Avoid "Basket Case" Data Products Through Iteration
Taking inspiration from the world of music to persevere with iterative loops to create great data products
Welcome to the Data Score newsletter, composed by DataChorus LLC. The newsletter is your go-to source for insights into the world of data-driven decision-making. Whether you're an insight seeker, a unique data company, a software-as-a-service provider, or an investor, this newsletter is for you. I'm Jason DeRise, a seasoned expert in the field of data-driven insights. As one of the first 10 members of UBS Evidence Lab, I was at the forefront of pioneering new ways to generate actionable insights from alternative data. Before that, I successfully built a sell-side equity research franchise based on proprietary data and non-consensus insights. After moving on from UBS Evidence Lab, I’ve remained active in the intersection of data, technology, and financial insights. Through my extensive experience as a purchaser and creator of data, I have gained a unique perspective, which I am sharing through the newsletter.
As September ended, it's almost a tradition to flood Twitter with “wake up” messages for Green Day's Billie Joe Armstrong, referencing their iconic song. September 2023 also brought another milestone for the band. They released the 30-year-anniversary version of their breakout album “Dookie”. On the album is their most popular song, “Basket Case,” which in 1994 got non-stop MTV airtime (back when the channel played music videos most of the day).
The anniversary version of the album includes demos and early drafts of songs. Hearing early song versions showcases their evolution. For some bands, it might take decades to release these early demos, even as their songs are continually cherished by fans. Sharing art can be scary for artists, especially if it is not yet seen as the final product.
The 30-year anniversary edition unveils an early "Basket Case" version – almost unrecognizable from the final song. The chord progression is there (A series of musical chords played in a repeated sequence), as is the melody. But it’s different lyrics, a slower tempo (speed), different drums (very minimal), and a different-sounding guitar (more acoustic and kind of strummy and jangly).
It’s a fine song. But this isn’t a song that would allow Green Day to reach the wide audience it reached… The song is kind of lame. If it were a data product, it wouldn’t be an MVP1 (Minimum Viable Product) because it’s not really viable… It's just an MP (minimum product).
What does Green Day’s “Basket Case” even have to do with data?
It's all about the iterative process—both in music and data product development.
The first working version may have a lot of vision but may miss the mark
Share and get feedback
Keep what works
Change what didn’t work
Share and get feedback
Repeat until you get to product/market fit2.
The first version and feedback
Data products are solving a problem by providing the missing insights needed for decision-making. An artist’s “product” also provides information, even though it’s not meant to solve problems.
Similar to artists, building a data product begins with a spark of creativity. Data products are solving a problem by providing the missing insights needed for decision-making. An artist’s “product” also provides information, even though it’s not meant to solve problems. Great art helps capture an idea in a way that connects with people emotionally and helps advance their thinking about the subject.
Data professionals and artists have an idea and inspiration for what will capture the insights needed and package them in a novel way that resonates. But that first version rarely captures what the original goal is. This should be expected and not a blocker to actually getting to the outcome desired.
At times, the initial attempt's shortcomings can discourage creators so much that they consider abandoning the idea.
Other times the initial version is held onto too tightly, and despite feedback to change it, decide to hold tightly to the details of the original version.
The reality is somewhere in between. As creators of either art or data products, we need to have a clear vision but be flexible about the details to get to it. We need to hear the feedback, adapt, adjust, and get more feedback.
Keep what’s working and change what’s not working
The initial version of “Basket Case” had the same chord progression and melody for the verse, pre-chorus, and chorus, which were kept in the final version. But the rhythm, tempo, and words were changed.
I find it interesting that in the third time through the pre-chorus, the guitar part is different from the rest of the song—but actually more like the final version of “Basket Case” in its steady rhythm and muted short notes in a row typical of punk rock music.
I wish I could find an article explaining their process to change the feel of the song. I did find an article explaining how the words were rewritten, and I would strongly recommend against that approach (especially for creating data products!). So I’ll just leave that there. But what was apparent about the lyrics change was that the rest of the band liked the new version, so they agreed to change it.
I suspect the same kind of feedback loop happened about the guitar style, leading to an “experiment” to try out what it would sound like if they played the whole song like the 3rd pre-chorus. They likely tried it out before committing to the change, got feedback (maybe from just the band or an inner circle of trusted advisors), and decided to move in that direction.
How did Green Day iterate on the song?
Here’s a later version of “Basket Case” that’s getting closer to the final version.
They changed the rhythm, words, and sounds of the guitar and bass. They cut the solo. They added a new outro. It’s getting closer to the final version. It’s not polished, but it is now a viable product (easy to say with hindsight!).
Data Products: Is this interesting? Is this useful? Is this critical?
That process is necessary for data products. To generate data driven insights at scale, the data product needs feedback on what’s working and what’s not working. The data team needs to be willing to completely change the details while continuing with the final vision.
