Privalytics.io
Privacy-focused analytics for websites
After my post reached the first page of Hacker News, I realized I had no way of knowing whether people was actually reading the article or just bouncing.
After tinkering the entire weekend, I came up with a solution to track the time spent on each page, without cookies and without risking any privacy violations. It also takes into account open but not on focus pages.
Now I can tell for sure that people spent on average 20 minutes reading my guide on how to write a programming book.
I have added dedicated pages for referrers. In this way you can explore the top landing pages for each referrer, as well as the page that was the source of the visit, not just the domain.
#ShowIH I have finally made it to finish the landing page and MVP. PRIVALYTICS is here. Any feedback would be greatly appreciated.
Even though the registration is open, you won't be able to create websites. Please, contact me if you would like to test the beta version of the service.
-
1
I haven't yet added it to my website, but I agree with @apd. Your biggest benefit even over other privacy-oriented analytics tools like Fathom and Simple Analytics is that you don't store cookies — which is an INSANE PLUS for landings like mine where I don't care about counting returning users (they're like <2% anyways), and I don't want to add a cookie statement popup (which would be ignored for sites like mine — the user would either go to app store or close the page, and I'd get no analytics at all). You definitely should stress more on that.
-
1
The cookie storing is a big doubt I always had. Some people don't need it, some people find it crucial. In my case, for example, I like knowing what articles and topics are more popular, but I don't really care if it is the same person coming over 10 times, or 10 different people.
However, sometimes you want to know how long a user stayed on your website, or the path they followed to reach a special page. This is very important if you sell something, and you need to optimize the path the user follows.
The cookies concerns are noted here. And thanks for the comments, they help me keep an eye on what I need to restructure on the landing page in order to make clearer!
-
1
You can consider cookies as an option later down the road, but not a mandatory thing. Besides, you can totally consider charging more for cookies support — this will be justified by multiple reasons:
- more computing costs
- more value for customers (more insights into visitor behavior)
- a price customers have to pay for something that’s not 100% true to your vision (that's more for you to be in peace with charging more).
Double besides, you can already charge more. You have a USP, which is a solution that won't require a cookie banner — that's a reason good enough, and I think many makers will be fine with paying a ~$10 monthly fee for taking away the hassle of implementing the cookie consent solution and still getting insights into their page performance.
-
-
-
1
Just visited Privalitics. Here's my experience:
-
Read "Privacy-focused analytics." My thought was "Maybe it's like Poverty-Focused Development Assistance which gives aid to reduce poverty." I wasn't sure how Privalitics would solve privacy isues with analytics.
-
Read "Clear and affordable analytics for your website, respecting the privacy of your users."
Me: "No no no. It's web analytics that respect users' privacy." -
Scroll down to the screenshot.
Me: "It's like Duckduckgo for analytics." -
Scroll down to the pricing.
Me: "Why should I use this instead of Google analytics? What's the benefit?"
In case of Duckduckgo, they try to protect (or not violate) their users' privacy. On the other hand, Privalytics offer Google analytics-like services without violating the privacy of the users of their users. The end users won't know if they're not being tracked by web analytics, unless it's boldly stated, like Duckduckgo does. But is it necessary?
Maybe it's not for all websites. A statement like "If you're A, this is why you should use Y instead of Z" would help to convince your potential users.
GOOG'LUCK :)
-
1
Hey!
Thanks for the feedback!Indeed, I need to make a clearer landing page. And as you say, analytics is not for end-customers, but for developers and designers, who do care about the privacy of their own users online. At least in Europe, privacy is a growing concern, and more and more people are considering what is the proper course of action. Just as a simple example, if you work in a larger organization and you are in charge of releasing a micro-website (either public facing or for internal customers) and would like to add some statistics, it becomes a conundrum to include GA because you will need to comply with the broader privacy policies of the company, etc. This happened to me when I had to build the website for the research group where I was working, and that is the reason of including the free-tier (10k visits a month is quite a lot for most people).
I do agree with the feeling, why would you pay 3U$ per month if you have free tools like Google? But that, I guess, goes with what every developer thinks it means having a healthier internet. I do see a growing community of people who are taking these concerns into account, and the idea behind Privalytics is to validate whether it is a large-enough pain that is worth solving.
The other thing is that Privalytics aims at being an open resource, not only the code, but also the process of building a product. My idea, was to develop an open knowledge base for people starting the same path. For, example, the your comments were duly noted here.
-
1
I think this would be suitable for a landing page. To estimate the effectiveness of a landing page, we could compare the number of times the landing page is visited and the number converted to $ or signups.
Having to click OK on cookie pop-up after previously doing the same on 10 previous websites does ruin the user experience. Instead of increasing leads, tracking users' session could potentially increase the bounce rate. However large websites usually do care about their number.
-
-
I have moved the code to a separated repository but more importantly, I have also added a project to do management of the creative process, experimentation, etc.
The landing page looks much nicer. More content can be added, but I think it is enough for a soft launch.
I have played with the dashboard and I think it looks much better now. It gives me the flexibility of quickly expanding it to other sections, such as analysis of referrals, etc.
I have finished the bulk of the website and it is online for some friends to test. I still need to include a payment procedure and finish the statistics on server resources in order to settle on the price.
