5 months in: Lessons learned on agile, data, and writing 🗣️
July 2024 progress vs. my STELLAR roadmap goals for writing here on Substack: recap on post sizes & readability, new lessons learned, actions, and plan for August (audio: 19:20)
I committed in March to doing (and sharing) a retrospective each month on how I’m doing vs. my “STELLAR” goals. Here’s what I’ve learned in my 5th full month on Substack, and what I’m going to do differently in August.
Thanks for reading Agile Analytics and Beyond! Subscribe for FREE to receive new posts (retrospectives are published monthly).
Some acronyms I use:
6P = my 6 ‘P’s in AI Pods newsletter
AAaB = my Agile Analytics and Beyond newsletter
RF = Readability Formulas tool
WIP = Work In Progress
How It’s Going
In retrospective #4, these were the goals I set for July:
15-20 Posts in a mix of lengths (still aiming for 50% short/50% long overall), and
24-32 Notes plus Comments
My effort budget for July posts stayed at 6-8 hours/day, or 30-40 hours/week.
Two changes are worth noting:
I merged my agile Teams newsletter into my personal newsletter sections (Agile Analytics and Beyond, i.e. this newsletter) and I sunsetted agile Teams.
I’ve stopped logging my LinkedIn posts in my data file and measuring them. At this point, I don’t see that effort as helping me improve as a writer.
The data in this retro post reflects these changes, including data shown for previous months. To be clear, these post targets and my actuals don’t include:
posts written primarily by my collaborators, or
my forewords on any posts I cross-post from other Substacks.
Aside from that, how’d I do in July? Here’s a look at each element of the STELLAR framework (other than ST, the starting position, which isn’t changing).
E - Progress towards End goal
Quantity Goals
I exceeded the high ranges of my goals for July by writing and publishing:
Posts in (now 2) Substack publications
Goal: 15-20 posts. Achieved: 24 posts.
Goal: 50% short, 50% long. Achieved: 58% short, 42% long (14 VeryShort or Short, 10 Medium or Long)
Goal: 24-32. Achieved: 43. (Most were brief; I didn’t measure sizes). Includes restacks with comments.
Count does not include comments or restacks without comments.
Archive links are below in the References 1.
As of the month 4 retro, I had written and logged 61 posts and 74,849 words (that’s as counted by MS Word and RF). During July, I added 24 more posts and 22,482 words (that’s as counted by RF, and does not include footnotes).
Averages for July, per workday: ~1 post, ~1000 words, and ~2 Notes per day.
Totals since March: 85 posts, 97,331 words. (That’s a whole book!)
Post Sizes and Distribution:
The average length for my 24 July posts is 933 words, lower than 1242 words in June. I’m ok with this ballpark.
58% (14 of 24) July posts were Very Short or Short (1-799 words). That’s a bit on the high side of my original March goal of 50% short and 50% long. It’s a swing from June, which was just over 40% short after excluding LinkedIn post data.
Figure 1 shows my post size distributions through July. The ‘VeryShort’ bars look different from the graphs in previous retros because I’ve removed all of the LinkedIn posts from the data for all months.
Quality Goals
During July, I continued to work on improving my writing. I estimate that I stayed at quality level 6 in July.
My measurement goals center on more readable writing and offering a good distribution of post sizes. Post size distributions were covered earlier under Quantity Goals.
For readability, I’m aiming for grade levels 9.0 or below (lower is better).
For action (A10), I scored all 24 of my July posts with both Readability Formulas (RF) and Grammarly free. The Flesch-Kincaid Grade Level (FKGL) values are consistent within +/- 1 grade level between the two tools, and Grammarly is imprecise anyway. So from now on, I’m only going to report on RF values, not Grammarly.
Here’s what the data is showing me on reading ease and grade levels. All of these numbers exclude LinkedIn posts.
Readability Scores per Publication
Table 1 below compares the data for my 24 July posts to my 17 June posts for each of my two publications. Some key points:
My 13 6P posts in July averaged:
18% longer (608 words to 720 words), and
slightly less readable (2.4% worse grade level, 0.9% better reading ease score). Average grade level for July was 9.6, still above my target of <=9.
My 11 AAaB posts in July averaged:
29% shorter (1673 words to 1193 words), and
slightly more readable (2.5% better grade level, 4.3% better reading ease score). Average grade level for July is 8.2, under my target of <=9.
These changes are all small. The 6P posts now include company profiles, which tend to be longer and more technical. So I’m okay with these values for now.
