Developing successful products in the Eurorack market

Posted on Oct 8, 2022

This research was done with the support of NYC Data Science Academy. A public version of the code used to develop this project is available at this github repo.

Overview

Eurorack is a set of design standards for synthesizer parts. The products are called 'modules,' which each serve a limited number of functions. Modules can be mounted in a case and wired together to create complex, custom synths. The Eurorack market is remarkable in that many brands are small, independent manufacturers, alongside a few larger audio equipment and electronic instrument companies. Despite being a niche market, it is growing in popularity, with hundreds of new brands arriving to the marketplace in the past five years alone.

While it is known that the growth of the global music synthesizer market is accelerating, in particular analog synthesizers, there is little market research on what makes certain products successful, particularly for Eurorack modules. Especially since many Eurorack manufacturers are small companies with limited resources, this research could be invaluable for determining product lines to focus on and develop. This post will summarize research which provides methods for (1) analyzing product features that correlate with successful Eurorack products, (2) finding promising market "gaps" where more products could be developed, and (3) determining strategies for building successful product lines for a brand at different stages of growth.

Context

While Eurorack manufacturers range from DIY to large instrument brands, the majority are small companies. We will look at products from companies of any size, while focusing on strategies which could help smaller brands.

Though some brands produce hundreds of Eurorack modules, the median number of products made by a Eurorack brand is only five.

Data for this research was acquired from the highly popular website modulargrid.net. Modular Grid has a nearly exhaustive catalog of Eurorack products, which additionally has social media components. In particular, users can design synth racks on the site by adding modules from the catalog. As these designs often represent purchase plans or wish lists, the number of racks a product was added to can work as a proxy for popularity or product success.

All results were based off of the dataset of all Eurorack modules listed on Modular Grid, containing the product name, brand, price ($), size (HP), list of features/functions, and number of racks/"likes" as described above. Almost all Eurorack modules have the same height, while HP is a measurement of width of printed circuit boards. This quantity might be important, as Eurorack users are often limited by the size of their rack, and must take the HP of each unit into account. The list of features/functions typically consists of functions the module can perform such as drum, reverb, oscillator, LFO, etc.

Analysis

The dataset was cleaned to include only branded Eurorack modules, with accessories (like blank panels or expanders) excluded. We will first look at the initial EDA, since it suggests transforming price and popularity (number of racks) and splitting the data into subsets (single vs. multifunction modules). This will allow us to answer the three research questions from the overview more clearly.

EDA

We can coarsely split Eurorack modules into two types: single and multifunction. While imperfect, simply checking if the list of features/functions has one vs. more than one descriptor works well for most products. Single function vs. multifunction modules have significantly different average prices (p β‰ˆ 2.1 x 10-35 < 0.05), which can be seen below, with multifunction being significantly more expensive.

The average price of multifunction modules is significantly higher than that of single function modules.

While both price and log(price) have somewhat irregular distributions, log(price) is almost perfectly normally distributed within the 'single function' and 'multifunction' families of products.

Similarly, the distribution of log(number of racks) over the entire set of products (single and multifunction) is almost exactly normal.

The distribution of log(number of racks) over the full set of Eurorack modules is approximately normal, hence will be our proxy for product popularity/success.

(1) What factors contribute to a product’s popularity?

We will show that the functionality/features of a product is a good indicator for its success. It is a better indicator than price (which does not seem to have a big impact on popularity), and easier to analyze than width, which likely has a complex relationship to popularity.

For both single and multifunction modules, cost and popularity have low correlation (r<0.25). This suggests price is not overall a competitive feature.

Price and popularity have a very low correlation for either product type

To begin with, width has a much more irregular distribution, even within the subsets of single and multifunction products. Moreover, the relationship to popularity is highly nonlinear, likely resulting from interactions with other properties like the brand or particular combination of features/functions. While this is a rich feature deserving of more exploration, we set it aside as it has few global patterns associated with success. Companies may want to incorporate width into their analysis only after having determined the types of modules they intend to develop in their product line.

On the other hand, multifunction modules are significantly more popular than single function modules, despite their higher price (p β‰ˆ 1.86 x 10-34).

Before putting the nail in the coffin for single-function modules, we check if there are any features/functions such that single-function modules with that feature are more popular than multifunction ones. Ultimately, there were no functions for which single-function modules had a significant advantage. For the only 5 functions where single-function modules were more popular, the difference in popularity was not significant.

