Transcript
Jayesh Bhayani: I'm going to spend some time in sharing our learnings of transformation at scale at Tesco. That's why modernizing technology at Tesco. How it has changed the culture, not just within the company, along with systems, but enabled the growth for the company. From that perspective, after 10 years being at Tesco, I thought it would be good to share the learnings we had from here and how we are looking at the future.
Context: Huge Scale of Our Business
I was in California, in Silicon Valley, for about 25 years, when Tesco reached out to me, saying, would you like to work at Tesco? I had got no concept of Tesco, because I had never come to London before that. When I looked at the website, it was a pretty basic website. The outside-in perspective, when you look at the website of a large grocery chain, feels like there is no technology. This went on for a few months. After that, they convinced me to talk to a few people in the technology department. The state that this technology was, and what was needed for growth and the transformation that was needed at scale, was what motivated me and my wife. We were empty nesters already by that time. We decided, why not? Came here. Just around that time, Trump got elected. We felt even better.
First of all, some context into how large Tesco is. There are about 5,000 stores. Some of them are super. Some of them are medium. Some of them are small. They are across five different countries. We used to be in Asia as well, when I started 10 years back, in Thailand, Malaysia. We divested those businesses. We are still in five countries. We have 20 distribution centers. We have 25 million customers. That's almost one-third of the UK population. That's not small. We have 350,000 colleagues across our stores. Out of that, maybe around 15,000 in head office. Out of that, around 6,000 in technology. Again, it's big. It's big operations, large number of colleagues. We do around 50 million baskets. Not number of transactions, the baskets, which means the orders across our estate in our stores. We do 1.3 million online orders in a week. When COVID came, many companies talk about scale.
When COVID happened, when entire business shifted towards online, our challenge was to double the number of online slots that we can make available for our customers. By that time, for the e-commerce part, our platform was already live. When we doubled it, that's equivalent of putting almost three other competitor companies online in one month. Many companies stopped taking orders. Many companies stopped signing customers. We doubled it in a month. That again shows the scale. Fifty million cases through supply chain, huge. We are not even talking about number of miles the trucks travel and the logistics within the distribution center. We do 4.5 billion calls a day. We have 98% UK coverage when I talk about 25 million customers, but coverage-wise in terms of online business, that can cover the entire geography of UK. We cover almost 98%, which means you can have your home delivery of grocery done in 98% of the UK.
Tills, as we call it here in the UK, POS systems, as called in many other countries, almost 50,000. That's just POS devices. We are not talking about all other thousands and tens of thousands of handheld devices that run on the same platform. That's just the context, because that sets a very different tone for the kind of technology, how mission-critical technology becomes. Then driving transformation through that is quite challenging, because I worked at startups. I've worked at mid-size companies. I had my own startup. Being in California and not having a startup is a crime. I tried that, too. I worked at Yahoo for 10 years. I worked at large companies, worked at startups. Tesco is yet another scale. I wanted to reflect here, because it drives quite a lot of complexities in decision-making of what would you do with your transformation and technology choices.
Why This Matters Beyond Retail?
This goes beyond retail. Any industry that has got physical estate, digital, and real-time operations will face these complexities. That's where we learned about being both efficient and effective. Efficient is do things well, fast. Effective, do good things, which are more effective. A slight difference in nuances. You can be doing fast all the things, but if those are not the right things, you are chasing wrong goals. Do the right things and do things the right way, efficiency and effectiveness. I wanted to share here, this is not just about retail. This will apply to many of the industries who operate at that kind of scale and complexity.
The Starting Point - The Big WHY
Let's start as to why. Ten years back, why did we need transformation? It was quite a poor state within technology I'm talking. When they talked about a large grocery company and I was thinking there is no technology inside, there was a lot of technology, but it was very siloed. Multiple technology stacks tied to historical business structures. When I'm talking siloed, this is not just, store had a POS system and e-commerce had something else. I'm talking multiple stacks within online or e-commerce business itself. Grocery, its entire stack. We had marketplace back then, completely different stack. One Microsoft, .NET, one Oracle. F&F, our clothing, Tesco wine. Customers across these, forget having single account, did not even have single login. Jayesh@groceryhomeshopping is different than Jayesh@marketplace, is different than Jayesh@f&f. We had no single customer view, no data platforms, no product management, nothing. It was that fragmented. There was no single identity, as I just explained.
