May 6, 2019

Marketplace not doing so hot.

Andrew @andrewpierno

1,200 visits, 30 users, 0 sales.

I build a developer marketplace where you can buy and sell code and docker containers. I have a few containers for sale mostly around machine learning but haven't sold any yet, they're priced between $49 and $99.

Either people don't understand the site or the containers themselves aren't getting in front of the right audience. I had one person on reddit say they didn't understand what the product was, but the above the fold text says

"Buy and sell docker containers and code."

which, if you're a developer, should make sense to you otherwise you're probably not a developer 🤷‍♀️.

in the doledrums for sure.

  1. 8

    Either people don't understand the site or the containers themselves aren't getting in front of the right audience

    Or option #3, there's no market for the product.

    My immediate thought is that maybe developers have the savvy to build these for themselves and that they don't want to use someone else's docker/code (because they want it done the "right" way, aka their way).

    What have you done to validate there's a demand?

    Other possibilities:

    • You're not getting enough traffic (10 visits a day? 100?)
    • You're not getting enough traffic from the right sources (no point getting 1000s of daily visits if they're all from Fremantle Dockers fans)
    • The docker setups you have for sale aren't of interest to your audience (maybe people aren't into ML or the other dockers you have available

    Also, looking at the SugarKubes website:

    • Your search/graph tags aren't setup properly (no meta description when I search for sugarkubes.. can't find you at all when I search for "buy and sell docker containers"
    • Your website doesn't render properly on my browser (15" windows 10 laptop, Chrome)
    • Ditch every animation on your website. I don't need things wooshing in and fading in and out as I scroll like it's a powerpoint from 1998.
    • Your hero is self-defeating. Get rid of the giant logo, get rid of the "login or signup" bit (the key CTA is "buy this docker", right?), and make the categories and products front-and-center. I would envision hitting your website, seeing a list of categories across the top, and then a default category selected/highlighted with the containers available from that category directly underneath
    • Everything after the "Popualar Containers" section is irrelevant. FAQ should be a separate page (or at the bottom of each product page)
    • I don't understand why I need to sign-up? Is this a subscription-based service where I can download unlimited containers? I thought each one was an individual sale. Why not make all of the products in your store visible to the public?
    • I clicked Join in the menu, and then I have to click Join again.
    • There's no pricing anywhere
    • Why can't I click to buy the ones on the front page?
      *If I click the title of the containers on the first page, it tries to make me login... why? Surely additional detail about the container can be presented on the public page, so I can decide if I want to buy?

    Generally speaking, I'd say there are a lot of barriers between me visiting the site and becoming a customer. Think of every click as a reduction in my likelihood to buy.

    1. 1

      Thanks so much for your candid thoughts, I deeply appreciate it and think you've unlocked a lot of new options for me to try. Will definitely implement some of your suggestions and report back!

      In my developer-focused ways, I totally overlooked nearly everything you've mentioned here. I almost feel like I would have been better off with a simple wordpress ecommerce site.

    2. 0

      Also just some thoughts on the demand side. My day job is the CTO of a venture-backed computer vision company. I'd kill for a solid marketplace that I could easily buy and deploy high-quality containers (hence why I'm building it, so there's at least an N of 1).

  2. 5

    Marketplaces are tough. What are you doing to serve both sides, separately?
    You've essentially got to serve two different users in their own way. The first three words... "buy and sell..." is a red flag to me.
    Seems you got the content already, you have some sellers. Way to go! nice job. Now just create a single landing page "buy docker containers and code". and super serve buyers.

    1. 1

      Thanks Andrew! I agree with you. I'm playing with the concept of a managed marketplace (so-called marketplace 3.0, a16z definition not the 2013 book by the same name) so I've been building all the items for sale so far and will likely continue doing so for a while so absolutely agree super-serving buyers is the best option going forward.

  3. 4

    That marketing copy could use a work-up for sure. If I hadn't found your site from IH, I'd have scratched my head for a while about what exactly was going on, and I'm still kinda scratching about why I'd want to buy from this place.

    Try re-framing your copy in terms of benefits - I don't really care that you're selling docker containers, I care that you're saving me time, improving my security, and adding superpowers to my ecosystem, all for a reasonable fee.