Though dynamics, switch, frequency divider, comparator, and video single-function modules were slightly more popular than multifunction modules with the corresponding feature, these differences were well outside of statistical significance.

Based on these results, companies should avoid producing single-function modules. The categories above are the only ones where multifunction modules do not have a significant advantage, in addition to clock modulators, reverb, midi, and sampling (where multifunctions' advantage is insignificant). Should a company decide to launch a single-function product, it will likely not be able to compete outside of these areas.

However, simply increasing the functionality does not increase popularity. In fact, most multifunction modules have between 2-4 functions/features. Comparing the ranking of modules by number of functions/features vs. their popularity ranking reveals a low correlation (Spearman ρ < 0.2).

Increased functionality does not correlate with increased popularity. Here, 0 indicates the lowest rank.

Summarizing, price does not seem to be a competitive factor, while the relationship between the module size (width) and success is much more complex. On the other hand, multifunction modules are significantly more popular, despite their higher price. However, simply increasing functionality does not lead to more successful products, either. It is then necessary to look at the popularity of specific features/functions, which also reveals market gaps.

(2) Are there "market gaps" of highly popular, but underproduced, types of products?

More than just single vs. multifunctionality, the popularity of products varies greatly based on the features offered. Interestingly, the popularity of modules containing a given feature does not track the number of modules being produced with that feature. This suggests potential gaps where competitive products could be introduced. The graphs below show the average popularity and number of modules which have a given feature or function. Some disparities, where a feature is highly popular but rare in the market, are highlighted in red.

In particular, we can see that loopers, function generators, samplers, quantizers, reverbs, and envelope followers are relatively rare in the marketplace, while modules with those features are highly popular. Modules with these features will have small competition in highly popular areas.

Feature-pairings

We have seen that single-function modules are less popular than their multifunction counterparts, but simply adding more features does not necessarily increase the popularity. Additionally, we have identified some product features which are indicative of market gaps. Suppose a brand wants to develop a reverb, envelope follower, or sampler. Knowing that that they should add further functionality, how can they decide which features/functions to add?

To answer this question, we analyze the success and availability of different feature-pairings. Concretely, for each feature, we consider all other features it could be paired with. We then compare (a) the average popularity of modules with both features, to (b) the average popularity of modules with the first, but not the second feature (though they may have other features). To find successful feature-pairings, we look for places where there is a significant boost in popularity gained by adding the second feature compared to leaving it out. We then check if there is a market gap for this combination by checking the proportion of modules with the first feature which already have the second feature.

For each pairing, we compute a number of metrics: (1) the increase in popularity when the second feature is added, (2) whether or not this increase is significant, (3) the proportion of products with the first feature which also have the second feature (indicating rarity in the market), (4) the ratio between 1 and 3, and (5) whether or not that combination of features is already offered by a 'top brand'. We can think of metric (4) as the (market) advantage of adding the second feature, since this ratio goes up if either the boost to popularity is higher, or the proportion of products with this combination is lower. For (5), we define a top brand as one in the top 5% of total number of Eurorack modules produced.

This method helps identify potential successful feature-pairings which are lacking in the market. The table below shows the top five feature combinations in terms of increase to popularity for a sampler, where that increase was deemed significant (p < 0.05). For reference, the average popularity in log(number of racks) is 5.08 with a standard deviation of 1.74.

Top five significant increases to popularity. The heatmap colors highlight the strength of the metric.

For example, the top row indicates a potentially successful market gap, with an increase in popularity by almost a standard deviation. While samplers which have onboard distortion are significantly more successful, they currently account for only 3.3% of the samplers on the market. Additionally, no top-producing brand currently manufactures a sampler with onboard distortion. Similarly, samplers which can work as a drum module also have a significant increase in popularity, while only accounting for 17.6% of the sampler market.

We can perform a similar analysis over all feature combinations. Out of over 1000 different feature pairings, we can identify a few dozen as having the greatest potential for success. We highlight a few combinations which make sense in light of domain knowledge and their potential market advantage.

Some feature combinations which have a potential market advantage.

For each of the highlighted pairings, adding the second feature provides a significant boost to popularity, though occupying about 1% or less of the existing market of modules with the first feature. Additionally, in all but the last, these combinations are not offered by top manufacturers, indicating areas where small companies may be able to break through.

