Backlog prioritisation is one of the hardest things for a product manager. As soon as the backlog grows just a little and part of it is literally out of sight, user stories start getting forgotten or out-of-date, duplicates are created, and prioritising with the awareness of the entire backlog becomes increasingly more complex, more so when several team members can add new items to the backlog. A backlog is essentially a simple list with no intelligence. So, we decided to make it better.

What it does

Firstly, Frontlog uses a custom Natural Language Parsing engine that we built to analyse all the user stories in a backlog and identify key elements in them, such as the (user) role as well as its job-to-be-done (goal).

Secondly, a product manager can see all roles and jobs-to-be-done in a single place and set relative priorities for them. These are translated into scores.

Thirdly, the above score, as well as other criteria, such as the explicit priority (trivial to blocker) or a custom scope-of-impact field for bugs, are all aggregated into a single Frontlog priority score.

Finally, this creates the possibility to prioritise the backlog with a single click. Yes, that's right, automatic prioritisation, that is clear and consistent! We even take issue dependencies into consideration!

How I built it

Well, it took a while, starting with the development of the NLP engine, and then learning by building the Atlassian Connect add-on. But we got there in the end :)

Challenges I ran into

I must admit that Atlassian Connect documentation could be clearer. First of all, help understand what are the possibilities and limitations with Atlassian Connect. Secondly, a clear separation from the Plugins2 documentation, which more than once confused us.

Accomplishments that I'm proud of

This is the first JIRA add-on we have built, which is a great accomplishment in itself. Also, building an NLP engine, then decreasing the lines of code by 25% while improving the algorithm was really nice... Developers will understand :)

What I learned

Analyse the possibilities of the host system before building something. We had to de-scope quite a few things due to limitations of Atlassian Connect.

What's next for Frontlog

See what people think about Frontlog, and try to develop it further. With Frontlog we are trying to create a new category of Smart backlogs, which we are very excited about, and hope to make lives of product managers much easier.

Share this project: