Bad data rarely announces itself. It doesn’t crash a system or trigger an alarm, it just quietly sits inside contracts, supplier records, and spend reports until it costs an organization real money, real compliance exposure, and real credibility with stakeholders. Procurement leaders often sense something is off, but few can point to exactly where the “data nightmare” begins, or how to fix it without it feeling like a never-ending clean-up job.
That’s the conversation host Deepak had with Karthik Rama on a recent episode of the Beyond Procurement podcast. Known across the industry as “The Procurement Doctor,” Karthik has spent two decades diagnosing procurement dysfunction across people, process, technology, and data for global organizations. In this episode, he broke down why bad data behaves like a silent illness, why supplier master data is the “heartbeat” of any procurement function, and why most AI initiatives in procurement fail before they even start.
Below are the questions Deepak asked and Karthik’s answers, exactly as he spoke them, because some of the clearest procurement advice comes when it isn’t polished.
What triggered the name “Procurement Doctor”?
Well, if you look at it, many years ago I was in the consulting business trying to set up a practice for an organization I used to work at. As part of setting up that practice, there was a report called the Procurement Diagnostic Report, that we built together. And I had this lovely client in the Middle East who was walking through the entire report. Unconsciously, subconsciously, the report had stethoscope and human-body analogies related to their procurement function.
As I was explaining to this lovely lady how the entire thing works, how the blood pressure of a human connects back to bad data, or whatever the case may be, towards the end of it she called me and said, “So you’re like the procurement doctor, and I’m your patient, right?” And that led to another. She kept telling people, “let me refer you to the procurement doctor,” and it kind of caught on.
At first I couldn’t get used to it, but after some time it became a norm in the industry. People kept calling me by that name. In India it was “Dr. Sahab,” or by my friends in the US, “you’ve got to talk to the doctor.” I kept getting these emails for a while, and then I eventually turned it into a brand, and into a company, because one of my customers said, “I’m not going to do freelance work with you any longer, you should set up a company.” So the brand turned into a company about four years ago.
And my mom always wanted me to be a doctor, and I really didn’t want to spend so much time on studies, so this way at least I get to be a doctor, a procurement doctor, for her, so she’s happy as well. I intend to become a real doctor too, so I’m trying to do a PhD. Let’s see, that’s going to take a while.
Procurement leaders often underestimate the impact of poor data. What are some of the damaging procurement failures you’ve seen due to poor data?
Oh, there are a lot of horror stories I’ve heard in my practice over the years. There’s a lot of crazy stuff, but I’ll talk to you about a few things that strike really hard in my mind.
Bad data is like white jaundice, if I might say, a condition where you don’t even know you’re sick until it really gets bad. Most procurement leaders think of it that way itself: let it happen, then we’ll realize. They want to see the symptoms, or want to see something go wrong, before they act on it. It’s pretty typically like a silent killer, a virus or infection in your organization that you’ve not realized until it really happens.
“Bad data is like white jaundice, you don’t even know you’re sick until it really gets bad.”
It could be as simple as people not realizing that you have “Vendor A” named as “A Vendor,” and it creates chaos in matching the PO to the invoice. It’s almost like, in medical terms, you’re prescribing the wrong medicine to the wrong patient, and you can imagine how big of a damage that could be. It affects the bottom line whether you like it or not. Our entire sourcing and procurement process is driven around understanding the spend, and if that basic thing is not right, your entire base is not accurate, you’re missing on compliances, penalties, and all this stuff. If data is not right, then you will not be set out to become a world-class organization.
Would it be apt to say that data has an impact across not just naming, but entire processes, compliance, reporting discrepancies to stakeholders? Can you share an experience where a major goof-up happened because of something this simple?
One common thing, we feel helping a customer with their data is one of the easiest and most impactful things, because the customer doesn’t realize there’s so much hidden opportunity.
Imagine you have one key critical supplier, one of your big guys in a particular category, and they’re categorized the wrong way, meaning the supplier name is entered in three different ways, and you’re missing out on 10% of the spend. They’re already a big spend because they’re one of your key suppliers, and you’re not managing 10% of it, and your entire payment terms, discounts, and compliances are going as per the other name variations. You could look at that and see the amount of risk you might be undergoing which you did not mitigate for, because of that simple alignment issue of having all the names together.
