Read With Kaffae
Extension that helps you remember more of articles you read
It's been a few months since I implemented the premium membership plan in the app. I was almost sure it wasn't working by now despite the user growth after Product Hunt launch in the meantime. But Stripe gave me a nice surprise email!
Right now, I'm happy someone appreciated it enough for an upgrade.
I was preparing for this day almost for a year (It kept getting bumped in my todo list). But it wasn't me who initiated. A marketing guy Craig stumbled upon the Firefox plugin, and he just shared it with crappy photos. Ugh. I only noticed it when my friend messaged me on LinkedIn.
I thought ok, I might as well get over it. And started promoting it among my friends. That was already after 15 hours from the initial post. It had only 10 upvotes. So, I thought it'd maybe go up about 20 with my friends. (Keep in mind I shared it on HackerNews earlier last week which got 0 votes. My expectation was very low at this point).
I shared it on my Facebook, Twitter and LinkedIn. I sent direct messages to 10 of my FB friends. And I sent about 50 messages on Twitter who were following me as well as Product Hunt. That's about it.
On top of lots of congratulation messages, I am quite surprised I get connection requests from LinkedIn from a few new people. And I've received a message from one of the venture capitalists. Also, I've received an email about an interview request. I've been developing it silo for quite some time, so this attention has been quite overwhelming so far to be honest.
I'm grateful for this outcome. But more than anything, I am happy I can get it off the long time todo list, so I can focus on other stuff like pitching journalists or writing a blog.
I have a few takeaways from the launch:
- Don't stress yourself over Product Hunt launch. It's just a free forum after all. It doesn't make or kill your project. It's just one of the many places to receive product feedback.
- Be prepared to launch on Product Hunt 4-5 times. InVision used to launch multiple times with a major update. And I'm planning on doing the same. You just build lots of features. That continuous launch mentality helps those who suffer from perfectionism.
- Once you build public reputations, people become kinda nicer haha.
Originally posted: https://www.indiehackers.com/post/product-hunt-launch-what-happened-and-takeaways-bcf33a052e
This morning I was surprised to find a spike in the web traffic. Google Analytics in the source sections says email. Email? I'd never used email.
It looks like Hackernews included Kaffae into their newsletter last week. Yay 💪
It's in the Cutting Room Floor section. That means the idea is worthy of mention even though it didn't get enough attention. I interpret that as the idea is pretty cool. I'm gaining confidence I should start pitching ideas to journalists:)
https://newsletry.com/Home/Hacker Newsletter/b7ca3ed3-a252-4a2d-cac5-08d7d216ff4b
Over the past year, I launched on Hackernews 4 times. Their community guideline is it's ok to launch if there's a significant change in your product.
I think it's the very valuable place for indiehackers to experiment with the different tag lines.
To be honest, the first time I shared, it received no like. And I was quite devastated. That is the end of the product right? As an engineer, I never knew the power of different phrasing because it's essentially the same product.
But the last time I shared, I received 8 likes. It's not big, but the learning is how you phrase in the headline really matters. For example,
Good: Show HN: Build a Personal Library of Articles Automatically
Bad: Show HN: Kaffae – Chrome extension to track your online reading
I am not trying to encourage spamming on the forum. But the most they'd do is to just ignore. I think the experience you get from experimenting different tag line is very valuable. I don't want founders to give up simply because they got so little attention on HackerNews.
Is there a better way to track your readings? I've been long dissatisfied with Chrome History. It is almost impossible to find articles I read. They get buried under the jargon of random urls such as Facebook. So I built this Chrome extension.
Inspired by Fitbit for reading, I first shared this project with Hackernews 10 months ago purely as an analytics tool. The challenge has been properly identifying what constitutes articles and parsing content properly for accurate categorizations. Since then, I've been exploring what type of useful information that can be built on top. Its article recommendations became ready for tryout so I wanted to share this (https://kaffae.com) again.
Recommendation features include an algorithm that caters to longer time-insensitive content, and an ability to walk through related articles directly from the browser page you're on.