No single source of truth for stock or pricing. From that perspective, when you have multiple monolithic systems running in private data centers, we were not in public cloud back then. It was not even private data cloud. It was just a bunch of machines. You would be lucky if you asked for a server and you get a server in two months. That was the state. Decades old POS system, 20 years old software running on our POS systems in the stores, completely batch oriented. You couldn't connect to them. The deployments were the engineers would go with the floppy disk and then will deploy the software in the stores. Third-party solutions, not just third-party solutions, they were heavily customized, because when you are at the scale of Tesco, you can demand saying, I need a slightly different thing in your software. Sounds good initially. Then you're locked in with a very customized version of vendor solution for multiple years.
We had software running from vendor solutions that were at least a decade old. You can't even upgrade them. Then heavily outsourced resources as well. These were where we had lots of consulting firms providing resources for us. Not a good culture, not good quality. You can imagine what the landscape was when we started 10 years back. At the scale of Tesco, that becomes quite complex to say, how do you find your journey through that?
Strategic Imperative for Change
That got us the imperative for change, saying we definitely need omnichannel experience, because customer expectations were moving very fast. The old style of doing business was not working. Technology was not helping business. Business had all the motivation to change. Technology was stuck. We had hit the wall, literally. Global coherence across physical and digital estate, both. Rapid change cycles. Back then, as I said, if you talked to a vendor and asked for a solution in our POS systems, it would be six months for minor changes. Regression on top of that, so three things would be broken. You do the testing, and they charge you for fixing the bugs again. It would be six to nine months before you could get anything out on our Tills or POS systems in stores. Forget e-commerce. Then along with that, of course, we needed higher reliability, higher capability. We needed omnichannel experience, higher security, everything.
It needed more capability and capacity, both. That led to us chasing from fragmented stacks to unified platforms that could serve all our channels across all our markets and develop a multi-tenant platform. That's what we developed. It was called only platform. Sometimes platform terms are confusing. I'm not talking just the tech platform. I'm talking about the retail platform that runs our business. We'll talk about tech platform as well. That's part of our learning as to how tech platforms are super important if you don't start early. When we started, it was a retail platform to say, how do we now enable all these businesses and develop the retail platform?
Where Are We Now?
Where are we now? All of our entire UK business and Republic of Ireland business runs on this platform, 100%. We are making inroads in our Central European businesses. There are multiple countries there. They all have their nuances. The trust in technology is much higher. Technology strategy is very well integrated with the business strategy. It's not like service-oriented mindset of say, we just need three things. Go ahead and can you build it? That's how the siloed happened long time back. We'll talk about that in our transformation journey. As I said, the platform is much more scalable but also capable. That's where we are now. We'll talk about how we are looking into the future of agentic commerce from that perspective. Technology has become an integral part of the entire business ecosystem. It's no longer a separate unit that sits in a corner and gets three orders to say, can you develop this for us?
Ecosystem of Transformation
This I describe in my own language. Let's now talk about the journey of transformation. Most companies most people assume that when you talk about transformation, you're talking about system transformation, which is, develop the platforms. We'll talk more about that. That's what I talked about until now. That's not enough. What we had to do along with that was talent transformation. Was very service-oriented mindset, as I mentioned. Most of the development was outsourced. We had to completely change our career paths, the titles within that, the job definitions within that, and completely transform the talent internally. When you want to insource that, that's a huge task. Back then we were only across two different locations, UK and India. As part of this transformation, we started two more, Hungary and Poland, because we had to scale and we needed good talent. The combination of that was required. As I said, the career frameworks were redefined across product, engineering, and program.