Once you fix that, they’re like, “Wow, I didn’t know this, I could have saved so much in the last couple of years.” And they’re actually very excited, but they’re not excited enough to do the work or ensure it’s done the right way the first time. That’s one classic example that I’ve seen.
What are the top three challenges that stop procurement leaders from fixing bad data, even when they understand its importance?
In global organizations, there are three challenges I typically see. First, inconsistent supplier master data, because supplier master data is the main thing for us. If you look at IBM, for instance, everybody writes it differently: “IBM,” “IBM Gulf,” “IBM EMEA.” You have so many different names, and it’s not standardized, and it shows up directly in your spend visibility.
Second, fragmented systems. You might have an ERP in one corner, the sourcing system in another, and then spreadsheets lurking all around managing your data. There are multiple ways to do the same stuff. It’s almost like, if I have to diagnose a patient, I’ll ask for the most recent report, a blood report, but I won’t take a blood report from last year and say, “Okay, you’re sick now, this is what you need to do.” Basically, that’s what the data is doing. We have old data, inaccurate data, or duplicate data, and we’re expecting it to be real-time and give visibility to our leadership.
Third, and this is the bigger point, you cannot look at data in isolation. That’s where we have the Procurement Doctor’s diagnostic assessment: we look at people, process, technology, and data. If the first three aren’t right, data is going to be messed up for sure. People aren’t trained, they don’t have the knowledge of how to populate the data, it’s all a mess. Or the system doesn’t have the mandatory fields or drop-downs to capture the data properly, and everything ends up being manually entered, there’s no point having the system at that stage.
And data itself, at the end of it, needs to be monitored. It cannot be that you leave it there assuming it’ll stay clean. If any one of these misses, it’ll show up in the data, but it’s more so a postmortem, unfortunately. If you don’t do a premortem, a regular check or self-diagnostic assessment, you’ll not be able to stop the bleeding. You’ll keep going, and the infection keeps spreading. It’s a never-ending problem.
Now that you’ve summarized these key challenges, how would you recommend procurement leaders start fixing this in a way that’s doable and not overwhelming?
It’s like a triage. What do you do first when you find an emergency case? You first attend to the wound, you stop the bleeding. Similarly, we’ll have to first look at the existing data, look at how you can cleanse it, gather feedback, it’s kind of a miniature Six Sigma project, if I may say. Understand the root cause, whether it’s people, process, data, or technology, and you have to plug the gap there, or else it will happen again.
And it’s not just about calling in an independent data person who comes, points out the mistakes, fixes them, and leaves. Someone in your own team needs to sit and redo the process to fix the root cause. So first you attend to the wound, stop the bleeding, then cleanse the data, whether it’s de-duplication or removing duplicates between two systems.
Then you apply external data intelligence to it: market intelligence, D&B, financial risk scores. It comes across as intelligent data that you can make sense of, and you get proper outputs, close to 20% additional value out of the entire exercise, as such. The language should be the same, uniform across different systems and processes. And you should ensure this becomes part of your everyday process, an exercise you do on a recurring basis, not something you did once during an ERP implementation and never revisited.
Most big organizations have realized this. They have a team in-house, sometimes called a metrics-and-reporting team or a market research team, and it’s better for them to own this so the data stays robust and usable in real time.
You need to keep it maintained rather than treat it as a one-time exercise. Is there anything else procurement leaders overlook here?
That’s right. And you need to keep in mind, whenever you’re acquiring or selling off part of your business, if you’re acquiring, especially for the big pharma companies who do a lot of this, data becomes a critical element for you to map. You cannot just, in haste, do the merger and add the system. There’s a chance of adding more messy data to your existing portfolio. You need to take some time, have an expert look at it, and do it the right way, or you won’t get satisfaction running a bigger organization.
I’ll tell you a little more about how I’ve seen problems in the past. Your contracts get merged. An acquisition or sell-off is easy in one sense because the pain goes to the other side. But if you’re acquiring, you’ll need to create an entity in your ERP. They’re running their shop in a particular way, do you use their way of running things, or do you make them work in your ERP? Is the data getting replicated correctly? There are a lot of things when you connect two systems, they call something the same thing very differently in their company versus yours, so there has to be a common way of doing it.
