Coupa is a company of talkers, passionate about sharing tips, tricks and advice for improving finance and procurement and saving companies of all sizes time and money. But we’re not the only people with opinions and ideas. We’d love to hear from you so join the conversation!
- September 20, 2017
- Rob Bernshteyn
In the C-O-U-P-A alphabet, “P” is for prescriptive, as in prescriptive recommendations. Over the long term, this is the biggest opportunity for Coupa to differentiate ourselves from our competitors, and for our customers to do the same with theirs.
As the world goes digital, many of us find ourselves awash in data. While we have all heard about the promise of big data, the truth is that it doesn’t matter how much data you have if you don’t know what it means, and what to do with it. That’s where prescriptive recommendations come in.
A case of the flu
When I think of the word "prescriptive," I think of a prescription for medicine, written by a doctor. In fact, that’s what first got me thinking about the opportunity for software to be prescriptive.
In the year 2000, I was living in Boston, attending Harvard Business School, and I got sick. I went to see a doctor just outside campus, in Harvard Square. Naturally, the doctors here see a large volume of students. Many of my classmates were sick, and sure enough, the doctor diagnosed me with the same flu they had, and handed me the same prescription he had probably written two dozen times that day for a three-day antibiotic, scrawled on a piece of paper in his sloppy penmanship.
This was around the time that the internet was starting to change the world by making it easy to access information using personal computers. I had already decided that I wanted to devote my career to solving big business problems with technology, and in my business school courses, we were studying different types of technology companies, and what they were doing with this incredible new capability.
Local vs. global prescriptions
With these ideas percolating in my mind, it occurred to me that knowing exactly what to prescribe was based on a special combination of what was in the mind of that particular doctor—his education and experience, his knowledge of flu bugs in general, and his real-time awareness of the Harvard community, and whatever bug might have been going around there. I must have been the umpteenth person from the community complaining of the exact same symptoms; it probably wasn’t too difficult for him to figure out what to prescribe me.
But what if I had gotten sick while traveling, and turned up at a hospital in downtown Los Angeles? In LA, it’s warm and dry at that time of year, the doctors are dealing with a different population altogether, and there’s no flu going around. Would it have been just as easy for the doctor to know what to prescribe?
Extrapolating from what was going on with the internet, it seemed like there would be an opportunity in the future for medical diagnoses and prescriptions to be based on the real-time collective intelligence of medical professionals around the world, and for the prescription to be specifically tailored to me, no matter who I was or where I lived, based on my genetic makeup and medical history.
Fast-forward to 2017, and that’s exactly where we’re heading. Medical records are quickly becoming digitized, creating a huge health database. In addition, services such as 23andMe that are collecting health and genetic data from millions of people are leading the charge toward individualized medicine. There are roughly seven and a half billion people alive on our planet today, and an estimated 20-25,000 human protein coding genes. Given the amount of computing power and technology we have available today, we should certainly be able to combine data from a community of seven and a half billion people with our know-how in medical science to eventually figure out the best ways to heal all human beings, one at a time.
Looking for an opportunity
You might say that my flu was the germ of an idea about what might be done with the power of the collective in any number of areas. Working in enterprise software for the last sixteen years, I've been looking for opportunities where it made sense to harness this power to create value.
It certainly wasn't easy with on premise software, because everyone's data store was separate, just as every visit to your local doctor was separate. Sharing of observations, knowledge, and best practices was not digitally enabled.
Later when I worked in the HR realm at SuccessFactors, I thought something like that might be possible. Because we were in the cloud, we had data on the skill sets of millions of people through our software. We knew their career paths, and the typical career paths of someone with their skills and experience. Based on that information, we could prescribe to any individual what they would be best suited to do next, and show them relevant job openings. Unfortunately, other company priorities always got in the way, and the truth is that we didn’t have enough normalized data, nor the technology that we have available now, which made solving this problem much more difficult.
By the time I got to Coupa, there was no question in my mind that we would, at some point, apply the power of community intelligence to advance our mission to help companies use information technology to spend smarter. In 2009, we began setting the foundation for this future by introducing our first set of patent pending cross-company benchmarks as part of our offering.
Our first prescription
Looking across the entire Coupa community, which, at that point, comprised of a few thousand users, we calculated the average time it took to approve a purchase request. We communicated this information to customers through the platform, put it in the context of their company, and offered our first prescription: “The average time to approve a purchase request across all Coupa customers is 72 hours. In your company, it's 15 days. Consider removing some potentially unnecessary approvers from your approval workflows to speed things up.”
This was our first iteration, and many customers saw the power and potential of it. But there were concerns: Was it a big enough dataset? How valid are the results when mixing large companies and small companies, or companies in different industries? While the concept was clear, it was clear also that the quality of the dataset and the individualized applicability was not yet there. So, we continued to offer these capabilities as part of our platform, for free, for dozens of different metrics, as we continued to build out our dataset.
A couple of years ago, when we got to hundreds of billions of dollars in spending through the Coupa platform, with the corresponding massive set of data, we began to take the concept further by slicing our data by industry, by time frame, by company size, to provide a greater level of fidelity for individual customers. And then we started offering them new, more personalized prescriptions.
This time, it felt different from the start. We saw customers get real value from our new prescriptions. The ability to compare themselves to best-in-class companies in their industry, or of similar size, and the resulting ability to target specific areas for improvement, has helped them become more operationally efficient. In fact, some customers have told us that they compete with each other to see who could be the first to have all their KPIs show “green” in their Coupa dashboards at best-in-class.