Product did not exist. Product management did not exist back then. It was only program and mostly service engineering. The right org structure is super important as part of that as well. You can have talent, but if they're not set up in the right organization, it does not work. It generally should map the business domains and the business structure. For example, in Tesco, there are four major business units, supply chain and distribution, commercial unit where you decide as to, should we sell potatoes and tomatoes that leads to almost a ranging of 40,000, 100,000 SKUs saying what are we going to sell across our estate? What should be the price of that? What should be the cost of that? What kind of promotion should we run? What kind of store design should be there? How many aisles should be there? Which product goes in which aisle? That's completely very complex business.
That's all trading decisions. Channels, which is where online and stores where customers actually buy things. Then customer area, which is mostly loyalty programs and stuff like that. We had to mimic that within technology to say we are going to completely map ourselves into how the company is laid out. You can't run your own ship in a large company and say that we are just different. That doesn't work. Because if you're not sustainable, transformation will not work. That's talent transformation, and we had to change recruitment processes well along with that. The third pillar of this one was process transformation. You can have the right talent, you have the right motivation because you need good challenges for right talent, which is system transformation. If your processes are not conducive, then that fails. The reason why we had silos, not that anything is wrong with program management. We still have hundreds of program managers, but the role of program manager was very different.
Some business person would have a favorite person in technology, as I said, it was a phone call away. No one knew about technology. They will talk to a program person and say, can you get this done? They will run around, talk to a bunch of service engineers, outsource it, comes back a solution. The silo gets deeper and deeper and deeper. We had to establish that from program led, we are going to be an agile product-driven company. We develop products, not just solutions. We'll talk more about some of these, mostly in the systems area. I'll refer to a couple other things as needed, because today is just a teaser of the transformation. I'm trying to compress 10 years of our learnings.
System Transformation: Strategic Framework
That led to, in the system transformation, what's the strategic framework? One of the very early things we had to decide was build and buy. Build versus buy, many people call it, we call it build and buy. Why? Because while we wanted to insource and build the platform internally with insource resources, you will need to buy many technologies along with that. What we said was, we'll always tend towards building things which are core differentiator for our business. That's how we will win. Something that is not a core differentiator, we'll just buy. Most of the enabling function, as we call it internally, which is back-office applications like people system, finance system, everything, they're not a core differentiator. The retail platform is a core differentiator, including the POS system that runs on every single Tills. You go to Tesco, that's our homegrown system. When you use your scan-as-you-shop device, some of you might be shopping in Tesco, you use those handheld devices, completely synced, runs on the same platform.
Every single device runs on our platform. Cameras runs on our platform. Everything runs on our platform. Those are the core differentiator. Where we buy, platform integration is a must. That's how we say. Platform and domain boundaries. Now, centralization versus decentralization. The reason why I mentioned this here is if you're going through this space, and now even more so, 10 years on it's a different world with the agentic workforce that's coming. You can't centralize things. Gone are the days of change control board, architectural board, queue up in front of those pedestals and get your design reviewed and get your code reviewed and everything. We had to decentralize that. That's where we had to come up with very clear domain boundaries, good set of standards. Freedom within the guardrail, that's how we call it. Freedom within the framework with those guardrails to say how are we going to do it, and then you can decentralize that.
Tech excellence. Tech excellence is not just tech debt. We call it tech excellence, because it's proactive as well. It is not just, do we have a few bugs that we need to do? Or, do we need to patch our operating systems? Of course, yes. How are you taking care that you're not going to face into those challenges on an ongoing basis? Back then, there was no concept of total loss recovery. How do you plan for total loss recovery? This is not DR. DR is basic. When you're talking about total loss recovery, it's a completely different ballgame. The kinds of capabilities you need for that, we start to put that into tech excellence. Because it's very easily assumed, even within technology, and I'm not talking outside technology. Of course, business teams don't have that level of understanding. Even within technology, anyone who is not in engineering will assume that this gets done by some engineers who had a good coffee on a good day on the side of their desk.
It doesn't happen that way. You have to dedicate time for that. You have to dedicate focus for that. That requires education. Outside engineering, that requires education. Outside technology, as well, at some point. Where now our business leaders talk about tech excellence saying, are you sure it's built right? It's needed. That's the cultural change we are talking about. That creates the framework for a much faster pace of delivery and clear ownership, as I said, with the clear domain boundaries, and strong and long-term resilience.