    Also, consider that you probably have a difficult problem in front of you - the people who are likely to find you are developers, not managers, and the developers frequently don't control the purse strings. You also need to frame your copy in ways that will help your users sell to their bosses and convince them to pull out the credit cards. ("Save days of work developing and testing powerful, secure containers", etc).

    1. 0

      Appreciate your thoughts Dan. You've touched on a core problem here. Developers are used to getting everything for free. The target is likely engineering managers, CTOs, and other decision makers as opposed to individual contributors. Initially, I thought going after developers themselves would work as they brought things they've found back to work but that's likely a longer process than going after decision makers directly.

      And I wholeheartedly agree the copy is proper garbage. I'm going to take a lot of the knowledge dropped on this thread and do another rev.

  4. 3

    I'm a developer but I'm not a real docker user.

    Some thoughts:

    • Your "How it works" section doesn't really answer the question, I would expect a real description of the process in straightforward steps,

    • Your demo section is not really a demo of your product but of a component,

    • I think that the "Use Cases" section should have a "see more" link which allows the visitor to see a list of popular/useful components for the selected topic,

    • The deployment problem could be important so I would expect to get something like a SugarKubes "package" to download and almost a basic interface to easily deploy the component after buying it. I'm not really sure that it's the case here, something in your landing page and your demo should show that this part is super easy.

    1. 2

      Oh interesting. I didn't spend a lot of time thinking about developers who don't really use docker all that much. I guess I'm now so reliant on Docker / Kubernetes I forgot that perspective.

      I love your idea on the deployment problem and showing users how easy it is to buy, pull and deploy. Step one I guess is that I have to make it easy haha.

  5. 3


    Have you reached out to the 30 users who've signed up so far? If you could do a call with them that would be high value for you to gauge where the gap is.

    Second, your ratios seem to be pretty accurate: 2.5% conversion from visitor to sign up is pretty standard for a young site. Of those who sign up typical ratios for young sites tend to be 2% of that. Which would have you at ~ 0.6 paid users.

    Keep pushing!

    1. 1

      Hi Ryan, that's interesting to note I'm not that far off on the ratios. I've reached out to every individual that's signed up (when they sign up I manually send a mixmax templated email) but need to ask them again as zero people have replied to the email. They probably thought it was automatically sent.

      1. 1

        No fancy templates. A personal touch goes a long way!

        Also take a look at :
        for messaging inspiration and ux.

        1. 1

          ok I stripped the fancy template down to just functional components. Hope this converts better. Actually it should convert worse but those who do convert should have a higher likelihood of buying (since the only reason to sign up is to buy and keep track of what you've bought).

        2. 1

          Thanks for the tip! I've seen algorithmia. I'll steal their messaging but digging into the apis it looks like they just scraped github and put the models online. Also, there's actually very little usage (which might be a sign ML models are not a sustainable path going forward).

  6. 2

    You cannot sell docker containers, you can sell solutions that are delivered as docker containers. Also, it is unclear what software is used inside concrete container, what is requirements, what is API. When I clicked on 'popular' container and wasn't able to get any details without login - and my next action was 'close tab'. Don't push users to login, this might be a reason of high bounce rate.

    From 'maker' point of view, at this moment no motivation to use your marketplace - as you don't have auditory yet (chicken and egg problem). Also vendors are more motivated to use cloud provider's marketplaces (AWS, Azure, GCP) where user can deploy the container in a click. BTW, I'm rather relevant audience in this question - I'm .NET developer and I sell some components, and one of them is 'microservice' that is distributed as docker container (it is hosted on dockerhub).

    And the last, it might be possible that niche for developer-oriented ML 'components' is too small. Here you again compete with ML services provided by AWS, Azure and Google Cloud.

    My advice: for some time try to forget about 'marketplace', and try to sell just your ML-oriented solutions.

    1. 1

      Those are some valid points. I agree the focus should just be on the sell side of the marketplace. And you're right, the decision to make a few ML containers was just a guess, and is perhaps too small.

      I've been interviewing early stage companies and also found that I overestimated how complicated many MVPs are. At least for the ones I've spoken with, they would value a simple auth service more than ML containers.

  7. 1

    Hi Andrew, thanks for sharing your story. Could you share the tech stack or website builder you used to build the marketplace?

  8. 1

    This comment was deleted 2 years ago.

Recommended Posts