This leads to better outcomes, and a spirit of continuous improvement, not just locally with certain customers, but globally, across the entire Coupa customer base. Whether they are competing directly with one another, or with benchmarks, we know they are likely to be running things better than companies who are not Coupa customers, and that is something we all started to take pride in.
We decided to not only continue offering this capability, which we call Perfect Fit Insights, but to push the community intelligence concept even further—prescriptive recommendations not just for process improvements, but for supplier selection and pricing.
With more than a trillion dollars of spend data analyzed, and over three million suppliers in our community, we can see now start to see which suppliers are doing the best job, which are not, and who the alternative suppliers are for each category. We also know what buyers want, based on their industry, company size, and their buying history in terms of quality, price, volume, and a host of other factors. And, we can bring the power of community intelligence to bear so that we can make individualized recommendations for our customers to consider.
For example, we have a product in early access that can tell customers things such as, “We suggest you reconsider working with a given supplier because they have over-invoiced a large portion of the customer community during the last year, shipped X percentage of broken goods that had to be returned, and have had X disputes with different buyers.” Of course, our buyers don’t have to accept this recommendation, or they can simply consider it as part of a broader set of variables they are tracking, but in either case we are helping them get smarter based on the power of the Coupa customer community.
At the same time, we are working on creating visibility for suppliers, so they understand how the community views them, so they can become the best suppliers in their industry, and begin to understand what it takes to be the leader in their market. This ultimately should contribute to an unprecedented level of openness, across all industries, on a global scale.
We are also working on capabilities to offer deeper transparency, so that buyers know where they can find the best pricing, service level agreements, on-time deliveries, and so much more. As we continue to grow and collect more data, the prescriptions will become better and better, offering our customers the opportunity to become radically more efficient and competitive.
A trusted recommendation
Now, back to my doctor in Harvard Square. After I left his office, I could have chosen to not take my prescription from him, but I did, because I trusted him. I trusted him, first, because I knew he had trained for nearly a decade so that he had a high likelihood of being able to solve the problems that I had. Second, I was a Harvard student, and he was affiliated with Harvard, and this connection provided another layer of trust. Finally, I knew he had seen some of my classmates that had the same flu and given them medicine that helped them get better.
Similarly, why would our customers trust our prescriptions? First, we’ve been in business for more than a decade, staying true to our mission of helping organizations around the world manage their spend more intelligently. Second, we are not simply a solution provider to the global spend management community, we are a part of this community ourselves. We partner with our customers, working closely together to solve problems.
Finally, we have solved similar problems for hundreds of organizations around the world of all sizes, from pre-revenue startups to globally distributed Fortune 100 conglomerates. Those are our credentials, and we’re committed to a common cause with our customers. The prescriptions we provide today will continue to become more personalized and valuable as our data set grows.
The natural next question that our customers ask us is, will other companies see my data? Of course not. Just like my doctor didn’t talk to me about anyone else’s case individually, but still used the insights he gained from solving all their cases to help me, the insight behind each Coupa recommendation is anonymized and aggregated from a massive community.
And lastly, they want to know, can they trust our data? Up until now, spend data has been notoriously hard to get a handle on. For example, let’s say you’re buying things from GE Healthcare, and you just record the transaction as GE. Then someone else buys something from GE Capital, and they also write GE. But, those are completely different spending categories.
You also have to understand which departments, divisions, and locations you are working with. Products also have to be normalized. What SKU are we talking about? This is why we bought Spend360: to apply machine learning and deep learning techniques to our data to normalize it, and to ensure that accurate, normalized data is the foundation for all of our prescriptions.
Our customers can further explore prescriptions because you're able to drill down, investigate, and verify them for themselves, much like many people today use the internet to learn more about prescriptions their doctors give them. For example, if we’re advising you to avoid a certain supplier, it’s not just us saying they're bad. You can double-click and see the number and type of disputes that the community has had with that supplier. The data is in front of you, allowing you to investigate further.
A prescription for success
Just as elsewhere in the world, our data set is growing exponentially as we add new customers and our existing customers grow their businesses and their use of Coupa. In fact, more spend has been processed through Coupa in the last 12 months than in the previous eight years.
We’re growing the database and we’re developing the technology. We’ve piqued the intellectual curiosity of our customer community. We're using the data generated by the community for the benefit of the community itself; which they are opting to be part of.
Together, we’re planning to take this concept to a host of areas within our platform. Of course, just as any individual is the final decision maker regarding matters of their own health, each company could ultimately decide what’s the best medicine for their company. We’re working on helping them make the very best decisions. That's the vision that we're pursuing in this area, and relatively speaking, we’re just getting started.
- Value optimization is the ultimate promise of SaaS
January 23, 2017 | by Ravi Thakur | IT & Technology
SaaS is not just changing the way software is delivered. It’s changing the way software vendors and customers work together. What’s changed the game is data. Vendors, with...
- 5 benefits that will help suppliers learn to love e-sourcing
March 09, 2017 | by Andy Chiang | Procurement
For buyers, the benefits of e-sourcing are clear, but suppliers aren’t always fans of reverse auctions, which are a core component of e-sourcing. However, when done right,...
- Advanced ideas for using new expense tracking technology to cut travel costs
January 19, 2017 | by Ethan Laub | Expenses
In my last post, I wrote about three relatively obvious ideas for trimming unnecessary travel spending that are easy to implement with new expense management technology. If...