matm

17 ms·

121.
Heap (YC W13) wants to bring data science to the masses(heapanalytics.com)
1 points by matm 13y ago
122.
by matm 13y ago
For us, I/O isn't as big a cost factor, since our data processing mostly resides in ephemeral storage where I/O charges don't apply. (Our permanent data backups live in S3.) I/O pricing does apply when using EBS.
123.
by matm 13y ago
Very true. This is basic stuff, and we simply overlooked it. In the future, we might even need to add increased redundancy to our data, which we need to be aware of.
124.
How We Estimated Our AWS Costs Before Shipping Any Code(blog.heapanalytics.com)
49 points by matm 13y ago|6 comments
125.
by matm 13y ago
Yes, we do all our querying on Postgres, and shard per customer. We'll definitely make sure to write a post soon that describes this all in more depth.
126.
by matm 13y ago
Good point. We'll definitely keep an eye on this. One of Heap's most important value props is retroactive analysis. You forgot to define an event upfront? No problem - it's available for you to analyze across your app's
127.
by matm 13y ago
Not currently. But we're eager to learn more about your use case to see if we can figure something out! Care to email team@heapanalytics.com?
128.
by matm 13y ago
Interesting. With regards to the unit of pricing, our thinking was quite the opposite. There's a high variance in value across your userbase, but our previous (per-user) model would price them the same. This was an issue. We tried to m
129.
by matm 13y ago
We already have one. It's part of our enterprise tier.
130.
Build the analytics tool the web is missing – Heap (YC W13)(heapanalytics.com)
1 points by matm 13y ago
131.
Heap (YC W13) wants to make everyone a data scientist(heapanalytics.com)
1 points by matm 13y ago
132.
Heap (YC W13) is building a new approach to analytics: just capture everything(heapanalytics.com)
1 points by matm 13y ago
133.
by matm 13y ago
Great, thanks for the thorough feedback! Re: your first con. We offer a heap.identify() API call that associates activity with one canonical identity. For instance, if a user logs in on a new separate device, you can call heap.identify({han
134.
by matm 13y ago
Huh, interesting! Will keep that in mind.
135.
by matm 13y ago
Yes, first time we saw the user. It counts as two, since by default we identify users with first-party cookies. We offer a heap.identify() API call that can be made to associate multiple devices with one identity.
136.
by matm 13y ago
You can try out the analysis part with some sample (web-only) data at: https://heapanalytics.com/dashboard/demo We let you define events after-the-fact based on the gesture type and the corresponding target variable. Once that event is de
137.
by matm 13y ago
Similar to my comment above: it'd be great to understand what sort of price point works for people like you. Feel free to send us an email at team@heapanalytics.com.
138.
by matm 13y ago
We'd definitely appreciate more data points on pricing - care to send us an email at team@heapanalytics.com, and we can discuss more there? It'd be great to understand what sort of price point works for people like you.
139.
by matm 13y ago
This makes total sense! Manually syncing is a good idea. We need to make sure we have reasonable defaults in place, though, and so far, we've had no issues from our beta testers.
140.
by matm 13y ago
CPU-wise, our SDK adds up to 2% overhead to normal UIEvent-handling. This is with frantic and constant user input - in practice the overhead is much less. Network-wise, we batch user activity and by default make a network request every 5-10
141.
Heap (YC W13) is analytics that captures everything. We’re hiring
1 points by matm 13y ago
142.
by matm 14y ago
"Capture everything" need not imply "measure everything." It should imply "measure anything you need to, immediately and retroactively".
143.
by matm 14y ago
It's quite different. Clicktale and Mouseflow capture a similarly large volume of user data, but they do it at the expense of flexibly quantifying that data. Most of their visualizations - while useful in their own right - are pre-built rep
144.
by matm 14y ago
We initially tried to build this on Mongo, and it was a huge pain. I'll need to elaborate on why at some point, but the very ad-hoc nature of all our querying precludes any non-relational database from being our store (at least any that I'm
145.
by matm 14y ago
Our thinking is that it's an efficient way to grow responsibly while still gathering user feedback early and often. We tried to include as much tangible product as is possible with an invite-only product.
146.
by matm 14y ago
Correct.
147.
by matm 14y ago
This is interesting. Our naive thinking was "fewer numbers" implies "easier comprehensibility", but I don't think the pricing was quite as understandable as we would've hoped. I like your suggestion a lot.
148.
by matm 14y ago
With the current model, your charges would spike. Which is bad. But we plan to offer caps and the option to auto-sample if you exceed your tier.
149.
by matm 14y ago
Feel free to reach us at sales@heapanalytics.com. We'd love to talk about fair pricing, and we certainly need more data points to evaluate what that fair pricing is.
150.
by matm 14y ago
Oh, whoops. Good catch. That should be min, not max.

More ›