2 ms·
The whole time I'm reading I'm wondering, "Why did he build a relational analytics system? That can be painful to scale." and "Why did he roll his own analytic
by tedmiston 10y ago
The whole time I'm reading I'm wondering, "Why did he build a relational analytics system? That can be painful to scale." and "Why did he roll his own analytics stack vs using something drop-in like KeenIO?"
The last two paragraphs about signing up for his service instead of rolling your own clarified all that. I'm a big fan of this blog and it is high value, so I don't fault the author for a little self-promotion, but I do wish this would have been mentioned at the top.
Nonetheless it's a solid article and I think rolling one's own analytics system is a fun exercise for the reader, especially when you have time to do so. On the surface it's such a simple schema but making analytics fast at a big scale is challenging.
On Unwaffle - I like that the pricing is based on number of new users per month vs total users. I didn't see the length of the trial period defined on the pricing page (please add). P.S. Trying to load the pricing page in Mobile Safari breaks a little bit — I had to refresh to see it.
- jasonkester 10y agoOne nice thing about SaaS is that scaling problems are almost always a good thing. This was something I actually first built to shake the trial churn out of S3stat, and to be honest there's nothing that would make me happier than to see that trial funnel grow to the point where it starts choking the little analytics ping catcher. "[personal assistant], be a doll and grab a few hundred thousand out of the petty cash drawer and hire a few contractors to scale the analytics server" (Sent from my BennetonShipToShorePhone) As to Unwaffle, right now the trial period is "30 days, plus however long it takes Jason to get around to implementing billing". I'll experiment with 60 and 90 day trials once it becomes an issue. For now, though, it's definitely at the beta stage.