test this working solution in three different call centers with, well, an advisor, and well, with customers. And the validation process includes technical appreciation, but includes satisfaction of advisor, commercial performance, processing time and so on. So we chose the solution based on the real experience during several months on the field. And at that time we choose Pega. So how we choose Pega with such call tenders with a working prototype. It was an idea to be sure, that it exists as a solution to our product, in fact, a global solution. Important: If you do that, you have to put, you have to provide a financial compensation for the loser, because if you want to have a working prototype, you need to be sure to have a good solution workable. And to do that, the only way is to put money on the table. So we start in the beginning of '19, 2019 with the contract, and the beginning of the project, it was like a flying start. We started very quickly. We decided many things, but the first idea was to reuse and to keep the initial prototype on the field. So we keep the prototype, we deploy the prototype on the three dedicated call centers involved during the call for tenders. And the idea of that was we want to keep our, the advisor involved in the project in order to have feedback from the field. And it's a key decision to the success of the project. Second idea was to make a quick development on a specific urgent case, which was in the summer 2020, the opening of a gas database of increment providers. It was a unique opportunity to address a lot of customers, a lot of prospects with data about their consumption. So we decided to use this solution to a massive marketing approach in order to keep new prospects. And we did that in less than four months. So it was like, not a miracle, but it was quite, a very good success. And we covered this urgent need very quickly. The problem at that time was it was only a dedicated case. And after that we have to think about the global organization, the industrialization, in order to go at scale. We decided first to use our agile scale methodology, the safe methodology. We launched the first agile train during COVID with our full remote PI planning. It was not exactly what I imagined before. We decided to keep the prototype on the field, to work on the prototype and to increase the capability of this prototype. But we decided to build the final industrial solution. And in fact we were able to manage the two goals simultaneously, that means to provide to the business a quick answer to use case and in parallel to build the final solution. At the end of the day, it was a little bit longer than I expected, but I will explain that just before Emile will explain that. So this phase was our about two years phase. And now just to explain what we are doing exactly, I just propose that Emile will show you a recorded demo with some examples on the specific use case, an omnichannel use case just in order to understand what is this application. And after that we will explain the main recommendation and the success of this project. So Emile.

- So I led the--

- Yes, there is a plane over the building. Okay, so I will so show you one use case of omnichannel, or principle case to make a quotation. So I will begin with the prospect and we are going to finish with the advisor.

- [Olivier] Click on the button.

- [Emile] Click on the button and go. Okay, here you are, we are on the website and the prospect wants to have a quotation. So you click and you go in the Pega portal website. So here, what we want, we are going to ask multiple questions here. It's about the location. And continue. We are going to type of home, size of home, number of inhabitants. So we need to collect different information about the home, the habit and the usage to determine which offers to propose. After that, we're going to call multiple APIs in EDF, and we are going also to call the API inside the container of CDH to propose the best offers to this prospect. In fact, we manage offers and strategy inside CDH with designer. So here we have this calculation.

- [Olivier] It's a well-timed demo track.

- [Emile] We're staging it, recorded, but it's, so here you can see the bundle provided by CDH and the three categories of offers that we provide. Now the prospect can finish the letter. So he give the email to receive an email to come back later with a link. So we have different information. And to continue with the advisor, the prospect decides to call the advisor and the advisor asked the prospect to give the email to find him. So he got the email, and he will find it. So you find the prospect, and now we're on the interaction of this prospect, we find the last case of quotation. We have the view of this quotation, and we can go in modification of this quotation, this case here. So we can add, remove some of, first, but we can also change the answer, add some information or change information. And so, was a question. This is an upsell. This is an upsell proposition in fact. And so we call, again, CDH, we have DIH-BIN-DEN, we have geofence, and we have also the exploitation of business. We can remove, we can add, and now we can accept. So this journey, so it's finished here, but these continue for the contract in another application. We can retrieve all the information already.

- [Olivier] Maybe we can just add at that time that, this is only a new interface for the advisor, but in fact it's fully connected to the world system. And today at the end of this part, the contract is downloaded in the SAP application. So it's a real efficiency. We have real efficiency in terms of cost to sell.