People just go with it, “let’s wing it,” and after three years they say, “why is this not working?” Maybe it’s not a bad acquisition at all, you’re just running it the wrong way, without the right people to do it. So you need to give the right amount of anesthesia. You cannot overdo it, or it’s dead. You underdo it, then there’s pain. The integration has to be done seamlessly so it makes sense and you can use it in the future.
Everybody wants to implement AI, but without clean data these systems fail. What are the biggest mistakes you’ve seen companies make when rushing into AI without fixing data quality first?
You see, in the last 20 minutes, whatever we’ve spoken, that’s all basic stuff itself, not advanced things at all. Unfortunately, we’re sitting here talking for 45 minutes still about the basics, because there’s not enough light shed on it, or people don’t care enough. They’re more so running after the next shiny object, Gen AI or whatever, without enough background. They’re too impatient, they don’t go deep enough.
You’ve got to go deep to learn, and you need to grow breadth-wise as well. If you go too deep and just keep going deep, you’ll be just a thin trunk of a tree that falls off anytime. That’s what data is, this was told to me by a leader long ago, and I’m just reusing the analogy.
If you want a good Gen AI or AI adoption within your organization, I think you should have at least three things under your radar, let’s call it the Procurement Doctor checklist. One is spend being under-managed. That should be a key metric, because the more spend visibility you have, the more use cases for AI, the more automation. The second, which is spoken about very little, is contract compliance, or contract leakage, tying the contract all the way to the invoice. That itself is a humongous task for a lot of leaders right now. I know many CPOs where, if you ask “which contract does this invoice tie to,” people go around like headless chickens. Our processes and systems aren’t built for that yet, but they’re at least aiming for it.
Third, Supplier performance, because if the supplier doesn’t do well, you’re gone as an organization. Your top suppliers, or at least your major, critical ones, should be reviewed on a regular basis. Give them feedback, learn from them. And it’s not just delivery time, quality, and quantity, data your ERP has captured forever. It should be more about sustainability, compliance, and how you drive innovation and strengthen the relationship, more so partnership.
If you have these three things on your radar, spend management, contract compliance, and holistic supplier performance, you will do better, and these will evolve into something else in the future. It’s going to be like your proper health report: you doing your steps, eating the right diet, getting good sleep, having good mental health. It’s somewhat like that, and it’ll do well for you as an overall procurement organization.
Can you share an instance where a company or leader maintained data beautifully and reaped the advantage, whether that’s savings, visibility, or a more resilient operation?
I can’t give you the name of the organization, but I’ll give you a couple of examples. One is an organization based in the US, one of the biggest giants, and they’ve done this for a long time, even before AI was branded as AI. They didn’t call it AI, AI was there for a long time, it’s just been coined and marketed the right way now. A lot of old features are being rebranded as AI today.
These guys had a state-of-the-art contract database tying back to all their data sources, invoices, POs, everything, and they had partnered with an amazing negotiation chatbot organization. They reaped a lot of benefits by just feeding it the contract data, because you as a category manager might not know all 12 of your contracts by heart, all the line items, the ones spending more. But a chatbot knows, it has the memory of a computer.
Most suppliers would sit around thinking, “oh, this is a chatbot, I can fool it.” They went into the negotiation thinking they were winning against the chatbot, whereas the chatbot was bringing 25 to 30% savings very easily by looking at all the contracts at once, giving the most favorable scenario for the organization, while making the supplier feel he was winning.
There are many procurement organization running smoothly, without much conflict between finance, legal, procurement, and other stakeholders, is often that way because of good data. We don’t realize that a complicated process or inaccurate data gives people a reason to point fingers at each other. If it’s all clear and you’re having data-driven discussions, showing data, then trust builds, and the relationship becomes much smoother, even though we rarely credit the data for it.
A lot of times your stakeholder thinks, “oh my god, procurement is always trying to reduce the price.” But when you show the use case with data, along with your domain knowledge, “no, I’m not going to blindly reduce the price, this is what my idea is,” it gets sold much easier. Just talking in the air won’t work.
Can you share an example of how good data changed your credibility with a CFO?
I’ve done reporting in the past where procurement started reporting into the CFO, and the CFO always thought cost avoidance is not real savings, and questioned how procurement was impacting the bottom line and supporting the overall executive management. We found there was a communication gap, so we started building a report to him in balance-sheet language, showing how procurement was making an impact.
