CUT OFF HAND-OFF

A work item (ticket, task, whatever) spends, on average, 15% 1 of the time in the hands of a performer. The rest of the time, it just sits around waiting for someone else to start it.

Well, maybe it's too tough to be true. In another source I found 40% 2.

A careful reader would ask me, "Where the hell did they get all these numbers?" and it's a very good question. Although many have written about these 15%, I couldn't find any decent research. All I got were opinions from personal experience, surveys of agile training students, or just arbitrary "industry standards".

But that doesn't mean I disagree with the general idea: tasks wait an uncomfortable amount of time before someone starts working on them. And the more links in the production chain, the worse the situation.

Let's CALCULATE

For example, one of my company's teams has 5 stages in the delivery chain: Business Analysis, Development, Code Review, Quality Assurance, Release.

Each stage has two columns:

• todo, to hold work items that are waiting to be started
• in progress, to hold work items that are being worked on.

Using Jira Flow Companion, I exported the time (in days) spent in each column and calculated the ratio of Waiting Time (sum of all todo columns) to Touch Time (sum of all in progress columns). It turned out to be 43%, which is surprisingly close to David Anderson's opinion 2.

I didn't take into account situations where a work item was in the in progress column, but no one was working on it due to blocking or other parallel work. So if I wanted to be more precise, the result would be even worse.

HOW TO DO IT ALL WRONG

Let's imagine a team. For example, the average time for a ticket to go from left to right on their Kanban board is 10 days - Lead Time. The average time in all in progress columns is 4 days - Touch time. The ratio is 4 / 10 = 40% - Flow Efficiency

For example, you have decided to increase work effectiveness of your teammates. Your intention is to do something to make them work faster.

LIST FULL OF INEFFECTIVE MEASURES

1. Employee's professional development

You can find a budget for some professional courses. Or start an internal book club. Or ask your seniors to mentor juniors.

2. Technological process improvements

You can simplify project architecture to get a code writing boost. Or start using more effective work tools. Or you can organize a technological environment that can safely handle fast experiments.

3. Specialization by work types or modules

If someone does a great job with some specific class of tasks or specific area in your code base, start giving those tasks to this person only.

4. Financial motivation

Give gifts to those who did great. Or implement KPI (in case you don't have a soul)

5. Classic management

Control, pressure, threatens to fire

6. Pleas for help

7. Agile training

AREN'T THESE GOOD ACTIONS? NO.

All of that is either too difficult, or too long, or too expensive, or immoral, or gives a short-term effect.

And let's face it, not many of us have had careers where 1 or 2 was the choice. Typically, items 4 and 5 would be chosen - for their quick effect. Carrot and stick. Classic.

Everyone forgets one thing: carrots and sticks motivate people to move faster, not think faster. No one gets smarter when the stakes get higher. The only exception is if you are Hugh Jackman in Swordfish.

Even if you've managed to increase the work speed of the performers by +100%, all you'll get is a 2-day reduction in Touch Time which gives you only a 20% reduction in Lead Time.

SO WHAT?

The next time you want to improve the efficiency of your production process, instead of thinking about how to do the work faster, think about where you are losing time.

Try to focus on the low-hanging fruit - reduce the amount of time tickets spend in the todo columns.

The theory and practical advice on how to accomplish all of this will be covered in the second part of this series.

References