- [Emile] Yes. So to continue, you are going to speak about the success in our recommendation and we begin with the technical difficulties. So if you understand, well, so we begin with the prototype quickly and dirty. But now you want to create a project and to address in the long time and accelerate. So you need to pay back your refactoring. And so as first about the design system, so the first advice that I can provide is to create your component with an expert UI directly inside Pega, like that your developer can use that directly. The second is about the case. So about the case, think about the logic of the business. We want to change to divide your case as well. And to continue, it was about the data over there. So you need to master your data model. You need to master your data model, you need to manage that, and you need to put your business data outside of the case because you want to use omnichannel. And omnichannel, what you want, you must do is to manage your data, also engage in your database, et cetera, because the omnichannel is to have your source win, so data source win, okay? And centralized, so your transformation of this data, like that you will accelerate your evolution. And most important also thing for your developer, to be consistent. So about the recommendations, so the first thing, take time to have a good architecture, take time to have the best practice in your development is very important. Being company with experts of Pega is very, very important at start and during the project. Those are difficulties that we have. It's about the language because we are in France, so we want maybe to speak French, but it's difficult when you speak in French, in Spanish, in English, et cetera. So we try to unify this with speaking to facilitate your exchange, et cetera.

- In fact, the question is about French skills on Pega in New York, choose French skills. So you have to tackle with English. And when we work in a company in which English is not the working language, it could be difficult. So it was a difficulty definitely.

- [Emile] Yes.

- Yeah.

- So to continue with the three other things, so keep it simple. You are going to have a strong hub. Start with DevOps. Very important. You have all the tools that you need to have a good maturity of DevOps inside Pega. So go on. And if you have other tools outside of Pega, you can connect. It's not a problem. And thinking pattern, it's important to reuse to accelerate your development.

- And maybe an idea which is not on this page, I think that when we use such a new solution, you have to think about the final user, about the marketing, about the business, the business manager. And they have to be, they have to learn the function, they have to learn about the tool. If not, if only IT has knowledge about that, it's very difficult to have an intelligent use case. And it's a key point I just discussed today, but it's very important.

- But it's correct because, about the build of the case?

- [Olivier] Yeah.

- The business must understand that. It's very important.

- So it's a key point.

- So are you, did you continue the--

- And just to conclude, so the project, the project duration was about two years and a half. It's longer than I expected before. At the beginning with the first move in first months, we expected to go very quickly. It was not the case because we have to rebuild a lot of things. We have to put the basis, and maybe it would be better to do that at the beginning. But nevertheless, the project is today a success. We have many, many, many interesting issues. First, we were able to deal with urgent needs at the beginning and during the work project. It was very important, thanks to the safe approach to do that because for each PI planning, we we are able to tackle with, to have the trade off between the business case, the urgent business case, and the necessary technical improvement. So that's, that was longer for the technical improvement because of the business case, but at the end of the day, we deliver the value. So it was important. Second point, we have guaranteed a very good acceptance of the final advisor, the final users, because they're involved during the whole process. They're involved by testing the prototype, by using the different version, and by contributing to the design of screen, the design of the march, the design of the new sale behavior. It's very interesting about that. At the end of the day, we have good commercial results because we're able to compare the commercial results on call centers using the tools and the user and researcher tools. We increased the upsell of services by about 57%. On the digital, we increase the transformation of about 30%. So with such a result, we decided to deploy the tools on the field. I think we have now a quite reliable solution without any interruption, even when we put an upgrade on the, we put an upgrade on the system. And thanks to the NBA, we are able to change very quickly the commercial policy because we have made a very parametrized bundle of policies, commercial policy. And two years ago it was not a need. But with the existing crisis in Europe during the last two years, we are obliged to launch new offers and to change every quarter, the commercial policy. It was not the case before, but now it's our work and it's a good thing for the business to know that it's possible to change the commercial policy next July with the decision in the beginning of June. So that's a good point. And so for this reason, we decided in last April to deploy this tool on all advisors, on all sales and presales situation, and the rollout is on the way and will be finished at the end of June. And I hope that it'll be a success. We'll know in next September if the results we have measured on the test call centers will be replicated on the field. We'll see. It's a hope at that time. So I think that the project is quite a successful idea, and bigger too. But for the next project, I keep in mind, and we keep in mind all the difficulties we have uncovered and all the advices we strongly recommend to apply. And just to keep it simple, it takes more time at the beginning, takes more time to think about the solutions and to find the good competencies, the good skills. It's very important to take more time and more expertise at the beginning, expertise from the editors and maybe others, but from the editors first in order to be able to go on the good track. Thank you.

- Thank you.

- So just in time, if you have a few questions, we are okay to answer. If not, it's beautiful weather outside, so no problem.

- [Person] Have a question.

- Yeah, there's a, yeah, I think there is a microphone just in the... I think there is a microphone in the area just there. There is two microphones on the, towards me.

