I’d like to start by saying I’m not totally tech-inept. My 9 to 5 is an analyst role at a martech startup where I work with super talented developers and data scientists. But proximity does not equal proficiency. I think a better way to describe me would be “tech aware” — I have a good understanding of what problems tech can solve, but I lack to knowhow to build those solutions myself.
So when it came time to start Outdone, my co-founder (a fellow non-developer) and I had a clear idea of what we wanted to build. We wanted to create the web’s first gift recommendation engine powered by machine learning. Our mission was to take the stress out of gift shopping. However, this Ai approach required knowledge in a number of arenas that we lacked. So it was clear that we would need to secure outside help — read: we would need to invest a decent chunk of money.
This was scary. And borderline irrational for young professionals. Sure, we were confident that the value prop would resonate with users, but this was an unverified hunch.
Well, as disciples of Ries’ Lean Startup method, we began thinking through the quickest, cheapest, and lowest-tech MVP that we could come up with to test this hypothesis.
In practice this was a (very) poorly designed Wix landing page with a Google Form that asked users a few questions about who they were shopping for — like age, gender, and hobbies. Then, we built some (again, very) poorly designed Facebook ads to promote the page. The ads simply read, “Take the stress out of gift shopping with Outdone.io”.
The idea was this: users would complete the form and submit their email address. We would then take that info and scour the internet for products that aligned with the giftgetter’s lifestyle. And in the end, we’d email the original user some gift recommendations in 24-48 hours.
Within a day of pushing the ads live we were overwhelmed with requests. On top of that, our CTRs and form conversion rates blew the benchmarks we had researched out of the water. This was shocking considering we expected virtually nobody to enter their email into our sketchy looking LP.
So after just a few days of running this experiment we paused it and set off on the journey to build Outdone V1, the version of the site that is still live today (and what we consider to be a higher-tech MVP).
That no-tech MVP experiment was super validating for a couple of reasons:
And, most crucially, all of this was uncovered with just a couple weeks of prep and for less than $100.
In case you’re wondering where that experiment led... the positive traction we received gave us the confidence to invest more time, money, and freelance support into our idea. A little more than a year later we now have a version of Outdone that is truly powered by a machine learning model and was recently named to Built In’s The Future 5 of Boston Tech. https://www.builtinboston.com/2022/01/25/boston-future-5-startups-q1-2022
We are also successfully working through a Pre-Seed round to take the current MVP to new heights in the near future.
So stay tuned and happy gifting!
Great stuff! I can relate to this problem as I was doing a lot of research trying to come up with gift ideas for my wife’s major number birthday recently.
I went through your site to come up some recommendations and was suggested a few clothing brands. It is a simple to use questionnaire process, but how ‘effective’ have the recommendations been? As I wasn’t really compelled with clicking through with the suggestions.
hey @acedesigner, we just launched a major update. I'd love to know if V2 helps you out more than the MVP did. Check it out on Product Hunt!
https://www.producthunt.com/posts/outdone-v2
Thanks again @acedesigner!
We're actually gearing up to launch a new version of Outdone later this summer! If you'd like to stay in the know and do us a huge favor, feel free to follow us on Product Hunt! https://www.producthunt.com/products/outdone
Fair question! The current site is definitely limited — we don't have a huge amount of brands in our engine and our model isn't trained on an ideal amount of data yet. But it's still an MVP in our mind.
That said, most feedback we have received has said that the recommendations were very accurate — with some exceptions, of course.
For V2 we'll be including almost 3x the number of brands (and expanding beyond the clothing category) and we will be using a substantially larger dataset to train the model with. Our expectation is that this will yield more accurate recommendations for an even wider range of individuals.
Our hope is that you'll have a very different experience when you use Outdone a few months from now. (:
Congrats, my brother had a similar idea a few years ago but never built it. Nice to see a market for it and good luck!
Hey @jabajac we just released a major update and launched on Product Hunt! Check it out and show some support if you like what you see!
https://www.producthunt.com/posts/outdone-v2
Thanks again @jabajac!
We're actually gearing up to launch a new version of Outdone later this summer! If you'd like to stay in the know and do us a huge favor, feel free to follow us on Product Hunt! https://www.producthunt.com/products/outdone
Thanks for the kind words @jabajac! We figured if we didn't build it soon enough, someone else would!
👏🏼👏🏼👏🏼👏🏼