-
1
This is simple. Seems good. Started using for ihkr.ml
However, the account page returns a 500 error.-
1
It should be fixed now, I hope...
-
-
1
I think you've built something very interesting here.
-
1
Thanks!
-
-
1
Man, this looks just what I needed. Cookie-less is very important to me.
I’m a bit concerned about it being not on-premise though. Technically you’re still a data processor, and I have to specify you in Privacy Policy, perhaps even get end user consent. I know I can probably download the source and self-host it, but my backend is not running Python.
-
1
Hello Paul,
Happy that you saw the update. This is still very work in progress. The definition of data processor only applies to personal data. In my case, I am not storing any personally identifiable information of the people who visit your website, the only personal information I store is about my own clients (i.e. the people who sign up for the service). I am considering adding more options to the code you need to include on your website to give developers full control on the data they send, etc.I am planning on releasing soon, and need some beta testers, would be great to discuss further with you, you can drop me a line contact@privalytics.io
-
-
1
This comment was deleted 7 years ago
-
1
Yes, it does a Javascript post-request to the backend. However, the backend also uses the request itself to analyze the user agent. There are few design-strategies that I still need to settle on.
-
-
1
This comment was deleted 7 years ago
-
1
Hello!
As you mention, at some point there is a degree of trust, since you can't really know what I do with my own server. I could be logging IPs, etc. The idea is that I am also bound by the terms of service, it's a two way street.I have consider dockerizing the website, but first need to clean up the code which is a bit of a mess and improve the design. Matomo offers a very complete solution for self-hosting analytics.
In this particular case, I still need to find the product-market fit. So, let's go step by step and see.
-
Design is not my strength, and I really dislike all the things associated with making a cute website.
I have been accumulating data regarding server load and everything looks very promising.
I have also been playing around with automating the deployment, but Fabric upgraded and most tutorials are out of date.
I have developepd a very simple model for handling post-requests from a website with ~800 visits/day. I can only say that Django is great. Got it set up in half a day of work. Now I need to focus on the front end to display the data that is being acquired.
Today I decided to start a project right in the open. The product is simple: Analytics software with a focus on respecting using privacy.
I will build everything from the software, to experiments, to thoughts directly on Github. You are free to join me in this journey, let's learn from mistakes, celebrate successes, and create real-knowledge on how to build #lean #startups.
8 Comments
Your product looks nice but I have to say that I think you're really undercharging
I agree, I wouldn't trust sending the data of 1M users to a $6/mo service. I mean, even the server to host/process this data would cost more than that, what's the catch?
Processing the data is an incredibly efficient processes.
Let's do some math: 1M users means on average 1000000/30/24/60/60=0.39 visitors per second, assuming 6 page views per visitor, that gives 2.3 hits per second. My current, single-thread, non-optimized (i.e. too many queries to the database) script, on the cheapest digital ocean droplet can consume around 100 hits/second. This means, my server costs of 5U$/month are my starting point for charging.
By the way, I am building the project completely open, not just the code, but the entire business so any discussions are always welcome (also code contributions). I am not trying to make a living out of Privalytics, but a tool that both satisfies me, and makes a healthier internet.
The server is already full with users (non-paying), and none with 1M visits/month, and I had to close new sign ups while I evaluate whether it is wise to scale up the infrastructure. The 1M mark was left there, but I think I want to target low traffic websites with a very accessible plan (many personal websites, or blogs like mine, etc.)
It's great that this is a side project but I think you still shouldn't choke its potential by doing this. Server costs are usually cheap but in the future you'll probably need to hire and pay yourself etc. What if a bigger company is thinking about HIPAA and you have to hire specialists, pay for data centers etc. Having the money available allows some of these things to happen.
If someone wants to host, you're basically making it free for them to get all your work since it's open source. However, if they pick you to host for them, you should charge a premium that will enable you to potentially pick this up full time. Of course, this is just my opinion and you have the full authority to decide how you may since it's your project.
I totally understand what you say. However, I don't want it to be my full-time focus. I'm actually the co founder of a high tech company that builds devices to characterize nanoparticles. I just have a broad range of interests, and Privalytics is a way of keeping me sharp with some web technologies.
My target (at least spiritually) are not the people with 1M/month, but all those out there who have a personal blog, a side project, and want to learn how to serve their audience better.
For instance, my websites generate around 50k impressions a month. I wouldn't pay for analytics more than for hosting, which for 50k is 5u$/month. (and even for 500k would still be 5/month, but I assume with those numbers, the websites would be generating much more revenue).
Anyways, I appreciate your feedback. It is something that is always around my head. Choosing what to maximize, including personal development.
Woohoo!
20 mins avg on guide? Fantastic. Means they're getting good value from it to spend that long reading it.
Are you running ads to the guide or simply organic reach?
Indeed, I added the feature because I was not counting bounces, and wanted to see the real numbers. But they came in as a big surprise for me! I reached out to some readers and they told me they actually study the guides. Let's say it is hit or miss, if they stay, the stay for long.
Everything is organic, no ads. By the way, the stats you are seeing are for Python for the Lab.
That's amazing. You're obviously doing it very right. Congrats! 🙌