In the 1st part of this series, I…
- explained why you should think about building an app on the Google Workspace Marketplace
- collected all 5,300 apps on the marketplace in a Google Sheets spreadsheet
In this post, I’m going to come up with several criteria to filter out the noise from this app dataset, so I can focus on researching valid opportunities.
Apps for educational institutions
If you look at the top apps by install count, you’ll notice quite a few targeting the educational sector: SketchUp for Schools (135m installs), Pear Deck for Google Slides Add-on (90m), Kami (67m), to name a few.
I don’t know much about selling to this vertical, but I get a feeling it won’t be easy. School budgets tend to be tight, and they’re unlikely to engage with a brand new startup. Teachers won’t pay out of their own pockets, unlike, say, marketing professionals.
There are exceptions, like Arvid Kahl’s FeedbackPanda, I know. In his case, the stars aligned—his wife was an online teacher, giving him firsthand insight into a pressing problem in that niche. We don’t have that luxury.
Apps that deliver little value
Looking at this list, you’ll also spot some apps that you just can’t charge folks for, no matter what: ZIP Extractor (66m), Easy Accents - Docs (39m), ChangeCase (4m), among others.
Again, I’m probably missing out on some valid opportunities. To which my response is - do you want to deal with this kind of feedback I’ve seen for one such app?
Applying labels with AI
The above 2 filters sound all nice, but how do I quickly apply them across all 5,300 entries in my spreadsheet? Well, I can create a JavaScript function in Google Apps Script, that calls OpenAI with a prompt, and then use it as a custom function directly inside Google Sheets.
My Google Apps Script function
I asked Gemini to help me write the function, and then created a prompt for each row in Google this format:
categorize this Google Workspace app as: 1) if it's for education, and 2) if it's valuable enough to be a paid service. return as 2 values - education/noneducation and premium/free, separated by comma - example "noneducation,premium. The app description: '''<DESCRIPTION_COLUMN>''. The app developer name is '''<DEVELOPER_COLUMN>''. The app name is: '''<APP_COLUMN>'''10 mins and $0.10 of OpenAI API credits later, I am now able to filter out 1,000 apps!
Drive / Admin apps
Scanning through the results after that 1st round of filtering, I noticed apps that reference Google Drive / Admin:
- Gopher for Drive (15m installs) - insights into Google Drive usage
- Mover (10m) - migrate users to Google Drive
- Backupify for Google Workspace - backups
As you can see, they are system administrator tools. Another area that’s hard to break into as a crappy little startup. So we’ll remove those out from our analysis as well.
Number of installs
An obvious metric to filter by. The question is, what threshold? Let’s look at the stats (after the previous filters were applied):
After some spot checks at different cutoff points to see if there were any legitimate businesses in the mix, I came up with the magic number of 1 million installs.
Next steps
After a great deal of filtering, we’ve now managed to narrow down our analysis from 5,300 apps to 205 apps. In the next blog post, I’m going to look into these apps, and hopefully find that startup opportunity that I’ve been talking about. See you next time!