The Retail and Tech Platform
That strategic framework gave us, what are the core tenets of our platform? Only customer is not enough. Yes, we are driven by customer. That drives our business. If you only think about customers, that's not enough. For the size of Tesco, when you've got almost half a million people working for you, colleague experience is equally needed. All the systems that we run in our stores, the operational systems that we run for our business operations, everything needed to be transformed, not just the customer-facing applications. Partners, we had very poor rating from our suppliers back then. We had to establish the trust, but also build the systems to make sure that our suppliers, partners, people who give us potatoes and tomatoes to sell for grocery, I'm referring to those partners, not tech partners, we had to improve experience for that. Reuse across multiple different markets and business units.
Offline operations. When you are talking about thousands of edge devices working in your stores, one of the most critical requirements for us is a network can be flaky. Because if you get in the business of replacing network of your entire thing, you might be spending hundreds of millions. Grocery is a very low margin business, it's around 4%. Some are 3%, some are 4%, some are 4.5%. It's a very low margin business. Every penny that you spend counts, goes in your P&L. Cyber risk and compliance built-in into the platform. Super important. This is where we had to innovate across all three public clouds. We did create private cloud, not just a server sitting in our data centers, and edge platform as well. As I was just referring to low margin business, cost is a feature for us, not a constraint. We don't see cost as a constraint.
It's a very different mindset. As part of one of our learnings, we'll go through the learnings, but since we are talking about that, when we jumped into public cloud, and this now will apply into agentic code development as well, and we are thinking through that. The value that gets created against the cost that you might be spending, you've got to be very careful about that. We were spending upwards of £50 million, £60 million on public cloud. I'm talking four, five years back. It was growing exponentially. If you don't control that, then you've got £50 billion business, which was back then, with 4% margin, you're talking about £2 billion. When you talk £100 million of just cloud cost, I'm not talking our salaries, I'm not talking about entire technology department, this is just cloud cost. You can imagine how much P&L you are impacting at that level. It's the responsibility of the technology department to say, how do you keep cost as a feature, not as a constraint?
Then we came up with good principles for our platform, which, as I said, domain design was very important. Create composable services. Data pathways and data quality both matter. You need very clear data pathways as to how data would flow between the various different operating datasets to data lakes, but also data quality across each one of these steps. Security in a layered defense model, all the way from perimeter to built inside. We got tons of different layers of how do you think about security. Multi-tenancy by default, which is, you build the platform once. Even from the physical deployment perspective is the same deployment that works and enables multiple countries for us, multiple different businesses. Build product capabilities, not just solutions, not point solutions. Build products, build capabilities. Then abstractions, all of you will understand that. It's not just for hiding complexity but also when you got mobile applications running across thousands of microservices, running in the data center or in the cloud.
For example, add an item to your shopping cart has got at least 40 different microservices called. Get the identity token, get the product ID, get the customer UUID, get their personalization preferences, get the price, check all that, get the product description. After that, it goes in the basket. Now if you are not creating abstractions that's going all the way from someone's mobile phone all the way to your data center or public cloud, and adding an item if it takes 40 network round trips, you are extremely slow. Yes, it is of course for hiding complexity, but at this scale all that matters. That's what went into our principles because that created cohesion across, as I was saying, digital and physical estate without becoming bottleneck.
Let's jump into cloud and edge platform architecture from that perspective. On the left side you see the retail platform. It's many different microservices. These are domains, and within that there are multitudes of microservices. While there are 60-plus domains within that there are hundreds of microservices. We are across both private and public cloud. As I was saying, when we started, we had only servers in our data centers. We very quickly jumped into saying, this is not going to scale. There is no elasticity. If you ask for a server it takes couple months before you get a server. We literally handed out credit cards to our developer saying just go sign up in the public cloud, because we had to do it in a rush and that led to huge amount of cost. No clear value what was getting done there. Then we had to do the repatriation of many of those systems and re-migrate.