Other Activities
I did quite a bit of writing-related work in July that will help me get to my goals but is not reflected in the data above.
Mentoring collaborators: Although I again finished and published some partially-written 6P profiles, I still have significant work in progress (WIP). A few collaborators are now contributing on some of the WIP 6P articles and helping me finish them, and starting on new articles and topics. Although mentoring and coordinating has initially taken more of my time, I expect this investment to pay off for my subscribers in August and beyond.
Combining newsletters: Analyzing my data made it clear that my agile Teams newsletter subscribers were a complete subset of my personal newsletter. To simplify life for my subscribers and myself, I merged my agile Teams newsletter content into sections of my personal newsletter, then I did a soft shutdown on agile Teams. So now I have 2, not 3, newsletters.
Audio voiceovers: I set up external podcasts on both newsletters and recorded audio voiceovers on 24 old posts (more on this under Actions).
New feature: I initiated a new 6P interview series on “AI, Software, and Wetware”. This included: defining draft interview questions, contacting prospective guests to be featured, arranging initial calls, collaborating on content, and (for some) recording and processing audio. Although this yielded zero new content for July, it’s giving me a great start on August. The first audio interview (with
!) was published on Aug. 1, and the second (anonymous, text only) on Aug. 5. Several more interviews are in final review and post-production, and even more are scheduled for the next few weeks.
LL - Lessons and Limits
Which RF metrics to rely on (or not)
This month’s retrospective showed me a key lesson on the RF metrics I’m using. Table 1 shows data from RF’s “Average Reading Level Consensus Calc” (an average of 7 different grade level metrics) along with the standard Flesch-Kincaid Grade Level (FKGL). As this data shows, RF’s ‘consensus’ average tends to run about 1.3 grade levels higher than FKGL on my writing. For my target audiences and the range I’m in, lower is better (more readable). And I’m not under my target yet. I’d rather under-estimate than over-estimate the readability of my work. So I’m using the RF Consensus calc for evaluating my posts from now on.
Related to this: The data in Table 1 says my FRE (Flesch Reading Ease) scores improved slightly for both publications. But both FKGL and the consensus grade level show that 6P readability got a little worse. That inconsistency just proves that different readability metrics value and measure different things. It’s why I didn’t want to rely just on Flesch metrics for improving my writing 😏.
Planning and executing interview posts
Another lesson is that the new interviews I set up in July, and launched on Aug. 1, are going to be quite different from my usual posts. I’m not writing most of the words, and I’m not going to count words in those posts in my quality metrics. But coordinating and delivering those posts makes quite a bit of work for me, especially if the interviews use audio. I need to plan my interview effort differently from written posts.
Planning and executing audio voiceovers
I also need to plan the effort for audio voiceovers on my own writing. I am mostly getting through the manual recording in one take with minor edits. But it does take time, and there’s often a delay in me being able to make a good-quality recording. I’ve also found that posts with a lot of data or tables are hard to ‘speak’ effectively. Time-sensitive or data-heavy posts will likely not get audio, at least at first. Some may get a short audio summary and not a full voiceover.
A - Actions
Here are updates on my previous self-assigned actions.
April-June Lesson 1: Connecting better with readers - Voiceovers
(A9) Change my publication process when adding voiceovers: Hold off on sending posts to my subscriber list until I’ve added the audio. (resumed)
During July, I did 3 things:
Set up podcast sync with Spotify, Apple, YouTube, Overcast, and Pocket Casts for Agile Analytics and Beyond and 6 ‘P’s in AI Pods.
Recorded and released audio voiceovers for 14 previous Substack posts and 10 imported pre-2024 posts.
Held 9 new July posts until the audio voiceover was done, and then published them.
Unfortunately, I don’t yet have great visibility into whether people are listening to the audio voiceover tracks, whether through Substack or through most of the external podcasts. If you’re finding the audio useful, or have suggestions for improving the voiceovers, please let me know!
June-July Lesson: Becoming a better writer - content planning
I had self-assigned an action in my July retrospective on planning my content.
(A12) Revisit my publication plans from May and realign. For July, be more intentional about thinking through my target audience and what I want their takeaways to be, before I dive into writing more content.
I spent some time analyzing and thinking about what questions people are asking, what they’re interested in, and where there are gaps that I can fill. This drove 3 changes to my 6P content plan for August.
