One of our powerful clients, Daniel from Teach Me 2 — a tutoring company trusted by 1000's of parents and learners – has been building sales analytics for quite a long time. We asked him to tell us about his insights and failures along the way and share the sales analytics dashboard which he created as a result of that work.
Two years ago, we didn’t have a sales team and the only analytics we tracked were in our front-end website. As the business grew, it became increasingly difficult to evaluate employees and processes. Our initial solution was to request detailed CSVs from our developer, import them into a spreadsheet, and try to find patterns. Adjusting the CSV meant expensive developer time, rebuilding the spreadsheet, and always meant you were looking at old data. We needed a better solution.
As soon as we started using Statsbot, we realized we had struck gold. We started being able to analyze and steer the business in real time. We could make iterative improvements to the metrics themselves without any developer time. Now, all our data is available to teams instead of being hidden in the database.
I started building the metrics we would use to measure and manage performance. It’s been a highly iterative process and has led to us fundamentally changing how we incentivize our sales. In this post, I’m going to share my insights on sales analytics and explain which metrics are needed for a sales analytics dashboard.
Usually, you want data as quickly as possible so that you can make a business decision. For example, you discovered that a salesperson is not working well enough and you need to help him manage this transaction with customers paid out, or that this person was doing really well and you’d better put more business towards him to maximize that.
All the fun is in building new things, not checking things. But if you want to push things forward, you need to spend a lot more time validating.
Meanwhile, you want to hit the sweet spot between business impact and data reliability. It takes a lot more work to get simple and understandable metrics, which would allow us to take the right actions.
Obviously, you want to be aligning your sales people with the profit. Our top performing sales manager was my mom (she was the oldest too!), her numbers were irrefutably good. But, she wasn't necessarily earning the most – we had holes in our commission structure, the most important sales metric.
By going through some of the data and looking at it from different angles we realized that volume of leads is only one piece of the puzzle. The other key ingredient is the quality, which we get by how many leads we plant, and then how much money we generate from how many we plant.
So, I’ve rebuilt the commission structure with the right allocations and joins in Statsbot, and I can now rely on that information. Having key insight into, and being able to run different models on what’s happened throughout the year, who’s done what, I can now have a much better commission structure.
The three key ingredients there are: how many people have you dealt with, what have you generated from them, and over what period of time.
Previously we were just looking at how much you have generated, not how many customers did it take you to generate that. And that’s the key. That’s worth a lot for us. I’m happy we managed to find that out and that Statsbot could help us figure it out.
Building commission structure also helps us in hiring because now we have a better understanding of who generates the best results for the business.
One of the key metrics of our new sales analytics dashboard is productive hours. People can have very productive hours in the day and then have no productive hours for two hours at the end of the day, doing little work.
Calling for us is a key aspect in engaging with customers. For example, you’d make forty calls in a day and I’d make twenty calls in a day, but, I would do two hours of calling at ten calls per hour, and you would do eight hours of five calls. You were working consistently through the day at a very high degree of productivity, whereas I was cramming my work into a short period of time and then taking it easy.
You would have eight or seven productive hours in a day and I would have only two. That’s one thing we’ve shifted in how we measure and evaluate the calling.
Now, we not only just measure calls, we also measure hours in the day when people were most productive and who was working the hardest.
If we could identify people who were doing little work, we could take corrective action. We also reorganized our days around the most and least productive hours – now we avoid creating meetings during productive time.
Number of inquiries
On our sales analytics dashboard, we have a certain number of inquiries being generated each day. The number of inquiries is needed to make sure that the priority is right and also give people an indication of how quickly they need to work.
It’s important to remember that there will always be a backlog. For example, you finish the day at 5 PM, your backlog is 0, and you come in in the morning and your backlog is maybe fifty, maybe twenty, or maybe about three thousand leads, depending on your business. You don’t want your backlog to be too big, but you also don’t want there to be nothing in the backlog.
Quality of inquiries
If I put you first in the queue and ten people behind you, well, you’re probably going to be more valuable because I’m speaking to you first. The person at the end of the queue is gonna be going, “Well, I’m gonna go somewhere else.” But, he might have been a really good customer, so it’s one of those tensions to try and work out.
We have a rating system for the quality of inquiries that come in, and then we measure how many of the top-quality inquiries we have called within thirty or ninety minutes.
The quality is indicated by how urgent the inquiry is, how long of a commitment they want, and whether they used us before.
It’s no secret that the biggest indicator of value is whether the customer has used you before. Those people go right to the top.
You can just look at your metrics and think, “Okay, looking at all these factors, which have the biggest bearing on the amount that they spent at the business?” Often, you can’t make such decisions in real time, you need to wait a while and then you can do some analysis on that over a couple of months.
Imagine, you and I claim a hundred inquiries each, but I’ll hand over twenty and you’ll hand over ten. But, of my ten, all ten paid. And of your twenty, only five paid. It means that my handovers were of a much better quality than yours. So that would be another metric which I call a to-payment rate, or handover rate.
A tricky thing to work out for your business is an appropriate time frame of how much time leads need to make a purchase. Obviously, three months down the line is enough to get the idea, but when you’re looking at your data immediately, you’re thinking, “It’s been three days, is that long enough or not?”
So, how quickly does your data mature? That’s one of the toughest things that I’ve found in working out some of the time frames.
The best way I’ve seen is to take several portions of your data and then hit multiple time points in days or in months. Then, you can analyze the variance functions to see how soon you get a strong enough signal that you can rely on.
Tip: Rates are a better indicator than your number. For example, looking at the handovers as a number we see, ah, we are up 50% on handovers so we must be doing well. However, if you look at your handover rate you could see a drop in what you should’ve been handing over.
I can’t help wondering that you always need to double check your data and the way you align things. Otherwise you can make devastatingly bad decisions, since the data isn't supporting you.
We were measuring the customer creation date on the point of handover. What was happening was that most of our customers were being created on Monday through Friday, and weren't being created on the weekend. We were looking at it going: all the customers who come in a weekend are worth nothing. We can see how many inquiries are coming in, but if no customers are being created who inquired on that day, we should just turn off all weekend marketing.
This is one of the trickiest things with data. Spidey-sense sort of intuition told us that something was fishy. And then, sure enough, when we investigated, we saw that customer creation dates were put in with the wrong timestamp. When we switched the customer creation date to when they inquired, we found out that customers that come in on a weekend are valuable.
As we get more of sales metrics in place, it becomes easier to fold them into our process. Analyzing if it was a 5% improvement, it probably wasn't worthwhile. If it's a 10% improvement, it definitely worthwhile. And if it's a 20% improvement, that's excellent for three months’ worth of work.
So, the changes which would have a significant impact on the sales process itself are coming down. Saying, "Where in the data will I see the improvement from this three-month project in technology, and if I don't see it, why not?" Those are conversations we never had before because we were never able to validate. Now we are.
If you are looking for ways to make your data available across all departments, give us a chance to help you and try Statsbot for free.