My company, Veeva is a provider of cloud-based applications with a strong concentration of customers in the life sciences and healthcare sectors. The company has over 2,500 employees and its share price has increased by 500% since the company went public in 2013, in part due to impressive growth of free cash flow. Since starting with Veeva, I have overseen a major accounts receivable and collections systems overhaul at Veeva. This work built on the work of my predecessor, Glen Olson, who got us started down the path to automating tedious AR tasks. This process has included both putting in place new technology to automate and streamline collections but also new processes and a new cultural approach to collections. The results have been tremendous:
- 75% reduction in bad debt write offs over 12 months
- 50% reduction in 90-day aging accounts
- Zero growth in collections headcount
- Significantly improved accuracy in collections forecasting
- A far more pleasant and user-friendly collections process for the collections team
- A better customer journey through Veeva collections.
Here’s our story. A few years back, we did not have any collections automation. Invoices were sent out of NetSuite and there was some NetSuite automation on dunning processes but much of what we did was using spreadsheets. We initially started using automation for all invoices under $1500 - very small balances - by sending them automated dunning letters via email. The initial thought was that the automation would take care of the smaller balances that the team didn’t want to spend time on. We wanted to focus our most intense efforts on the bigger balances. The campaigns that were initially set up and were sending a lot of emails. But it was basically one template and one letter. We were getting some results but I thought we could improve on that significantly.
So I decided to start using automated dunning campaigns for all of our customers - not just small ones - and to make the campaigns a lot more customized. We added numerous dunning letter templates to our automation systems. This allowed us to deliver more specific messages for specific customer types. We customized by region, used different wording, and made other changes. We also put in place very different messages for 30 days past due versus 60 days, 90 days and 120 days. With this customization, we could target bigger balances much earlier in the delinquency cycle and maintain strong customer relationships. It didn’t feel nearly as much like cookie cutter work.
Another aspect of the automation process that was helpful was giving us more control over assigning different types of accounts to different collectors and building workspaces. Today, I am able to go in and assign specific customers to specific collectors. Instead of everyone eating off the same plate, every collector has their own plate. I can make this rule-based to reflect balance sizes, geography, or past interactions with a customer. This allows me to have different people on my team more easily specialize and for hand-off from one stage to another to be more automated and less manual. We can even tag customers in the system, just like you can tag someone in Salesforce.
Here’s how it works in a bit more detail. We have a group of customers that we call partners. We have a cross-functional team of 5 that handles these customers internally. Partners are generally smaller deal size and smaller revenue customers. They are handled differently in how they are brought on board, how they are introduced to products - there are a lot of nuances. We have created specific collections campaign workflows and routines for partners. In collections from really large customers we’ll allow accounts to go out pretty far - to 90 days or 120 days - before we move to a manual process.
For “partner” tagged customers, the timeline to collections is much shorter. We have a campaign we launch at 45 days and at 60 days we are sending out suspension letters. This has really improved the responsivity of the partner-class of customers; we’ve significantly reduced aging in that category. And we’ve done this while actually spending less people-hours on partner collections! I have one collections analyst who handles partner transactions. The analysts formerly were spending 25% of their time each week on “partner” collections efforts. Since we put in place the automated campaigns and the newer strategy of earlier notifications and actions, that analyst spends only 2 hours per week. And this is as we continue to grow. Six months ago, that analyst had to send out 5 emails per week. Now they have to send out 10 to 15 emails. It’s scaling up while actually requiring less time. With the automation, as well, the analyst can quickly pull up all accounts tagged as “partner” and assess what’s happening and report on them. In the past, this was manual spreadsheet work that might have taken hours to build out and update.
Another thing that changed for the better - the automated systems can often help with payment prediction dates. (Tesorio has this capability via its machine learning engine). The collections automation system will make prediction dates are based on past behavior. (Our analysts can also enter their own information on when customers are saying they will pay). This can be helpful in structuring campaigns and alerting us to when an account we expect to have paid hasn’t. It brings things we should know about to our attention very quickly.
Additionally, the payment predictions rolled up is a really good reality check for my own collections and cash forecasts. If I build my own forecast and it matches what I see forecasted by the system, that gives me more confidence. If there is a big gap, either above or below my forecast, then it’s a red flag that I should go look at what might be causing that. Granted, the automated systems are only as good as the data they build their models off. If invoice and terms data, of past payment dates are not accurate, then the automated systems can make errors. So you have to be mindful of that. But as a way to fact-check my work, it’s really useful. What’s also great about the automated systems is that its much closer to real-time. In the past, you built your models, made your forecast and waited 30 days or 60 days to see if you were right. Now, because the system aggregates so much data, you can see whether collections are trending towards or away from your models and forecasts. You can adjust earlier.
When it’s easier to assign accounts and easier to manage accounts, then we can give each player on the team an assignment that suits their capabilities. One of our most experienced and effective collectors handles all of our biggest accounts. Just one collector. Those comprise a significant percentage of our business. When we get more junior hires, it’s easier for me to break them in and have them try less critical accounts. To understand how everyone is doing, I can review things daily, weekly. I can make changes on the fly. It’s easy for me to make changes to campaigns, assignments, letter templates. If we see something we don’t like, we can change it immediately. And we can still track it and see the results in the differences of the campaigns and strategies.
For any automation technology to work, it has to be easy to use. Really, really easy to use. If it makes you work more than spreadsheets, then the analysts will abandon it. During the relocation of the AR & Collections team at Veeva, we have added a number of new hires. Most new hires have collections experience. For me, I give them a day or two and say “Here’s the log-in.” We let them go in and play around and click around, without any instruction or tutorial. If they have trouble, that means the system is not well designed. If you want people to use technology, you have to make it incredibly easy to use - easier than what they were doing before. This is a good test, in my mind. If they can’t figure it out on their own, in a day, then it’s not easy enough.
As a parting shot, for collections and accounts receivable managers and analysts working in fast growing companies, here are two pieces of advice. The first is not technology centric but it is crucial; You need to have a strategy that is scalable. It’s easy enough to say “I know I need one collector for every 1,000 customers.” That’s fine if you have 1,000 customers, but as you grow, that’s not sustainable. It’s not just saying collector A works these, collector B works these. In my last job before Veeva, we didn’t have a collections automation system. We had a series of spreadsheets and formulas. It was not sustainable and scalable. We had to throw people at it. Instead of being smarter, we worked harder. In a situation like Veeva, that would have never worked.
Second, once you have decided on an automation system, don’t hold back on it. We were using ours at 10-20% capacity when I started. That’s good but once you know your system works, let it do it’s thing. Now, we dump every invoice we have in there. We built campaigns and tried stuff out. Some worked, some didn’t. We tweaked them and kept trying new things. Sometimes assignments fell into buckets we didn’t expect. We fixed them as we went along. The worst thing you are going to do is send someone a collections later that is Past Due. That’s not the end of the world. So once you are set up and comfortable with a collections automation system, let it rip. Don’t hold back. Because that’s the only way to really leverage technology to change your business.