Then the hair on his shoulder stood up, and he started paying attention. We weren’t talking about categories in spend anymore, we were talking in general-ledger language, cost of goods sold, the kind of language he didn’t have to learn our spend portfolio to understand. That made an impact, and he started investing. He let us buy some cool procurement technology, and invest more in specific people who could help drive transformation within procurement.
“It’s hard, or almost impossible, to argue with data. It allows you to ask for the right resources, the right people, the right tech.”
If I were a procurement leader looking at my own systems, are there any metrics or parameters I could use to gauge my data accuracy?
It ties back to what we spoke about initially: supplier master data is one key thing. Unfortunately, procurement is not like finance. Finance follows debit-credit, balance-sheet rules that are standard anywhere in the world, whether it’s India, the US, or Mexico. In procurement, people run their own kingdom in their own way, so I can’t give you one standard across the globe. I can give you a descriptive definition, but at the end of the day it depends on how complex your process is, or how customized your system is.
I’ve seen organizations, even in India, where the average age of the team is around 45. They’ve traditionally used offline work for a long time, and now all of a sudden you’re pushing them onto digital systems. It’s a very different way to manage that transformation depending on the region and the workforce.
That variability is exactly what keeps me going in procurement. No two days are the same. Every customer is different, every industry is different, every region is different. That’s why, once I got into procurement, I never wanted to leave, with other work, after a year or two you feel like you’ve done it all. Here, you never have. That’s the reason I took the Procurement Doctor’s approach: diagnosing, asking the right questions, learning, and applying that repeated methodology, because there’s no single “received fix” that works across every organization.
Is there a practical starting point, like a weekly data check?
Yes, or you could set up a vendor management team within procurement. A lot of big organizations have one; smaller ones don’t, and they simply expect procurement to manage vendor data without giving them the tools, rules, or controls to do it properly. You should have a subsection of your policy defining how a vendor gets created, what triggers happen when a new vendor is added or an existing one is changed. Without that structure, mistakes keep happening, even with good intentions.
For small and medium organizations, procurement is usually focused on invoices and purchase orders, less so on the contract or vendor lifecycle side, because they just want to get the PO done and move on. The mindset has to expand, think like the CEO who’s planning to grow to 100 people by the end of the year, and needs their operations lead to think ahead of that scale too. It’s important for leadership to actively engage with procurement, because unfortunately, in a lot of companies, the CEO himself ends up doing vendor negotiations directly, bypassing procurement entirely, which is exactly why procurement still struggles to earn a seat at the strategic table.
Many procurement teams struggle with real-time data integration across ERPs, supplier portals, and e-procurement solutions. What’s your go-to strategy for integrating procurement data without major disruptions? Is there a master system you’d recommend?
Firstly, you have to attend to the wound, clean the data first, before deciding how to integrate anything. From there, pull your spend data and invoice data, analyze the supplier data, and tie it back to your contracts. Categorize your vendors into critical, major, and minor, and understand the criticality of spend if one of those suppliers underperforms.
It’s always preferable to have a master system, and slave systems against it. Ideally, all supplier data is created first in the master system, and it gets replicated to the child systems from there. And if you have a front-end system, whether it’s Coupa or Ariba, where data actually gets entered, the flow should be one-directional, from that front-end system into your ERP as the source of truth. You don’t want people making edits across every system, master, slave, and front-end alike, because that multiplies complexity, breaks traceability, and makes the whole integration heavier than it needs to be.
Beyond the technical integration, what mindset should leaders bring to a data cleanup exercise like this?
You need a broad, open mindset. I’d call it the classic Elon Musk way, he trims as much as he can. There’s this example, I forget exactly which feature it was, in one of his Tesla or Cybertruck models. After two years of data showing a particular feature on the passenger seat was barely used, once or twice in a year and a half, he simply took it out of production and saved millions of dollars. He didn’t call a board meeting to debate it. No one even realized it was gone. That’s the power of data.
