Friday , May 7 2021

PSL Studio Kill: Xylo, Hacker News

At PSL, we kill nine out of ten ideas that we test. We haven’t talked much about these publicly before, but we wanted to share what it looks like when we make a “kill” decision. Our goal in publishing this is to help other founders think about how to do early validation the way that we do inside the studio.

The idea we tested was Xylo, which solved a problem that millions of parents in the US experience: finding great teachers for music lessons. Our proposed solution was an online platform to connect these teachers with students for remote (or in-person) lessons with a great user experience. Sounds like a great idea, right? Let’s dive in!

This email was sent by Peter Denton, our validation lead, to our whole team.

A section of the test site for Xylo. Note that we typically use very basic landing pages so that the quality of the design does not affect results. Simple templates allow us to ensure our convesion data is attributable to resonance of the value propositions, not the site execution.

—————- —-
From: Peter Denton
Date: Friday, January Subject: Xylo investigation
To: PSL Team

Hey everyone,

I have been looking into Xylo this week and wanted to share my findings with you. To cut to the chase,

I am recommending we kill this idea

, but of course want to share my methodology and see if anyone wants to keep going.

I am working under the assumption this is marketplace business where we take a cut of the revenue in a booking.

If we just start at the raw demand generation

around this, according to Google, over the next quarter, for the entire United States, we could get , clicks for $ 90, 600 in marketing spend. I would look at this as the entire world of actionable intent-driven demand.

We ran tests on both Google and Facebook to get real, in-the-wild numbers, which proved to be worse than Google’s estimated models. (As I discuss later, it’s probably due to high levels of competition both paid and organic).

For (Facebook , tests, the best CTR was 0. (% and the CPC was high with an average CPC of $ 3.) . That is really expensive, likely due to the holiday increase in competition and base CPM going up to reach consumers / parents.

. The “tech-enabled” ad that ran on Facebook / Instagram



, targeting the similar value props (affordable, convenient, tech-enabled, and exclusive) we are currently seeing just a 0. % CTR .

Note: It is important to be sure each ad is clearly messaging a distinct core benefit to understand why people clicked through.

We saw the following funnel , using a blend of Google’s model and our own CTR:

3% CTR (of

, (searches)==, visitors
and 23% conversion (for people “signing up” on the landing page) – 5, 480 registrations. (Note: we saw %, and optimistically assumed we could double on-page conversion rate by optimizing over time, which we have seen with other companies)
and 20% of those paying==~ 600 Total possible customers in Q1.

We are looking at (customers (which would cost around $ (K to acquire). If those 728 customers took 3 lessons each, at $ an hour, and we took a % commission, we are looking at only $ , of revenue .

As for market size , the best estimates from MakeMusic (a leading resource for music instructors), put the number of music educators in the US at about 154, 10. I got to a similar number (, 01 building a bottoms-up estimate by running Google searches by market and taking the sum. (Also, many people believe there is a lot of under-the-table neighborhood activity going on, so the number is likely higher, but not sure that is actually addressable for us).

Regarding competition:

First off, there are a ton of enabled small businesses competing in this space and I’m not sure there needs to be a middle man. For example, I get 4 ads from businesses in Seattle offering lessons for guitar on Google. Thus, they know how to market and get customers, they are offering free first lessons, and have availability. So I am unsure / doubt there is any real consumer problem .

TakeLessons is the largest VC-backed lessons marketplace and has raised ~ 27 MM over 3 rounds. TakeLessons charges % commission and has expanded out into “all things lessons” , which is likely an indicator that music lessons were too small to keep growing in.
(There is also Yousician , which is an entirely tech-based way to teach you music. Think: guitar hero for music lessons. They have about 154 employees, so it’s a sizeable tech team to keep building.
3152346 And lastly there is YouTube, where thousands of guitar instructors teach people a wide array of music-based education. While not the same as a 1: 1 lesson this is completely for free for the consumer.

In conclusion:

I do think there is a healthy demand for this – 1mm people are looking for it every quarter.
(However, the supply side is mobilized and there is a TON of competition, which is going to continually drive up costs perpetually. With % of consumers skipping AdWords, this means very limited early traffic.

  • It was not clear that you could create intent outside of Google , so you would need some unique differentiator in GTM to get real market share.
    Lastly, I can’t imagine that tech won’t win here with non-human based teaching. It’s affordable, convenient, and scalable to any genre, language, right from home on many platforms.

    Please feel free to contest that, and I mean it. This is just my summary after a week of looking at the opportunity as if it were my company. I wouldn’t proceed with it, but would love alternate opinions.

    – Peter

    (Read More)

  • About admin

    Leave a Reply

    Your email address will not be published. Required fields are marked *