Learning from a Blogging Experiment Failure

We’re about ten weeks into running an experiment in blog post writing. The purpose is to be less deadline-driven, allow time for quality checks and make the writing process less of a weekly struggle. It was going well until this week. There were zero articles in WIP. And I was all out of ideas. This wasn’t supposed to happen.

Lemonade from lemons and rabbit from hat, I decided to write about this circumstance. Blogging about blogging is a bit meta. It’s not something I want to make a habit of. But there are things we can learn from any negative results of experiments.

The Experiment in Brief

The idea of the experiment was to have a pool of finished articles ready for proofreading and publication each week. The old process pattern was to think of a topic and write an article during each week. For one reason or another, this would slip to the weekend, leaving no time for a colleague to proofread it. Building up an initial buffer was supposed to give me some weeks to develop more ideas into articles, rather than one per week. Creative work or processes that require iteration don’t always lend themselves to one-piece flow.

We set an upper limit of five items in WIP. It’s unclear whether that is appropriate. The initial minimum inventory level was two articles ready for proofreading and publishing. If there were two or fewer, that was a sign that I needed to prepare at lease one more draft in the coming week. These are experimental assumptions. Neither number is based on scientific data collection. Experience is proving they may be wrong.

The Recent Process Failure

Things were going well until about a month ago when I saw we were down to two articles, the lower limit. I found a topic, developed it offline and added a third item to digital WIP. However, what I failed to see was that one of the two articles was an early draft of this same “third” article. I noticed this only the following week. Down to one item in WIP, I had ten days to find an idea, turn it into an article and get it proofread. That sounds reasonable, but my calendar was a bit full. Soon, I was back to where I started, brainstorming for Monday’s post over Saturday breakfast.

The immediate causes of this process failure were a combination of deviation from standard work and poor visual controls. When is that not the case? Writing an article offline introduced human error. What looked like three completed articles were in fact two, and a half-finished duplicate. There was nothing in their titles to indicate that they were different versions of the same article.

The Countermeasures

Near-term, I need to mistake-proof the act of counting how many articles are in WIP through better labeling. That’s not difficult. I will also need to quickly rebuild the buffer of articles. That’s harder, but necessary to contain the problem. Resetting the buffer and make sure I don’t repeat the same mistake is only part of the remedy. The inability to refill the WIP on a reliable basis remains an issue.

To address root causes, I need to better understand the process. The original process pattern was to, on a weekly cycle, 1) find idea,  2) write a blog, 3) schedule it for following Monday morning, and 4) have it proofread. Step 1 was the bottleneck. Step 4 didn’t always happen.

The experimental process pattern was 0) a one-time build-up of initial stock of 5 finished articles, then on an non-specific multi-week cycle to 1) find idea, 2) write article, 3) add to WIP. Then on a weekly basis, the proofreader 4) picks one and edits it, and 5) schedules it for the following Monday morning.

Step 1 needs to be broken down and studied further. I maintain a list of potential topics that are interesting, but not fully developed. The current process is to look at the list to see if one of them prompts me to write. Between “find idea” and “write” there is an activity that involves looking through these ideas, adding to, refining and discarding ideas. This needs to be more of a process rather than inspiration-seeking treasure hunt.

Drawing General Lessons from this Failed Experiment

Sometimes it’s OK to run out of WIP. This exposes process problems. Perhaps it’s like fasting for the body, healthy when done once in a while. We need to be careful that running down WIP won’t cause interruption in supply of the product or service isn’t life-threatening or inconveniences to customers. Blog posts aren’t either of those things.

It’s tempting to increase the WIP level. That’s human nature, I suppose. Luckily (?) for me this was not as easy as buying more stock to put on the shelf. If it was easy, these process problems would remain hidden, felt but not seen. It’s possible that the minimum of 3 is too low. First, I’ll address the issues of visual control and standard work.

Coming up with ideas requires an unspecified amount of time and some serendipity. This is the grey zone that continues to defy quantification. That’s the nature of creative work. But there are definite activities and inputs, such as reading or talking to customers, that yield ideas faster than staring at a web browser. Even for creatives, it’s possible to define process parameters.

The new process ran pretty well for a couple of months. I was quite content with it. Perhaps too much so. Fortunately, failed experiments give us the opportunity to reexamine our assumptions and strive for a deeper understanding of the process.

---