Why do things happen the way they do?

### What is the difference between trend and compare views?

Trend view lets you see your lines moving in relation to each other (each has its own Y axis). Compare view should be used to see the difference in the line values (a single Y axis).

### I just configured a service. Why hasn’t the data shown up yet?

In some cases, it can take up to 2 hours for new data to show up.

### Why do I see more emails than expected?

There is a setting which allows you to choose between counting the number of emails or the number of conversations. If you are seeing more emails than you expect, switch this setting to count the number of conversations.

### How does your forecasting algorithm work?

Our prediction algorithm is a simple mathematical linear weighted approximation of the function by the least squares method.

We use the last 13 full weeks to feed into the prediction algorithm.

1. Most recent 2 weeks have a weight of 6.
2. The previous 2 weeks have a weight of 5.
3. The previous 4 weeks have a weight of 4.
4. The remaining weeks in the past have a weight of 3.

After this weighting, we build a system of 2 equations and use a Gaussian method in which the system of equations reduces to triangular form. The solution yields the linear approximation.

### You don’t integrate with a service I use. Will you?

See this blog post: Requesting New Data Extractors In DigMyData

### How is churn rate calculated for the Stripe and Spreedly integrations?

We took inspiration for our churn rate calculation from Steven Noble (@snoble) at Shopify here. The churn rate algorithm we are using is the following: 30 * (number of customers who churned on day x / number of customers at the beginning of day x). Note that we do not include lifetime or complimentary subscriptions in the churn rate calculation (they cannot churn). We do include accounts with complimentary time (assumption is that the complimentary time feature is used to offer a longer trial or sales incentive and the account will eventually carry churn risk).

### How is churn rate calculated for Authorize.net?

We calculate the Authorize.net churn rate with the same basic formula as we do for Stripe and Spreedly – but – because the data is not as complete we have to make some assumptions.

How we build subscription data in Authorize.net:
- We build subscriptions/cancellations using Transactions.
- We look at the combination of the CustomerID, Name, and Email fields to build a unique customer.
- If we find multiple transactions with this profile then we have a subscription.
- Calculate the average time period between neighboring transactions
- We identify cancellations when there are no transactions after last transaction + (ATP x 1.5)
- We now have the list of subscriptions/cancellations for our churn rate formula.

### Why don’t the MailChimp.com subscribers numbers match in DigMyData?

The MailChimp subscriber number that you see is the number of confirmed subscribers (double opt-in). Double opt-in is the industry best practice and the MailChimp API makes it impossible for us to show a full history that include single opt-in subscribers. For more information, see MailChimp’s support article on Double Opt-in.