What is our primary use case?
We're a service provider, so we work with multiple clients to build out Anaplan solutions across a lot of different disciplines. We have a couple of high-tech companies. We're using it to manage the PR process, the purchase request process for marketing, which includes campaign management, and also for how a purchase request gets funded. It's that modeling around scenarios where you want to really understand what costs are associated with activities.
Shortly, we're going to implement around connected with revenues so we can understand in more detail the ROI on marketing. This is for a high-tech Fortune 100 company.
What you do is you buy a workspace and then you basically have access to that workspace, and then you can build models within that workspace.
How has it helped my organization?
The solution improves forecast accuracy. We've seen forecast accuracy go from a variation of 10% to something like 2%. It also drives up the quality of analytics. What we're seeing is not only forecast accuracy improving but also the time spent on pulling data together to do a forecast. You get the benefits of both worlds and you get the forecast accuracy as people are spending significantly less time.
It goes from about 50% on data validation, and getting the data, and validating the data, down to about 20% on a lot of our clients. You're freeing up those finance people that are pulling those models together. You're freeing up 30% of their time during the planning cycle to really spend a lot more on analytics, and deep-diving into the numbers, and doing a better job at understanding what those numbers are owning up to instead of spending all that time just consolidating all the numbers into one space.
That's definitely a big advantage. We are seeing massive time savings for people, freeing up time so that employees can do more value-added work.
Especially over the COVID period, a lot of people have left and they haven't hired new people to replace them. The company is really looking at standardizing processes across the globe. This solution helps to build a platform where you can build a standard process across the world, and you get benefits from that idea of standardization where anybody can do anybody else's job as it's exactly the same all over the world. There is that advantage as well.
What is most valuable?
In corporate performance management where you're looking at building out forecasts, modeling data, and have really heavy interaction with data, it's extremely useful due to the fact that, unlike using Excel to build out models and things like that, it's very collaborative.
While you can put Excel in the cloud, and try to collaborate that way, it doesn't necessarily work as well. Whereas, this works really well as a collaboration solution. You can build complex data models. You can have lots of people entering lots of different pieces of data into them, similar to what you have in Excel, and yet, it's in the cloud and it's designed for the cloud, and it's a bit more structured than Excel.
It takes out a lot of the risks you have in building models in Excel. On the other side of the spectrum, you have your consolidation tools. You've got your Hyperion, you've got your adapters. Their primary purpose is to consolidate large amounts of data. They're typically quite purpose-built with little ability to customize them and their core function is consolidation. You can also add driver-based solutions on top of that. It's very rudimentary, and it's very designed around the application. These solutions typically are for corporate performance management. If you're a smaller company, and you just need a basic consolidation solution, that's where you go.
Where we're seeing the opportunity in the market is at the Fortune 100 companies that have far more complex modeling type scenarios where you need to be able to have a lot of people collaborating together, and you need a lot of information all at everyone's fingertips. This is where you see the big advantage of a solution like Anaplan.
What needs improvement?
Anaplan is a relatively expensive piece of software. It's definitely being applied to very complex problems, and if the price were to drop I expect it would be more broadly adopted.
From a product point of view, they have launched some new reporting functionality, which is pretty basic compared to something like Power BI.
What they have done to compensate for the reporting was to build native APIs into something like Power BI and Tableau, so that you can integrate your data into a reporting suite. The need to continue to develop this. There's new functionality all the time, however, some of the core functionality was lacking about three, four years ago. It's continuously getting resolved and improving, and I'm now pleased with the level of functionality, however, they need to keep going in this direction.
For how long have I used the solution?
I've used it for about the last five years.
What do I think about the stability of the solution?
I've been super impressed with the performance. I use other technologies for multi-billion dollar, trillion-dollar companies, and they can sometimes be a little bit challenging, especially if you haven't got a great PC. Due to the fact that this is all in-memory computing, the stability has been pretty good.
The challenge to that is building the model. For example, we just built a model for a client recently, and we created a holiday calendar, which had a relatively complex calculation in it. That was actually slowing the entire model down. Every time there was a change, we just had to hard code the numbers as opposed to having it calculating all the time. That said, the actual stability is pretty solid.
Still, when you're building the model, you can get performance issues, which may be caused by actual model design, as opposed to the software itself. My advice is to make sure to spend the money on a good architect. If you can get a good architect, you're fine with the builders as most people pick it up quickly.
What do I think about the scalability of the solution?
It's phenomenally good at scaling. We built a solution for a retail chain, which basically initiated transfers from a centralized hub down to stores. There were 800 stores and about 600 skews in each store. It was all machine learning that drove the purchasing. And we were able to, on a daily basis, send an automated transfer to the central hub. There were six hubs around the world, and it would automatically on a daily basis send anything between 100 and 250,000 skews to different stores based on predictive algorithms.
Typically, it's very unusual to deal with a small business. Typically Fortune 100 companies are using this product. There are a couple of different models, however, just the cost of being able to build a model, for a smaller company, that amount it would cost doesn't make any sense.
Also, typically, smaller companies don't have as many complex problems to solve. You can have a standardized model, whether it's a forecasting model or a headcount model, full cost and headcount, or a sales model and things like that. For smaller companies, Excel could handle it.
We do have plans to increase usage in the future.
