Ever longed for a tool that could give you feedback on the health of your email inbox? Listen in as I take on the challenge of creating one from scratch. Here in LiveLab 2 I invite Dr. Michael Einstein to help me build a brand new kind of tool – one that would provide users with an automatic diagnosis of their inbox health. In this episode (the first of three), we work on the underlying assumptions that such a tool must include…and discover along the way that there are *many*. Plus we start to look at the elements we think should be measured in preparation for the next episode which focuses on the actual formulae which the tool would use. It’s a far-reaching, one of a kind conversation which starts off in wobbly fashion as we’re not sure whether we will be successful. Here’s the link to the original calculator I created which we used as our starting point. Can’t wait until the final episode to see the end result?
This is the second part of LiveLab 2.
In this discussion, we delve into the formula that would make up a working email calculator.
Months before I called Michael, I devised the following formula in my very first attempt.
5.0 max(0, [2 x (total messages-unread messages-tagged messages) [
0.5 x (total messages – unread messages – incomplete tasks recalled)
5.0 x (unread messages – incoming email per day)]/10)
This is where our discussion started. It moved quickly and resulted in this formula which we produced in the first part of the call.
1. (total messages-unread messages-tagged messages) [more importance] “BlindSpot measure”
2. # read messages * (age in days -1) [more importance] “Morgue Measure”
3. # unread messages/ # incoming email per day [more importance] “Backlog Measure”
4. # threaded messages / # total messages [moderate importance]
5. # raw threaded messages [high importance]
6. # subscription email older than a day / # total messages older than a day [low importance]
We took a break for about a month, then reconvened to record the second part of this episode.
After another discussion, we agreed on the following.
1. (total messages-unread messages-tagged messages) [more important]
2a. # read non-subscribed messages * (age in days per message -1) [more important]
2b. # read subscription messages * (age in days per message -1) [more important]
3. # incoming email per day / ((#unread subs–sqrd> + #unread nonsubsc) [more important]
5. # raw threaded messages – for non-subscribes using geometric growth [high importance]
6. # subscription email older than a day / # total messages older than a day [lower importance]
Our next and final conversation didn’t take place until 10 months later.
Live Lab 3
In this instance we started by listing a priority of concerns. They are listed here from 1-5 in rank order.
1. How many days of stored email are accumulated? (read vs unread, subscribed vs non-subscribeds)
2. How old are these message? (read vs. unread)
3. How unique are these messages? (subscribed vs non-subscribeds)
4. How fast are they entering? (incoming email)
5. How complicated are they by being threaded?
During the hiatus since the last episode, I drafted a weight for each measure and after playing with the tool we would be using, producing the following formula which we discussed in this episode.
0.25 x Left Behind Index (i.e. (Total messages in your inbox – unread messages-tagged, read messages)/incoming email each day)
0.20 x Number of Days Surprise Index (i.e. unread messages – unread subscriptions email)/incoming email each day)
0.20 x Total messages older than a day)/incoming mail each day
0.20 x (Average age of non-Subscription messages/days)
0.20 x ( .50 x Average age of Subscription messages/days)
0.10 x Incoming Email each day / messages removed per day
0.20 x Threaded messages
The final input into the calculoid app used the following weights which were simply scaled so that they would sum to 1.0:
Field 1 – 18% Left Behind Index [i.e. (Total messages in your inbox – unread messages-tagged, read messages)/incoming email each day)]
Field 2 – 14% Number of Days Surprise Index [i.e. unread messages – unread subscriptions email)/incoming email each day]
Field 3 – 14% Total messages older than a day)/incoming mail each day
Field 4 – 14% Average age of non-Subscription messages/days
Field 5 – 7% Average age of Subscription messages/days
Field 6 – 11% Max(1, incoming email/150)
Field 7 – 7% Incoming Email each day / messages removed per day
Field 8 – 14% Threaded messages x #average active participants in each thread