- [Person] No Olivier, the question was maybe if you can share a sense about the magnitude of the deployment, the number of agents that are expected to use this application,

- The number of agents in terms of--

- [Person] Advisors, advisors.

- Advisors, yes, we have five, up to now, I think there is about 1,800 advisors using the solution. And this summer we will achieve 5,000 advisors. In fact, we have about 5,000 advisors in France. They split between sales and on the customer or customer self care, I said customer management. But during the summer, due to the amount of moves of residential, almost every advisor has to work on the sale. And it's very interesting to have such a tool for advisors having a low competence, a low skill about sale, because it's a way just to push a bundle to the prospect. So yes, it's about 5,000. And for the deal digital aspect, we hope that we can increase the numbers of sales on digital. Today, it's very few. But the first experiment we have made with this solution are quite impressive. So I hope we'll have more self to care subscription. It's cheaper. The average amount of margin on each electricity contract is very low. So we need to acquire our customers with less than 20 euro. So if the advisor uses 20 minutes to sale, it's more than one year of margin. So we have to do something. Yeah. Guys, another one there.

- [Person] Hi, very impressive use case.

- Could you speak a louder?

- [Person] Hello, can you hear me?

- Yeah.

- [Person] So as you say, you have followed the safe methodology. So you went live in incremental way or you made the whole solution and went live as a full rich product.

- You talk about the safe methodology.

- [Person] Safe methodology. Safe is all about incrementing in .

- Yeah, we use, we apply the safe methodology for many reason. First, we need to have about, at the peak of the production, the peak of development of the project, we did have four Pega developers, four Pega teams, development teams. But we need to have some digital skills, data lake teams development, and microservice teams expertise because part of the algorithm we use for the pricing are made inside, are made internally on the microservice solution, the microservice internal platform. So to do that, it was a, from my point of view, a good idea just to manage this agile scale using the safe methodology. But it was the first experiment for us, and it's not so easy to involve the business in the safe methodology, but for the IT, it was quite a good solution.

- And in addition it was more than 100. So when you addition all the team, 100 people, so it's important to these team.

- But one of the difficulties was in fact the COVID because the first PI planning was in March, 2020.

- [Person] Oh.

- So in March, 2020, we have to invent and to find the good tools to have a PI planning with almost 100 people everywhere at home. So, okay. It was a funny experience, not so funny in fact. But I think the agile scale is a good way if you have to manage a business case, and technical issues simultaneously. If not, it's very difficult to have the trade off between the development.

- [Person] Thank you.

- Thank you. Yep.

- [Person] If I understand well, you use CDH. We've heard this morning and the last two days about generative AI. What's your view about this? What's your next step in term of using AI?

- Yeah, it's a good question. In fact, it's not so easy to involve the business in using innovative tools. And CDH was used in the application in order to propose the better bundle. So the better bundle. So it's a way to modelize the commercial rule, to modelize the commercial policy, and due to that, we are able to change that very quickly, and it's understandable by the business, understandable by the advisor. It's a good way. If we want to put AI in that, there is a lot of barrier. We are trying some prototype now, trying to launch some prototype, but it's difficult because first we don't have any data, so much data to do that. Up to now, we deploy the solution only on a few parts of the activity. So if we, there's a lack of data for the deep learning or for the learning in the general. So it'll be possible I think next, in the last quarter of this year. Okay, first point. And the second point is that I think there is some, yeah, it's difficult for the business to accept that decision could be taken by the, by the tool. So it's easy with auto explanation, with an AI that is able to explain the choice, but it's a black box. Up to now I think it's difficult to do.

- [Person] So it's more a change management aspect that--

- From my point of view, AI could be a good AI, generative AI could be a good approach, for example, to personalize a little bit the emails and letters we send to the customer. We can work on maybe on code generation, but it's a little bit late in this project. So I will try to find some business oriented use case giving more value. And if I find that, if I will find it, if I find that, we will test to, we'll test it. And first we have to convince the business, it's not so easy. Not so easy. And it was the reason for what I think that, when the IT division decide to use a tool, we have to provide some formation from learning to the business. And especially the case about the marketing approach of Pega because the marketing approach is a very personalized marketing approach. And former use of the marketing was a massive, massive marketing traditional approach, and it was not compatible, in fact. So that's the game. It's a complete change management, in fact.

- [Person] Yes. Thank you very much.

- Thank you. Thank you very much for your attention.

- Thank you.