Typically the way it works is that you normally start off with a use case or a number of use cases, and then the client gets used to using it. They build up an internal team as well, and then they expand the use cases out, and that continues to kind of build-out. There are just hundreds of modeling opportunities where you can bring in two sets of data, where you can bring in lots of people putting information in and, you can bring in predictive analytics. So there's always a huge amount of opportunity. The nice thing about Anaplan, as well, is it's very connected. I can use that one use case and I can use data out of that use case in another model, and all the models are connected. If somebody changes something in one model, then it'll change in the other models as well.
How are customer service and support?
The tech support is pretty good. They're very responsive, and that typically means they're pretty good at solving problems. They can tend to take a little bit of time. Sometimes the challenges that we have are, typically, quite complex.
When we had a performance issue on one of our models, it took them at least three weeks to do a full review of the model as the models are quite complex. However, they did a detailed breakdown line by line and there were probably 10,000 lines of items in here. It's built as cubes, so we have line items inside the cubes, similar to a pivot table in Excel. They did a full analysis, and then we got a detailed report at the end of where we had performance issues on a line-by-line basis and we could easily fix those issues.
Which solution did I use previously and why did I switch?
Previously, if we were building models and things like that, we would typically build them in SQL, and, before then, Microsoft Access, and before then, Excel. Some of my clients still use Excel just to keep it kind of simple, however, the reason why we switched to it is that it's really easy to set up the user interface, so you can build something that used to take us six months in SQL with a web UI. Now it takes us six weeks to build something. It's just the speed to deployment which is significantly faster. This is due to the UI which is very well designed, so you can build out that UI very, very quickly. Then the model in the background is also extremely powerful as it's all in memory. The barrier for most companies is the cost to switch. A lot of our clients would stay with Excel models until they got to a certain point, or a company got to a certain size, and then they would move over to an Anaplan.
How was the initial setup?
My background is not technical. My background is in finance. Therefore, picking up a solution is relatively quick if you're familiar with building models in Excel, or some of the other technologies out there. However, for a non-technical user, it is relatively easy to pick up. A lot of our clients don't necessarily use their IT department to support Anaplan, however, it does require that the people internally are trained. My advice is also to get a very good architect.
A lot of the projects we do are actually fixing other people's models. For example, people have built out a particular model that hasn't been positionally well designed. What we end up doing is going in and redesigning the model just to optimize it. What we need from clients is to really get them to focus on ongoing support. Getting them trained up is a key part of the deployment. Making sure that they are driving the functionality, et cetera, however, where we support them a lot is more in that initial design phase to make sure that we're building the right model for them.
The deployments are typically pretty small. For a small deployment, probably two or three people are needed. A larger one might have up to ten people.
For deployment, what we normally do is we do it in phases. A typical phase will be about three months, and a lot of our projects last a couple of years. The solution is very agile, so typically when we deploy, we have something up and running in the first month so that we can start to set it against people's expectations and understand some of the challenges. A typical project lasts three months. It might be to build a particular model, and then we'll go on and then we build another model or enhance the first one and things like that. For a typical deployment, end-to-end, is about three months.
The product probably doesn't require maintenance. The reality is that if you're doing business modeling, business modeling is constantly going to change. It doesn't necessarily need maintenance, and yet, typically, you don't want to build something that's static for very long, so you're constantly updating it. Where we see companies doing well is to have them invest in the development upfront, and then invest in what they call a center of excellence that is an internal team of people, that's centralized that can help architects and build future enhancements. Then, you have Anaplan experts within the business that can update and build constant enhancements. If you think about the modeling capability, this is not something that is like a one-off kind of development. It's typically an ongoing thing.
What was our ROI?
We had COVID in between, so the dates are not a hundred percent accurate, however, a client was looking at a significant cost return just being able to get the right level of inventory in every store across the world meant that, there was a 5% impact on the bottom line. I haven't got the final numbers yet. They are still working them out, however, from what we were seeing from sales, there was a 5% increase at least in sales. That was all translating to the bottom line as well.
Some client ROI's are a little bit more difficult to quantify. It's ROI in saving people time and being more accurate. One of our clients went from 10% to 2% accuracy of the full cost, which meant that they were able to make far better decisions. They knew exactly how much things were going to cost, for example. This is an IT organization within a Fortune 100 company, a tech company. It made their lives a lot easier when they were going through this massive uncertainty. They were able to really understand exactly what their underlying costs were.
What's my experience with pricing, setup cost, and licensing?
The entry-level is anywhere from about $30,000 to $50,000 a year, however, it does go up significantly after that depending on the complexity and how much space you're using. It's really driven by the number of users and also the space. It's more the user base, however, sometimes you need to buy extra users just to get the space.
What other advice do I have?
Currently, we are working on a partnership agreement with Anaplan.
It's one of the areas that I probably haven't explored enough at the moment, however, they have added integration into machine learning as well. The idea of that is you can build out a model and then you can push it through a machine learning model and then you can get an answer out of that. That's an area that I haven't probably spent enough time exploring, however, we're definitely working towards that.
If you've only got one person that's going to be building, then you probably want to go with something else, however, if you've got a complex modeling problem, and you've got a lot of people that are interacting with it, your typical kind of forecast and where you want to collect data from loads of different people, and you're just finding that Excel isn't doing it for you, this solution is worth considering.
I'd rate the solution a nine out of ten. If there'd be any reason why not a ten, it is just it is that much more expensive. It's a lot more difficult to get more clients onto it. You need to be a certain size to get it, to use it, however, from a stability point of view, from delivering value, et cetera, it's well worth it.
Which deployment model are you using for this solution?
Private Cloud
*Disclosure: I am a real user, and this review is based on my own experience and opinions.