That was one of the lessons learned to say, while you are trying to do that, sometimes it's good to build more centralized capabilities first before you jump into that. It takes a little longer but saves you costly migration. From that perspective, any services that needs true elasticity and scalability we put it in public cloud. Public cloud is a dual-edged sword because when you start using PaaS, Platform as a Service, you get very tightly bound to that public cloud because many of them are very proprietary services. You will never get out of that. You lose your flexibility. As Tesco, at our scale, we like that flexibility. Whether it is when we buy the technologies for our use, deployed or in the cloud, anywhere, we care a lot about saying, does it give us flexibility or not? Because if you're not careful, the license fees keep on increasing exponentially.
If you are locked in, it's a very well-known fact that you will be held hostage. If you don't use PaaS then it's just a compute environment. Why are you in public cloud in the first place? You see, it works both ways. What we try to use is all portable PaaS technologies within the public cloud, that gives us that freedom. It's not perfect. I will not claim that. That does give us an optionality when it's needed. Then, many of our large systems, we are talking about supply chain systems, distribution centers, technologies related to that, we tend to run that into our private cloud. It's almost 80-20, 80% of our presence is in private cloud, 20% is in public cloud. Cost-wise, exact flip of that, 80% of the cost is in public cloud, 20% of the cost is in private cloud. It costs almost 8 to 10 times more for an equivalent setup in public cloud.
This is where as we are talking about agentic code development and everything, a developer could spend almost £500 a day without realizing or without clarity on what value is getting created. Again, if you step back historically this is where public cloud was almost 10 years back. Easy to jump in, hard to explain what value is getting created. The learnings here that we are talking while the technology will keep on evolving is pretty much where you can reapply.
Then comes the edge part of it. As I was saying, when you are at the scale and you want omni-channel experience, how do I now create a platform that is going to power your edge as well? What we did was, the same platform services, the microservices that run in our cloud, they run in our scaled environment with the right abstractions in our edge devices as well. There is a real consistency of business logic. There is a real consistency from data perspective, because when the network is there it's constantly exchanging data. For whatever reason, if the network is not available, we are able to identify the customer. If you have a club card, we identify your loyalty. We fail in favor of our customer. Let's say you did disable your club card literally last minute and network is chopped off and you are in the store, yes of course, that's a real-time requirement, but we fail in favor of our customers.
We take the payment, not cash, credit card as well. Very lately we have enabled contactless payment as well. That's a huge set of capabilities. It's pretty much, you don't realize the store is running offline. The pricing on the shelf edge, that has to be considered. Now as many companies are talking about electronic shelf edge labels, that's part of our consideration in the platform, saying, what happens when the network is not running. How would you update prices on the shelf edge labels? All this gets very taken for granted. Even I wasn't aware as to what's the level of complexity when you are in a retail business with hundreds of thousands of such devices running across thousands of stores. One part which we made sure was, it's the same team which is responsible for edge platform as well as our retail platform in the cloud. That made sure that consistency was there, that ownership was there.
In fact, when we were migrating from legacy stack and we were transforming, we made sure it's the same team that owned the legacy platform as well. You cannot compartmentalize and outsource the problem even within the technology department, saying, A team is building new services and A team will take care of the old services. Never works. You lose motivation very quickly.
This is where we were talking about workload placement. We touched upon this a little bit before, which is, a single team owns that thing. We do use portable PaaS technologies when we are using the public cloud. In public cloud, we generally favor the systems that need scale and they are closer to the customer from that perspective. We, for the private environment in data centers, enabled the cloud as well so that for the developers it's the same experience when you are in the private cloud. We centralized a lot of our shared services. What initially started as, and this is probably a misconception, me being in technology, I'm 35 years in technology, that engineers love doing all these things. Setting up your Cassandra's, setting up your Kafka's, figuring out all the protocols. Yes, all that is true in the beginning part of the transformation. As we started putting more and more production workloads, because early on as part of our talent transformation we decided there is no separate QA team, there is no separate support team.
