I think you guys need to rewrite your pricing page, at least tell the user what a data point is
If you send less than 20,000 data points per month,
Mixpanel is free.
If you will be sending more than that, we have a tiered
pricing model:
Contact us for high volume pricing.
* $0.35 per 1000 points - first 50,000 points sent
* $0.25 per 1000 points - next 150,000 points sent
* $0.15 per 1000 points - all points after that
This pricing includes data storage, analysis, and
bandwidth.
A little calculator tool would be nice to help me ballpark a price. That, or show some canonical numbers. "ABC company has X daily uniques who frob an average of Y widgets each day. That would cost only $123".
Remember: you guys are the experts on analytics, not us. We might have no idea how many data points we have now, or will have when we start using your service. Maybe consider a simplified pricing scheme as well.
Anyways -- this looks awesome. Google Analytics covers a lot of stuff for us without having to think. When we've got a spare moment to think about what else we want to track, we'll definitely give this service a try.
You may want to consider rethinking the metered billing model. Geeks love them, particularly if the algorithm to calculate the price is spiffy. I have yet to see evidence that people who pay for stuff like them.
Here's an anecdote for you: our industry is going cloud cloud cloud right now and I am pushing it internally. Enthusiasm among management for test projects is high -- I'm presenting one today, actually.
(I rewrote an internal app on the App Engine. It took two days for a prototype, will take two more for getting it production-ready, total cost and schedule less than 10% of when we had more senior, more talented engineers implement the version we are actually using.)
So in this process of "Hey, we don't have to actually host hardware here!" evangelism I actually tried to get a purchase order for Slicehost approved for a different project, where I said something to the effect of "It is going to cost us X,000 yen when I'm in development, and X0,000 yen when we move it to production, and Y0,000 yen if this app gets really popular internally."
This purchasing request broke my manager.
He told me that he'd rather get a request for ten times the highest price I quoted, than try to figure out how he was supposed to incorporate conditional logic into his budget and reporting forecasts for this quarter.
Are you sure you don't want to be selling my manager the low-cognitive-pain, easy-to-buy, ten-times-more-expensive version?
$0.35 per 1000 points - first 50,000 points sent
Boom, you just broke my manager. Not only are you requiring forecasting and conditional logic from him, you're requiring math.
Small Business: $20
Web Application: $100
Enterprise : Call Us
There, now we have an unbroken manager. I ask for $100 a month (not because we need it but because my manager won't approve the Small Business so I won't bother asking), he asks "Are you sure we aren't Enterprise?", I say "Yeah, I did the math", he says "OK, approved."
[Edited to add:
Incidentally, in my own business I'm a stats junkie. There are currently three types of tracking scripts on my pages, plus my rolled-my-own bits, and you very well might be #4.
That said: I pay $19 a month for CrazyEgg, which has given me some very valuable insights into user behavior in the past, so valuable that I continue paying for it on a monthly basis despite not actually using it in most given months.
At the moment, for example, I think I have no experiments actually running.
Now, hypothetically if I were already using MixPanel and savoring that time you made me a thousand dollars by increasing my conversion rate, but didn't currently have an experiment running, you might not be making money from me. That is sort of suboptimal for you, isn't it? I'm willing to pay you for nothing if you have previously delivered wins to me and if I think I might get another win at an unspecified point in the future. This dynamic has earned CrazyEgg about $250 of free money from me. Do you want free money? ;) ]
You may want to consider rethinking the metered billing model. I have yet to see evidence that people who pay for stuff like them.
How about Amazon's cloud services? Lots of people pay for those.
Your story is sad but it sounds like the problem might be more the relationship between you and your manager rather than a problem with all cloud servies.
Quick plug for zuora.com, started by ex-salesforce people (I'm not connected w/them). They make cloud billing models easier for purchasing to swallow. And yes, they're a SaaS provider themselves.
How much can they scale? And how much historical data would be stored? I am working on crunching huge weblogs right now and looks like they have already solved the problem (on a smaller scale? The website doesn't talk about traffic that they can handle)!
I think the original poster is interested in your service, but wants to be confident he can trust it.
"millions a day" builds confidence but "mostly under control" puts it right back here we started. Maybe a little bit more detail would put him (and others) at ease?
The reason I said "mostly" is because it's naive for anyone to think they have things perfectly under control. No scaling problem is always solved, there's always things that show up. We have it largely under control and are very capable of handling it.
A good way to end people's fears of your scalability would be to introduce an individually licensed instance of your service and inviting the client to keep things under their own control (if not in their intranet, then on a dedicated host under their control.)
Won't this involve handling over the code itself? I am toying with this idea for my startup but the fear of code getting leaked has prevented me from fully embracing it.
Depends on what their code is written. I handed out a lisp binary and an image dump; I update it once every 2 weeks and I send them a single .core file as an attachment.
If you insist on exact version compatibility, I am sure you can dump images from most mainstream dynamic languages or ship intermediate "compiled" object code.
There is also the option of getting them a standlone host and billing them extra for its monthly costs. Failing that, there is always the law.
Most companies just want to improve their business processes and use tools to make them more profitable in their line of work; just think of how much software you use, and see how many of your vendors do you feel like competing with. If anything; quicker money is made from asking the vendor for a reseller license and agree on a partnership and sales commission.
I've been using mixpanel for a few weeks now -- nice software. The integration (PHP, Javascript) was very easy and the data presentation is easy to understand albeit still a bit rough.
ditto. Adding Mixpanel tracking was incredibly quick in both Javascript and Ruby, and setting up basic funnel stats is far simpler than with Google Analytics. They've also been really responsive to suggestions.
we've been using mixpanel for our facebook apps since launch and it's helped us greatly, and our entire team loves it. even the pm understands the stats without asking questions (which is usually not the case).
understanding how a user uses the application has not only helped us figure out what features the user wants, but also which of those actually WORK (in real time).