Tracking/Analytics features are to understand how much you read in a given week, both the number of articles and time spent reading, automatic tagging of your articles into 81 categories, categorize your articles by publishers, sentiment analysis (negative vs positive) of the content you consume, and a goal settings for a gamified reading.
Who is it for?
You read about 10-80 articles per week. You read on a laptop (obviously). And this tool may be great for those who pay for publishers and want to check how much of their contents you consume.
Future possibility includes podcast recommendations based on your article readings and/or reviews of articles/publications especially of long-form ones.
I started my week by developing a leaderboard. This idea was inspired by many of the sports tracking apps such as Runtastic (now Running) and Strava. They take pride in their leaderboard. The biggest gain they get as an online platform is to have a shared feeling on a common activity and goal (in this case, to run longer and more frequently). It also boosts the community feeling as a side benefit.
That encouragement translates directly to Kaffae platform where it says to encourage more reading. It shows the top three of those who read the most in the previous day. It will set a standard in the platform, being inspired to read more yourself. With the applaud action, you can show appreciation to them as well.
The only question I have now is about privacy. I notice Running shows a private account on the leaderboard while Strave doesn’t. The privacy setting is one of the most things I become cautious and difficult to go by with…
Another change I made was adding a second line chart on your reading habit graph. It’s the first step the platform encourages you to spend more time reading. That data should be as valuable as the counts of articles you read through.
Technicality speaking, it was not as difficult. ChartJS has extensive support on adding multiple lines on one graph. The only part I needed to do was normalization and scaling of the number to fit with the read count properly.
The idea of a suggestion feed is to pull in contents from real people instead of machines. However, it has a chicken or an egg problem. When there are a few users, the feed would remain empty. So it needs to prepopulate it initially with existing contents.
The solution was to add RSS feed and treat publishers like another account.
This week I added 14 publishers: Popular Science, Science Magazine, AEON, Forbes, Vice, Vox, Nature, National Geographic, Literature Hub, The Paris Review, Vox, Artnews, Artspace, Frieze, and Artsy. They were selected on the basis of the length of each article. Since the feed suggestion is based on articles you read and the aim is to learn deeper, I omitted news like contents. Those 14 would be added on top of the existing 12 publishers so it would have a total of 26 publishers.
I was working on adding bot publishers. If there were more contents to give, the algorithm works much better. The concept came from Deep Learning where the algorithm was not changed but the amount of data available at hand made the difference. The focus was on adding longer articles. Longreads so to say. On top, I picked the topics around science, literature, and art since they were the primary interests of the existing users. I want to give the best suggestions for a few users and wait for their feedback. The idea of reading in the tracked platform makes sense, but there are so many places you can consume contents, I ‘m still skeptical if this idea would be validated. The feedback will be to come in a week or two.
For the past month, I've been testing who would most find the product useful. Especially there were 2 ideas I've been playing with. To share articles with close someone in a relationship. And the other one is more for discovering your friends articles. So, it'd be intimacy vs fun app. The underlying product architecture would be the same, but the marketing message would be totally different.
The part I'm having the most difficulty is where to get feedback. I could get feedback from friends who are using it. But I wanted to also get feedback from the market.
So, I've decided to try Facebook ads. I set up 2 ads with different images, different user segment, different text copy. I parituclarly liked the images. One with warm feeling with a girl sitting on a bench gazing at a book. The other is a guy staring intensely at newspaper burning. If one ad comes out way more effective than the other, that will be a good go sign. If there's no variation in click count, the test would fail.
I've put down only $2 on each ad campaign, since I want to run multiple variations thereafter.
It started to run only today. The result would come back soon after.
UPDATE:
Here're the screenshots of 2 different ads I made targeting different segments.
-
2
Sure good idea but you’ll need to spend more on ads and have more images text combinations etc to really see what works
-
1
Thanks Mike for the tip! Ya, I'll slowly boost up the ad budget!
-
-
2
You will probably need to gather 1000+ data points at minimum in order to do a proper A/B test. Otherwise at your current spend you will probably only get a dozen interactions. Also I wouldn’t just look at clicks: that only says how attractive the ad is, nothing about the product. You want to measure engagement as well.