It is important to note some limitations of this approach:

  • It only looks at feature combinations which already exist in the marketplace. This way, the quantification of the potential success of a pairing is supported by existing data. However, it cannot identify new successful feature combinations, which would require another analysis.
  • Families of modules with a particular pairing may have 'hidden factors' driving the success. For example, other functions in common, size, and/or price.
  • Relatedly, success may be driven by a small set of highly successful modules filling a more particular niche.

However, these comparisons can still identify underproduced areas where products have been highly successful, which can help limit the scope of further product research. Some of the issues above can be addressed by examining the products having the successful pairing more closely.

(3) To have more highly popular products, should brands specialize or offer a variety of products?

Our answers to the first two questions identified the importance of functionality. Popularity of a Eurorack module varies significantly based on its functionality, and functionality can also be used to identify market gaps. On the other hand, price is not a strong competitive factor, while the benefits of making a module a certain size are much more complex. Though the previous sections showed ways to identify functionality that can make a single product more successful, it now makes sense to ask how functionality across a product line can benefit a brand.

To analyze product lines, we created a redundancy metric which assigns a number to any set S of products. This number represents how much redundancy in functionality there is in the collection of products.1 For each brand producing at least n products, we defined the top n products to be the n most popular products made by the brand. We will say that a product is highly successful if it is in the top 5% most popular Eurorack modules on all of Modular Grid. We were interested in how redundancy among the top n products that companies make corresponded to how many of them are highly successful.2

For each n, we looked at the rank correlation between redundancy among the top n products and the number of them which were highly successful. This is done to determine whether increased redundancy in a company's most popular products led to increased or decreased highly successful products, compared to other companies making at least n products. This does not control for the confound that a company which simply makes more products overall may have an increased chance of more of their products becoming highly successful. So, we compared this correlation against the rank correlation between the total number of products (made by brands with at least n Eurorack modules) and the number of highly successful products amongst the top n.

For each n, we look at brands making at least n Eurorack modules. The blue plot shows the correlation between the total number of products made and how many of the top n products are highly successful. The orange plot shows the correlation between the redundancy in the top n products and how many of the top n products are highly successful.

For a given n, the two curves describe the two rank correlations above for companies making at least n Eurorack modules. When the orange curve is above the blue curve, increased redundancy in the top n products is a stronger predictor of how many of them are highly successful than the total number of products made by the company. We see that variety in functionality is desirable among the top 15 products, so companies with 15 or fewer Eurorack products should focus on introducing more products with varied features. When a company is making at least 15-50 products, increased redundancy in the top product line (specialization) is tolerated. In fact, redundancy among the top 15-50 products correlates with having more highly successful products more than the overall number of products made. When a company is this size, products overlapping in functionality with the existing top products can be introduced with success. However, this effect fades and becomes negligible after about 75 products.3

Summary & further research

The research looked at a number of ways to create successful Eurorack module product lines. Some of the key insights we can draw from the analysis are:

  • Lowering price is not essential to stay competitive
  • People want multifunction modules!
  • Looking at specific functions is useful for identifying market gaps.
  • There are many functions which are highly popular and underproduced (loopers, samplers, quantizers, …)
  • There are many underproduced but potentially popularity-boosting function pairings.
  • When a small number of the top 15 products are successful, a company may want to introduce products with different functionality from their current top products. When a company is making 15-50 Eurorack modules, they may want to introduce more modules which overlap in functionality with their most successful products.

There are many issues which deserve further attention. This analysis did not have access to sales data, which may be much more useful than popularity. Additionally, there was no data which indicated when products were introduced, and it did not include discontinued products. The product line analysis would likely be affected by looking at the available products from each brand over time, or other groupings of products. We only looked at pairings of features/functions, while we may want a higher-order model which takes into account general groupings of features. Finally, we may also want to reexamine width (module size) when restricted to certain groups of successful products and feature-combinations.

Footnotes

  1. Let F be the complete set of functions/features which are used to describe any product in a collection of products S. For each feature/function f in F, define nf to be the number of products in S which have feature f. We define the redundancy of a set of products S to be log(nf1 ... nfp) , where F = {f1, ..., fp}.

