TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Tell HN: How I pivoted or: please review my startup redux.

9 pointsby drxabout 14 years ago
A while ago I posted on HN [1] about my film/tv show/video game recommendation app called Archfinch [2]. It wasn't received well and it didn't even make it to the front page. I was a little disappointed that fellow HNers didn't care about my app and that no blogs wanted to write about me since I was overshadowed by a similar app a month earlier (which at the time I didn't know about). But such is life and I hacked on.<p>I tried a few ways to advertise around, but I noticed a common trend: some people naturally bounced, others liked the site enough to rate dozens of things, I got some nice emails from users who liked Archfinch a lot but retention was very low. To my amazement of people who rated anything at all, the average rating count was high:<p><pre><code> avg | stddev ---------+---------- 60.943 | 68.147 </code></pre> Clearly I was doing something right. But there was no way I could keep people based on movie recommendations alone -- people don't need that kind of service often, and by the time they do, they will have forgotten about Archfinch.<p>I implemented something I planned to do 'in the future': fresh links, HN/reddit-style, with a twist: you get personalized link recommendations. This seems to work well so far, people are coming back for more. And it has a chance to work -- reddit, Digg, etc. have all succeeded in the past and people need their daily fix of links.<p>In theory, link recommendations could be the end to the 'eternal september' syndrome that all such websites suffered in the past: the algorithm could recommend cat pictures to the masses and awesome links to people like us. Time will tell however.<p>Some of you mind find this boring but personally I like reading about people's mistakes and their subsequent analysis so maybe some of you will find this interesting as well.<p>If you feel like checking out Archfinch [2], feedback is very welcome, either here or at luke@archfinch.com, I read (and try to respond to) all emails.<p>References:<p>[1] - http://news.ycombinator.com/item?id=1601930<p>[2] - http://archfinch.com/

5 comments

revoradabout 14 years ago
This might sound harsh but I hope it's useful.<p>The landing page is terrible. Before I can make head or tail of what this site is, you are telling me I've been logged in as a temporary user, I might lose my data yada yada. What data? What are you talking about? Do you mind introducing your site first?<p>You need a simple one-liner describing what or who your site is for.<p>Next, please don't use plain text links for films. Movie posters exist for a reason. You need images asap. Engagement will shoot up. The same goes for users.<p>Why are you hiding your rate buttons? You know hover doesn't work on touch devices, right? Even when they are visible, they are too faint. And too complicated. Don't make people think if a film is 1 to 5. Everyone's now used to Like or Dislike. Just keep it simple.<p>Lastly, on a meta level, when you are looking for advice on your app or startup, you should always first search HN for advice given to other people with similar ideas. For example, a quick search for your case brought these up:<p><a href="http://news.ycombinator.com/item?id=2253574" rel="nofollow">http://news.ycombinator.com/item?id=2253574</a> for <a href="http://www.subzin.com/" rel="nofollow">http://www.subzin.com/</a><p><a href="http://news.ycombinator.com/item?id=519649" rel="nofollow">http://news.ycombinator.com/item?id=519649</a> for <a href="http://www.greatmovieexperiment.com/" rel="nofollow">http://www.greatmovieexperiment.com/</a><p><a href="http://news.ycombinator.com/item?id=1767309" rel="nofollow">http://news.ycombinator.com/item?id=1767309</a> for <a href="http://watchth.is/" rel="nofollow">http://watchth.is/</a><p>Good luck!
评论 #2377910 未加载
rabidonrailsabout 14 years ago
I think a good start would be a phrase explaining what service Archfinch provides. I felt like you should greet me with a phrase about your service. "Archfinch 'find what you like'" seems to sum up the site in four words. The service you provide to me are the recommendations, I pay the system with my ratings for those recommendations.<p>People like to accomplish tasks. Your initial list of movies to rate is too long. I'd get that down to five. You can continue to ask about other movies but you need to let people feel like the system has enough information after five (maybe ten). The site is boring after rating a list of fifty movies - I want the recommendations!<p>You must have a "haven't seen it" option.<p>I ventured into the lists page and was greeted with a "We don't know you well enough to recommend you anything." I may be willing to fill out the information that you need so that you can give me a recommendation but I want you to lead me there...where do I go to give you the info you need? (link?)<p>What is the Pics link about? Why are those pictures recommended for me? Are they supposed to be?<p>"Save your data by creating a Permanent Account" ...sounds serious, like I'm buying a burial plot. Why not just "save your data by creating an account?"<p>Just my .02<p>Keep chuggin, I think you could have something you just need to polish it.
评论 #2377916 未加载
erikpukinskisabout 14 years ago
It's not clear to me why I would be interested in Archfinch over, say, Reddit.<p>And the movie recommendation functionality is totally gone now?<p>As you were explaining your pivot, I thought you were saying that you were going to add a hn-style "hotness" feed to a movie ranking site, which I thought was a genius idea. Basically, people would write pitches for their favorite movies, why they're awesome, and then everyone would up/down vote them, and you'd get a feed of what rises to the top. It'd be even better if it was personalized based on your taste graph, but that wouldn't be necessary to start.<p>So, I don't know... different kind of idea than what you're doing, but it seems sort of interesting to me.
drxabout 14 years ago
Clickables:<p>[1] - <a href="http://news.ycombinator.com/item?id=1601930" rel="nofollow">http://news.ycombinator.com/item?id=1601930</a><p>[2] - <a href="http://archfinch.com/" rel="nofollow">http://archfinch.com/</a>
bmeltonabout 14 years ago
I like it, generally speaking. I played around with it a bit, and here are my thoughts:<p>- The site needs some character. The logo has some, but isn't enough; It's REALLY neutral to me, and I'm a pretty big fan of minimalism.<p>- There's no "I haven't seen it" option, or "Let me rate this later". Which means that when I got to the end of the list of things I could rate, I was at a loss as to what to do next. Also, it bugged me that I couldn't make 'Cloverfield' go away, even though I've never seen it.<p>- What do I get out of it as I rate them? I certainly don't mind rating these things to help evaluate, but ... is there ANY reason I might come back and do this again later?<p>- The lack of registration option is a little confusing. I hit the home page, rated a few things, I see a log OUT option at the top, which is a little quirky, and I see a log IN -- oh, never mind, it was in the middle. If that's the call to action, it should be more pronounced, I feel like.
评论 #2376945 未加载