The main 6P content plan still remains mostly focused on ethics of generative AI for music. However, with the help of a collaborator, I’m going to accelerate the WIP company profiles (maybe with less depth) to move on to the other PARTs of the music series, which I think will interest more people. (Each PART is already more than half-written.)
My new 6P interview series “AI, Software, and Wetware” arose directly from this action and exercise. People want to know what’s real in AI. I now have an interview planning workbook where I’m tracking planned interviews of real people, and working to ensure I’m getting a diverse set of global views on AI.
Based on questions from a follower about AI, I’m working on an article for the new AI Fundamentals section, on “What Is AI?”. This topic is also part of chapter 1 of one of my WIP AI books, so tackling it in an article also helps me get that track of writing work underway.
R - Review (Retrospective)
I committed to do a monthly review at the end of July to see where I stand on my goals and actions - status: this article is it 😊
Bottom line:
I made good progress on my Actions.
I spent lots of July time on newsletter admin and adding voiceover and podcast capability to try to improve accessibility of my articles.
I feel like I’ve made good headway on my content planning action. I’m hoping that my Substack data in the next few months will give me some insight into whether these topic changes are the right moves for my subscribers.
I reduced my WIP further by finishing and posting some draft content I’d already written. Thanks partly to using my WIP, and partly to a higher level of effort than planned, I exceeded my posting targets for July.
I’ll reassess my targets and plan in my next monthly retro.
New Targets for August 2024 (month 6)
My July goals felt “aggressive but achievable”. I overdid it a bit, though, getting the new interview series, podcasts, and collaborations off the ground. And we have some planned family vacation time this month. So I’m going to lower my August goals somewhat vs. July, but keep the Notes goal the same.
12-20 Posts in a mix of lengths (still aiming for 50/50 overall) - this will include:
2 AAaB / agile retrospective posts (this article, ~2615 words, and 1 supporting article on lessons learned from podcast setups, ~2490 words).
5-9 6P / People interviews (5 if audio-focused, 9 if all text).
Reality check: My AISW interview plan for August indicates I will have at least 3 audio interviews, likely 5, plus at least 2 text-only. Of those, 2 audio and 2 text-only interviews will already be finished and scheduled for delivery by the time this post comes out on Aug. 7. I am hearing so much interest, and having so much fun, and I want to get the interview series off to a good start this month; so it’s going to be tough to not over-achieve here ;).
As mentioned earlier, I’m not going to count or measure readability on the written words on these interviews. Most of the words come from my interview guests, not me; at most, I’m transcribing. It’s a ton of work, especially for audio. But it’s not what I want to measure as my “writing”.
2-4 6P / profiles on ethics of genAI music articles (companies and PARTs, much in WIP) - not counting articles authored primarily by my collaborators.
1-2 6P / AI Fundamentals articles (in WIP).
2-3 AAaB articles on whatever gets me fired up this month.
24-32 Notes plus Comments
My effort budget for August posts will remain at 6-8 hours/day, or 30-40 hours/week for 3.5 weeks. However, interviews will likely make my effort bursty.
New Actions for August 2024
I discovered through another Substacker’s Note that the (i) icon in the lower left corner of the Substack editing window shows estimates for word count and approximate reading & speaking time. This will reduce my effort for gauging size and time while I’m writing! 👍
So, as a minor tweak to action (A10) on measuring readability, I’m going to start:
Capturing the word counts and reading & speaking time estimates from Substack’s editor, and
Comparing the reading & speaking time estimates from Substack and Grammarly to the actual audio recording times for posts with voiceover.
Aside from that, I’m giving myself just one new action:
(A13) Figure out if or how I can get visibility into whether people are listening to the audio voiceover tracks, whether through Substack or through most of the external podcasts.
I’m hoping the audio is improving accessibility of my writing to people with more listening time than reading time. I want to know if it’s working! Measuring how many people are actually listening, if I can, is a good first step.
Other than that, carrying through on my current actions is going to keep me plenty busy in August. I want to give the new content plan a few months before making any further big adjustments.
“It does not matter how slowly you go, as long as you do not stop.” (Confucius)
What’s Next
That’s my writing retrospective for July. I commit to doing another one after the end of August. I’d love to hear your thoughts on any of this!
Thanks for reading Agile Analytics and Beyond! Subscribe for FREE to receive new posts (these retrospectives are monthly).
References
Archives analyzed in this retrospective
Substack:
Agile Analytics and Beyond (now includes former agile Teams content)
LinkedIn: