Segment.io (YC S11) Lets Developers Integrate Multiple Analytics Services
techcrunch.com
techcrunch.com
function track(event, category, options) {
_kmq.push(...)
_mxp.push(...)
.
.
.
edit: Also, I'm interested to see how eager startups are to add yet another $150/mo to their analytics bill when nearly every service starts at that level... While it may not sound like a lot of money, a few analytics engines can add up really fast, especially for a bootstrapped company.
Regardless, I'm a huge fan of what segment.io is doing! I really like their service so far! :)
We're also working on some dependency management stuff so you can include jQuery UI and it'll automatically pull in jQuery, for example.
Basically, if it's going to be me just checking each analytics service in its own tab vs. needing to dedicate hours of development resources to pool everything together, I'd rather just stick with what I have.
for example:
mixpanel and kissmetrics are great for tracking custom events and learning deeply about engagement and conversion inside your app.
but you're still going to want hittail or google analytics to analyze your referrers.
and you probably want customer.io, vero, mixpanel or klaviyo to do targeted email marketing, which is great for improving activation and retention.
so i'd say why wouldn't you want to use multiple analytics services? maybe our definition of "analytics" is just broader :)
Using it with
* Intercom * Mixpanel * Vero * Errorception
I'm not saying standards are a bad thing, I'm saying that it's likely counter productive for most vendors at this point in the marketplace.
The (hosted) product seems especially useful to non-technical staff, who don't necessarily care about standards and protocols.
- As a developer, integrating the various analytic libraries is trivial. Installing it? (npm install mixpanel) is mostly trivial too. Yes, it'll take a couple hours, but it will still take several hours to set up Segment too. (One might argue that's not true, but that's wrong. Even if you include the Segment.io code, you still need to provide a bunch of options and understand very well all the other analytic libraries.)
- As a manager, I don't want yet another 30$ or 150$ / month when I'm already paying that for the other analytic services. Also, I know for a fact that I can easily find a dev to set up all the analytics for less than 200$.
- As an investor, I ask myself: What's their unfair advantage? It seems like everyone could do such a service and even provide an open-source version of it.
So, yeah, I'm a bit lost. I feel it's a big like Grove.io. I really liked the idea and found it very useful. But I didn't see how it'd work as a business. But then, there are hundreds of huge companies that I don't even get their business model. (Twitter?) So, please don't take this comment as a rent, but more as an eye-opener for me to learn about where I'm thinking wrong : )
The main problem with analytics/metrics is that they're all useless - or very close to it. Each of them is horrible about getting data out in order to correlate it with other services.
If segment.io can consolidate metrics data, provide uniform interfaces for getting data in/out, and can start partnering with other services that should be using this data but can't (there are several dozen), they can turn useless metrics data into immediately (possibly automatably) actionable data.
Just a personal vision, and I have no idea if that's where they're goin or not, I just know that needs to happen, and the team behind segment.io has the chops to pull it off.
I also agree that metrics are not nearly as powerful if you can't tie them directly to action. Which is why we built our own, fully featured email solution so you can go from metrics, to important people to engagement automatically.
That's where we think things are headed and it's going to be incredibly empowering.
Hmmm. In what way? We (Chartbeat) provide all of our data through our API, and all of our products are built using those very same APIs.
There are a couple major benefits to this service:
1) Your marketing guy can go in and turn different analytics products on/off without touching code.
2) One call to track() pushes events to all of your enabled services. You don't have to add/remove calls to change which services they go to.
3) The backend code is asynchronous. This is a huge win, because you can call it in code handling a web request without worrying about it slowing down the user experience.
4) You don't have to deal with crappy APIs from companies like Marketo. We were able to remove a couple hundred lines of code that dealt with talking to marketo, in favor of one line of code that talked to Marketo via segment.io.
Assuming that they give you the option to store historical data, they truly are democratizing access to all the data that's out there.
You can start with one analytics solution, and then add / switch to an alternate solution at will. Your data is no longer locked in with one vendor.
If you guys haven't read Peter's piece on Thorium it's worth noting that these guys aren't just about analytics: http://rein.pk/thorium-reactors/
Are there any resources on this topics?
This is vastly different problem, which is mostly about distributed systems and algorithms.
Alex Smola's Hokusai paper is good place to start for analytics on truly massive traffic: http://www.auai.org/uai2012/papers/231.pdf
https://github.com/snowplow/snowplow
Worth checking out if you're too big for e.g. Google Analytics free tier (10m pageviews+events a month)
You can do this server-side, too. Record all of the data, but skim just a small percentage for a high level view of what's going on. You can always dig into the data later for a more in-depth question.
i wonder if they will be able to fix the leakiness problem across all analytics platforms... somehow none of my analytics reports are ever accurate.