We’re always looking to improve the accuracy of our data.
New product features are all well and good but the core of our business is mobile app market data. We believe that, fundamentally, our customers come to us for data and remain with us because of our accuracy.
So far this year we have released four model updates. When we do this, we usually incorporate new partner data - exact data shared with us by app publishers - that makes our data models more accurate.
“The distribution of app store downloads and revenue across top chart ranks follows a predictive curve,” explains Michelle Tran, our Head of Data Science. “With the help of data from our data partners, we can determine what this curve looks like. As a result, we're able to get estimates at every rank position.”
The more data we pull in from data partners, the better our estimates get. So far we’ve had more than 1,400 new partners accounting for 23,000 apps and growing - we definitely want to include that volume of data!
That’s why, when we get significant quantities of data from new partners, we reformat our models.
But it’s not all plain sailing
There are positives and negatives (when aren’t there?) to pushing model updates whenever we can.
When the estimates are improved, they improve all estimates - not just estimates going forwards from the date that we push the model update. That means the historical data in the platform will change.
That can raise problems with some of our customers. Here is an excellent summation from one of them,
“What we struggle with the most is that you change the data in past and we are not able reproduce the same numbers and the same insights that we have reported to our clients previously.”
That’s a hassle.
It always bothers us to making customers’ days more difficult. So we’ve decided to re-examine our model update process. We had a look back at the updates we put out and decided there were a number of improvements to be made in the way we communicate them.
Simply pushing out the information isn’t enough - we want to provide context to our decision and give customers better tools to explain the new data to their clients.
How we’re improving the model update process
There are two difficulties here: one is about understanding the changes to the data and the other is about communicating model updates to your clients.
As an industry, mobile app intelligence providers struggle to communicate around the actual data that is the meat-and-potatoes of our business.
- Giving you more warning
From now on, we won’t push a model update without giving our customers fair warning. We will give one week warning before pushing any model update.
That gives you a chance to look back at historical data before it changes, making it easier to compare the data before and after the change.
- Adding graphics
Internally, we use fully informed monthly report to understand quickly and effectively how good our data is by category and country. We decided to share it with you.
This will let you inspect where our data is strong and where it’s weak. We’re not interested in hiding our weak points - we’re interested in helping customers make data-driven decisions.
Another option is to come up with a way of providing historical data based on previous models for certain users. We’re exploring this internally - we’d also love to hear from you.
Help us improve our data, join our Partner Program
Join the Priori Data Partner Program and get up to 50% discount off the base plan of your choice.