The iterative process was discussed as related to finding product/market fit in “When to Scale or Keep Iterating”
The data product has to solve a problem for the client in such a meaningful way that they are willing to change how they work, or the data product has to fit seamlessly into the current workflow. A product with feedback that is just “interesting” is not good enough. Even “nice to have” is not good enough. Only feedback that the feature or product is “critical” and “can’t live without” is acceptable feedback showing product/market fit (PMF). Spend time with the users and find out what’s “just interesting”, but not useful. Go back to the product and iterate quickly.
https://thedatascore.substack.com/p/when-to-scale-or-keep-iterating
The final product is not the minimum viable product prototype
Artists throw away their work all the time.
Musicians record a track, review it, and decide on revisions or retakes. Then they re-record it. The prior track is archived, hopefully not thrown out, because then how will they have bonus material to release in 30 years? The musicians re-record the demo to create the final versions of each instrument and part of the song, which can be mixed and mastered as the final product.
Painters too draft and experiment, assessing how their ideas translate on canvas. The final version will reuse the learnings, but they may paint over major parts of the artwork or start on a brand new canvas.
The final version of “Basket Case” is very polished in terms of the technicality of the recording while also enabling the vision for the song to be understood and enjoyed by a broader audience. All of the instruments and vocals seem to have been rerecorded.
If artists can easily move on from their prototypes, why don’t data professionals?
In the data world, we should use those same principles. The work to discover the approach to generating high-impact, data-driven insights is hard. But that work is not meant to be in the final product. The final product needs to be built to be scalable3 and sustainable, allowing many users to benefit from the data-driven insights.
From my early days in data products, I recall how the rush to develop and drive adoption led us to keep prototype code in production longer than necessary. Once we were able to remove the prototype code in favor of properly developed, automated data pipelines4, the ability to expand the scope of coverage accelerated. The longer data products stay in production with prototype code, the more technical debt5 builds and the longer it takes to transition to a properly scalable solution. It may be a hard transition, but one worth doing.
This isn’t to say that all technical debt is bad. To innovate via prototyping is to create technical debt. As discussed in the Data Score entry on iteration, finding the initial signs of product market fit is best found via agile methods6 and fast, handcrafted prototyping. https://thedatascore.substack.com/p/when-to-scale-or-keep-iterating
But, once the product/market fit is found, the work needs to be done to create the scalable solution and throw away that prototype—the same way a musician moves from demos to their final version.
A word on the period after the release of the product: feature enhancements
The live version of “Basket Case” has evolved, which is not different from any product that has been launched and updated to remain relevant with its customers.
In the music world, the song lives on in live performances, which are gradually updated as time passes. Green Day has performed “Basket Case” more than 900 times in live shows. The performance has changed over the years. There are versions where the initial demo version is brought back in subtle ways. Like playing quieter for the pre-chorus (similar to the original demo version). Later, they added a new intro to build up to the surprising start of the song people know.
The analogy for data products is that both the insights needed and the data evolve over time. The data product is never actually complete. Constant iteration post-product/market fit is needed to add and cut features to keep the product relevant to the users. Ensuring the feedback loop continues between users and the data product team in some form of a DevOps7 structure allows for continued iteration.
Afterword: Influences from Music and Data
While I remind myself this is a data insights blog and not a music blog, the parallels with music are fascinating (to me at least). There’s lots of analogies between various disciplines of creating something new that is well received by audiences. This will be an ongoing theme of The Data Score, Composed by DataChorus: drawing influences from music as it relates to generating data insights at scale. Earlier in the year, the newsletter drew comparisons between Alternative Rock and Alternative Data8.
- Jason DeRise, CFA
MVP (Minimum Viable Product): A product version with just enough features to be usable by early customers who can then provide feedback for future product development. The key here is the viable part of the definition, which often gets missed in favor of the minimum description.
Product/Market Fit: The ability of a product to meet the needs of customers, generating strong and sustainable demand for the product. This term refers to the point at which a product or service has been optimized to meet the needs and preferences of its target market, resulting in strong customer satisfaction and retention. Achieving product/market fit is considered essential for the success of a startup or new product.
Scalable: A solution designed to handle growth, ensuring performance remains consistent as demand or volume increases
Data pipeline: A set of data processing elements or tasks connected in series, where the output of one element is the input of the next one, converting raw data into cleansed and enriched data, typically managed on an automated schedule.
Tech Debt (Technical Debt): the cost of reworking previously implemented code that is no longer suitable for business needs, typically created when applying quick solutions in the development process. The replacement of the initial work with sustainably written code reduces the technical debt, improving the efficiency of the product.
Waterfall versus Agile Methodologies: “Agile project management is an incremental and iterative practice, while waterfall is a linear and sequential project management practice.” - Atlassian definition: https://www.atlassian.com/agile/project-management/project-management-intro
DevOps: The combination of Development and Operations in a continuous cycle typically used in software development. The goal is for applications to be created quickly and deployed into production on an ongoing basis, where the operations team is involved in the process and provides consistent feedback to the development team to handle new feature requirements and the removal of bugs.
Alternative data: Alternative data refers to data that is not traditional or conventional in the context of the finance and investing industries. Traditional data often includes factors like share prices, a company's earnings, valuation ratios, and other widely available financial data. Alternative data can include anything from transaction data, social media data, web traffic data, web mined data, satellite images, and more. This data is typically unstructured and requires more advanced data engineering and science skills to generate insights.