Of course, first-line support is different, but then pretty soon the call does come to our engineers so they are fully responsible for support of production environment. What started as a good challenge, as a good hobby, learning new technologies, pretty soon became very onerous. When I talk about creating shared services, so I own the retail platform, I own the tech platform, we'll talk about tech platform at Tesco, and I own the applications as well. It's not as much fun for engineers today to say, I'm going to deploy my own Cassandra, I will do the partitioning, and I will do all these things. It gets very onerous at some point. We invested a lot in shared services to say, how do you now not worry about all these things?
The sequence is very important. When we started 10 years back, and we are talking about a company which is 100-plus years old, when we were talking about product development, the misconception in the business teams was, prepared meals in a box is called product. That was the journey when we started. When we started talking about platforms, we had to choose our sequence in the right way. You can never do Big Bang. You cannot ask a 50 billion business to stop and pause and say, we will come back in two years or one year or three years and we'll change everything. Even within that staggered approach, that you see from 2016, this is the real chart. In fact, we stopped counting at 2024, and we had only put significant items here. We had to pick our sequence. Initially it was very far and few in between. We had to develop and create that trust in technology from the business side to say, how are we going to figure this journey out.
There was a lot of distrust when we started. You have to find the right demand, where the business demand, the customer demand is there, and find the right opportunities to do transformation in small steps early on, and then it gets bigger and bigger. Now we get more demand from the business with more trust, less questions. You say, how can we do it right?
Beyond AI in Operations - Agentic Commerce
We have been using AI for a very long time at Tesco. That's nothing new for us. For example, start with all those business units I was saying, supply chain. One of the biggest challenges is forecasting how much customers would buy. If you go wrong in that one, it either leads to loss in millions or very poor customer experience. For example, if you forecast that you will sell 100kgs of tomatoes. It's in tons generally. I'm just taking an easy number. Except if you go wrong, either your shelves will be empty because customers are buying more and you need more than 100kg, or you will end up buying too much and it goes to waste. Waste is a big number. We try to control that, and we have done very well on that one. That's a tiny example, but that whole forecasting is dependent upon what's the lead time to getting the products.
Like tomatoes can be supplied four times in a day in our distribution center. Something like jeans might be six months in advance. Then the shelf life could be very different. Tomatoes will get rotten in few days, non-perishable items will stay on the shelf for longer time, so the combination of that. That's the supply chain side. Distribution, how are you going to lay out all that stock? What's the stock management? What's the truck routing algorithms? Because we are talking about hundreds and thousands of miles of trucks traveling every day. How are you going to optimize that? Good for environment, good for company, good for everyone. How do you fill the truck? This is just touching into the supply chain area. Likewise, when you talk about ranging, you talk about customer experience in personalization, you talk about pricing, everything is AI driven. How are we thinking about agentic commerce?
Because that changes the paradigm completely. What I said is agentic commerce is a business decision. Because if you don't participate in agentic commerce, you can be marginalized. Because, let's say most customers are spending time on ChatGPTs and Geminis of the world, and they are going into agentic commerce to say, we can do grocery for you. Of course, they want that. Nothing wrong in it. That's where customers are. If you don't go there, you can be marginalized. If you go there, you can still be marginalized because you're losing the first touch, customer experience. Because now you outsource that experience to someone else. Hence, it's a business decision. It's not just a technology decision. Getting ready requires complete shift. How would you be ready for agentic commerce? Because what used to be the simple SEO and SEM, the metadata and data on your website where the search engines will come and crawl your website and then you appear in the searches, organic search or paid search.
Now the agentic crawlers will come and you need much more machine-readable format of that data so that agents can understand. That's just the AEO, agentic understanding. We'll look at the architecture as to how you will prepare for that kind of integration. You will need the same kind of identity consent and policy frameworks as to how much you can allow the agents to operate on your website or through your APIs. Then multiple protocol interfaces are emerging, UCPs, ACPs. The platform completely evolves from a system of record, that's how it used to be, to a system of delegation. Because now those agents are working with your data, with your services, making decisions for the customer. It's a complete delegation.