-
2
Thanks gligildner for the advice. Ok. good to know. I might boost up the budget eventually. Right now, the site's traffic is very low, that I can technically analyze the traffic with a regular Google Analytics. But like Andrea said, I should look at Facebook Pixel too. It sounds quite handy.
-
-
1
You can’t split test ads with a $2 budget and expect statistically significant results.
If you installed the pixel and you are using the conversion objective (don’t boost posts, it’s completely useless), the FB Ads algorithm will need to run for 2-3 days with a budget that is at least 5x your average CPA to see meaningful results.
-
1
Hey thanks for the info. That's good to know. I haven't installed pixel, but it's a website promotion.
Do you know anything about the real click rates? Facebook suggested 10-28 expected clicks based on my reach. Is that accurate or more of bs?
-
1
That’s only a suggestion based on other advertisers’ past performances. However you shouldn’t focus on CPCs or CPMs, but rather on your CPA (cost per action).
What is your goal? Getting subscribers, making sales...?
-
1
I see. CPA is for more of the complete sales funnel yah?
The goal is very simple. I want to see which customer segment responds the most to the message (age, gender, interest, etc). Or even to check if the startup idea is valid for someone to begin with. So, I'm only looking at clicks on the promoted post, not so much on conversion at this stage.
-
2
@massanishi as Andrea says, your budgets are far too low to get any kind of statistical significance from this.
Plus, (and this is from doing this with much larger budgets), FB ads will really not end up telling you a huge amount about the real hook behind a service that (as far as I can see) doesn't yet exist. You are testing far too many things, which will mean that even if there is a big difference between one campaign and the other, it will be very hard to understand why one resonated and one didn't.
There's a general trend to try to 'let the market tell me' why my product resonates / what users value, but in my experience your time and money will be far more effectively spent putting the product in people's hands than trying to set up split tests on Facebook.
-
1
Thanks ambudge. I appreciate you guys' experienced marketers' input. You're probably right about the statistical significance. I was hoping to conclude the test if the variation came out very different.
I came across this Facebook testing from some Lean Startup folks. Maybe split testing doesn't work, but wonder if you can validate your startup idea with ads and a landing page (and signup) 🤔
-
2
Split testing works but you need to spend a little more money than $2....How about trying with $100 on each ad?
That'll give you a few thousands impressions guaranteed. Also I would let the experiments run for a while, no need to do it for just 1 week. You can also A/B test if seeing the ads multiple times have a better impact on your audience. Finally I would look at the "copy" of the ads you created. I think you can come up with something more catchy. The 2nd image caught my attention, the first one not so much!-
1
Hey orliesaurus! Yes, I might boost the budget. I wanted to test out the water in the first try.
Roger that on running it more than 1 week.
And ya, thanks for the image/copy feedback! The $2 campaign did come out more clicks on the 2nd image (I guess inclusive anyways statistically, but hey!). Honestly, that kind of qualitative feedback can be worth much more than the ad clicks:)
-
-
-
-
-
-
-
Yesterday, I've got a first word of mouth!
It's still from my friend's inviting her friend. But that's a huge improvement considering most of my friends have stopped using the app after 1 week.
My friend's been using this app for 6 months, and she genuinely uses it. But it's always sharing just with me and her. She never shared Kaffae with anyone else at this point.
This was funny to me because this word of mouth happened right after I consulted her if Kaffae can be used for a couple's thing - to know what your boyfriend is reading online. As someone coming from an engineering background, I have to admit this is the power of sales. Setting the right framework helps people see who to share to.
There's nothing that feels better than the someone talking about my app because they are genuinely interested in it.
It's been a nervous experience cold messaging on Twitter. After 50 messaging, I've got a first reply to try it out. More than anything, it was about context. The person was talking about needing to make a chrome extension to prompt the user because the complaint was about mindless browsing. This was the exact solution for that. I want to find that situation more. Maybe "browsing" or "mindless". The problem is that context is hard to show up on keyword search. That is homework to come.
Comment