    This is a nice metric of redundancy for a number of reasons. First, redundancy(S) = 0 iff each nf1 = 1. This happens iff no two products in S share a feature/function. Second, if S and T are two product lines which do not have any products with overlapping functionality, then redundancy(S+T) = redundancy(S)+redundancy(T), where S+T is the (disjoint) union of the two collections of products. Third, an inclusion of a subset S into a set of products T (representing an extension of a product line S) has the property that redundancy(S) ≀ redundancy(T). This is an equality only if all the products added have no overlapping functionality, and additionally they have no overlapping functionality with existing products. Finally, if we consider adding functionality to an existing product in S, the redundancy will only go up only if we add a function which another product in S already has.
  2. We chose this analysis because: (1) redundancy must be a property of sets of products, but we only want to compare redundancy between two sets of the same size; (2) the set of highly successful products a brand makes must correspond to their top n products for some n; and (3) we only want to compare the success of sets of products that are the same size. The reason we only want to compare the redundancy and success of sets of products of the same size is because larger sets of products have a greater chance of being redundant and having more highly successful products.
  3. It is hard to rely on the results past this point due to the small number of brands making that many products, and the fact that the sheer size of the product line pigeonholes redundancy.

About Author

Zach Stone

I am a data scientist with a background in linguistics research and math. I love to make it easier to analyze and draw insights from complex patterns using a combination of research, code, and modeling.
View all posts by Zach Stone >

Leave a Comment

No comments found.

View Posts by Categories


Our Recent Popular Posts


View Posts by Tags

#python #trainwithnycdsa 2019 2020 Revenue 3-points agriculture air quality airbnb airline alcohol Alex Baransky algorithm alumni Alumni Interview Alumni Reviews Alumni Spotlight alumni story Alumnus ames dataset ames housing dataset apartment rent API Application artist aws bank loans beautiful soup Best Bootcamp Best Data Science 2019 Best Data Science Bootcamp Best Data Science Bootcamp 2020 Best Ranked Big Data Book Launch Book-Signing bootcamp Bootcamp Alumni Bootcamp Prep boston safety Bundles cake recipe California Cancer Research capstone car price Career Career Day citibike classic cars classpass clustering Coding Course Demo Course Report covid 19 credit credit card crime frequency crops D3.js data data analysis Data Analyst data analytics data for tripadvisor reviews data science Data Science Academy Data Science Bootcamp Data science jobs Data Science Reviews Data Scientist Data Scientist Jobs data visualization database Deep Learning Demo Day Discount disney dplyr drug data e-commerce economy employee employee burnout employer networking environment feature engineering Finance Financial Data Science fitness studio Flask flight delay gbm Get Hired ggplot2 googleVis H20 Hadoop hallmark holiday movie happiness healthcare frauds higgs boson Hiring hiring partner events Hiring Partners hotels housing housing data housing predictions housing price hy-vee Income Industry Experts Injuries Instructor Blog Instructor Interview insurance italki Job Job Placement Jobs Jon Krohn JP Morgan Chase Kaggle Kickstarter las vegas airport lasso regression Lead Data Scienctist Lead Data Scientist leaflet league linear regression Logistic Regression machine learning Maps market matplotlib Medical Research Meet the team meetup methal health miami beach movie music Napoli NBA netflix Networking neural network Neural networks New Courses NHL nlp NYC NYC Data Science nyc data science academy NYC Open Data nyc property NYCDSA NYCDSA Alumni Online Online Bootcamp Online Training Open Data painter pandas Part-time performance phoenix pollutants Portfolio Development precision measurement prediction Prework Programming public safety PwC python Python Data Analysis python machine learning python scrapy python web scraping python webscraping Python Workshop R R Data Analysis R language R Programming R Shiny r studio R Visualization R Workshop R-bloggers random forest Ranking recommendation recommendation system regression Remote remote data science bootcamp Scrapy scrapy visualization seaborn seafood type Selenium sentiment analysis sentiment classification Shiny Shiny Dashboard Spark Special Special Summer Sports statistics streaming Student Interview Student Showcase SVM Switchup Tableau teachers team team performance TensorFlow Testimonial tf-idf Top Data Science Bootcamp Top manufacturing companies Transfers tweets twitter videos visualization wallstreet wallstreetbets web scraping Weekend Course What to expect whiskey whiskeyadvocate wildfire word cloud word2vec XGBoost yelp youtube trending ZORI