That's the harness, how we are thinking. We are still going to use the platform because the core capabilities are still within our retail platform. If you look at the bottom left, customers, the bottom part is the existing path, existing online channels or stores from that perspective, existing application gateway, existing experience APIs going into our platform, retail platform. When you're talking about the agents on the top, you'll need now the gateways for the agents. How would you allow them? How much would you allow them? Which one would you allow them? The commerce protocol, ACP and UCP, now there are severe regulations in the UK for making payments, not allowed for agents to make payments on behalf of customers. Again, there are differences between the countries and regions to say what is allowed, what is not allowed. Many of these technologies, and I'm talking about both UCP, ACP, we are in direct conversation, are not ready for production scale yet.
As much as there is marketing, they are not ready. For Tesco scale, again you need very different readiness. We are not even talking about risk and compliance and anything here. I'm just talking pure technology risk here. Then you will have your own sub-agents that might be running, which might be talking to these agents and hence agent-to-agent protocol. You will need MCP for discovery of many of these services, discovery of data. Many of our application experience APIs were built for the applications that we built, like add item to basket, for example. You might need those plus some more for the agentic experience, but going back into the same retail platform. That's how we are looking into the future and thinking how we are going to do that.
What Could Have Been Better?
What could have been better is, build central tech platform early. As I said, we jumped in early into the public cloud and then we realized that everyone had their own version of CI/CD platforms and observability and everything. Now when I was talking to my own teams, say, we got best CI/CD pipelines, and every team was saying that. I don't need 50 CI/CD pipelines, I need one CI/CD pipeline. Portfolio management, we introduced that a bit late, but that where the products are matured when you're talking about 6,000 people in technology and hundreds of products, you need a very good management of where you are going to put your investment, where the value is. The portfolio management becomes very important. We use product maturity model for that now to say, our product is mature or not mature, and it can go left and right in the maturity curve.
Location strategy. Most companies, they assume that engineering can be even in-sourced where they are your own teams. They can be in various different locations, but if you're not careful, product management sits somewhere else, engineering sits somewhere else, program sits somewhere else. Try to create self-sufficient teams. I touched on three parts of those transformation, a little bit of each one of them to say what's important. There are many more in that one. The biggest point was the more you step back and think about it, you will avoid costly re-implementation and migration. That's huge. It spends a lot of time and spend of cost.
Tips for Any Large Enterprise Transformation
Final two slides on what are some of the tips if you are following the transformation at scale. Three pillars, we talked about that. Just system transformation is not enough. Change business processes when required, because they are just organically developed like your technology systems as well. Question that to say, is that really different? Why should four different countries have different operational processes? Simplify operation. Build core capabilities, we just talked about that. Because now we have the tech platform which supports this retail platform for those same CI/CDs, the observabilities, everything, but we could have started that early. Start that early. Throughput is a long game. Today also we get a lot of pressures, and can we go live in two months? Can we go live in two weeks? Can we go live in three weeks? We can do that if the platform capabilities are there. Total throughput is much higher today, but the same thing could have taken us six months long back when we had to develop those capabilities.
If we didn't do that, now we can do 10 such things and we would never have done that because we would have just duplicated it. Evolve continuously. We used to be very horizontally laid. I ran entire engineering of Tesco, all the platforms, all this thing, but product management was separate. Now that platforms are matured, we are going more vertical. I still run the platforms because that we cannot divide. The core platform is still there, but the application domain wise now we are creating more partnership between business and people like me within technology. Evolution is necessary. E-commerce is not just enough. Avoid tech for tech. Communicate value early. It's very hard to understand outside technology for sure. What are you all doing, is the frequent question we get. How come you are so many people? What's going on? It has gone less and less now. Even within technology, it's hard to explain what value we have created for the business.
We talked about the cost, IP and vendor lock-in. We are very careful when we are talking about agentic code development. We are very careful as to what kind of IP leakage could happen. Vendor lock-in, we talked about it. Even when we use many of the SaaS solutions, we ask for those protections from data sharing perspective, from IP perspective, so that we keep that edge of differentiation for Tesco. Then last one is technology will keep evolving, but if you have set the right culture, I mean there was blockchain there, public cloud before that, MapReduce systems, now agentic part. It will keep happening, and the pace has increased. If you don't have the right culture, the right talent, and the right level of trust within the company, it won't work.
See more presentations with transcripts