You need that same mindset when you have multiple teams or categories, and everybody thinks, “I’m special, my category is uniquely complicated, so I need to do things my own way.” Everybody thinks they’re special, but often they’re just collecting data that’s never actually used, creating a data dump with no real insight. You need to re-look at the entire thing and ask whether this data really makes sense to collect at all, because even if it only takes two or three seconds to enter, if you’re never going to use it, why collect it?
Even in supplier onboarding, we ask suppliers a million things without cross-verifying any of it. We ask, “are you a certified diversity supplier,” they say yes, but do we ever check if that’s actually true? It becomes a tick-mark activity, and we’re just dumping data without ever really populating or using it meaningfully. You have to think about how the data will actually be used, not just how it gets collected.
Out of people, mindset, and strategy, what actually drives lasting change in data quality?
It’s a culture, honestly. If good data practice is part of your culture, it happens effortlessly. If it isn’t, then your culture is to create messy data, it’s just how you run your shop. It’s not a one-time thing where you fix it once and it’s gone. It’s more like exercising, staying fit. It’s not like when you feel sick you go to the doctor and get an injection. It’s an ongoing thing, a daily routine that matters, like going to the gym, but we don’t look at data that way, unfortunately.
If a procurement leader could fix just one data problem today, what would have the biggest impact?
If I had my stethoscope on, and I’m putting it on a procurement patient right now, the first thing I’d treat is supplier master data. It’s the heartbeat of the organization. If your heart isn’t well, the entire blood supply, the entire system, becomes outdated. Your contracts are like the nervous system of your organization, but your heartbeat is your supplier master data. We don’t pay enough attention to it, unfortunately. When it’s right, it gives you more negotiation power, more compliance, more visibility into your spend and your contracts.
“Your contracts are like the nervous system of your organization. But your heartbeat is your supplier master data.”
Your suppliers should basically be your friends. During COVID, people who didn’t care about their suppliers found that their suppliers didn’t prioritize them either. The ones who had genuine personal relationships got answered first. Good master data isn’t just a record with a number and a name. You need the details behind it, so when you talk to a supplier, you actually know them, what value they’re driving, what they’re delivering, how you could grow together.
I’ve seen situations during a supplier performance review where a supplier openly shared an issue he was facing, and instead of just noting it, we worked through it together, a real co-innovation. He was so happy with the outcome that he offered a price reduction and said he wanted to take the same approach to other customers. That kind of relationship can’t happen overnight by handing someone an award or a thank-you note. It happens by genuinely valuing the relationship and being authentic about it. And master data is the very first step of that entire chain.
Before we wrap up, could you introduce what you do, where you’re based, and the best way for people to reach you?
LinkedIn is my go-to way to connect, whether people DM me or otherwise. I’m always active there, my wife isn’t thrilled about how much time I spend on it, but that’s where you’ll find me regularly. I run an organization called Procurement Doctors. We’re close to 40 people based across India, the US, and the Middle East, and we’re here to write prescriptions to cure procurement’s woes. The more unknown or unusual the problem, the better for us, because we learn from it too.
We work across people, process, data, and technology, and we also partner with other specialists depending on what a client needs, like a pharmacy section in our medicine cabinet. My own team can handle a good level of data cleansing, but if something needs to move faster or deeper, I bring in a trusted partner. My entire team averages about eight years of experience, and most of them come from a procurement background who later picked up data or technology skills, some of them people I worked with 20 years ago who are now back on the team.
Sharing this kind of knowledge, through podcasts, keynote speeches, whatever form it takes, is genuinely satisfying to me. It feels like a form of giving back to the procurement community that’s been my life for a long time.
Conclusion
Karthik’s diagnosis is a familiar one for any procurement leader who has quietly suspected their data was messier than their dashboards let on: bad data rarely announces itself, it just erodes savings, compliance, and stakeholder trust in the background until someone finally checks under the hood. His prescription is consistent throughout this conversation, clean supplier master data first, build in continuous maintenance rather than one-time fixes, and don’t chase AI or automation on top of a foundation that isn’t ready for it yet.
That last point matters more than ever as procurement teams rush toward AI-led transformation. Clean, consistent, real-time supplier and spend data isn’t a “nice to have” before AI adoption, it’s the prerequisite. Organizations that treat data hygiene as an ongoing discipline, not a one-time cleanup project, are the ones who actually see the 20%+ value Karthik describes, rather than an AI pilot that quietly fails six months in.
