I don't get this. How is viewing individual customer activity supposed to be useful? Outside of a very few companies, any reasonable level of success is going bring far too many customers to make this practical. It seems like a feature that makes for great marketing copy but is practically useless in real life.
Activity feeds can give you great insight into what individual users are doing so you can test out your hypotheses in funnels and see if large swathes of users exhibit similar behavior. For example, one client noticed that users were scheduling appointments by clicking "day" 20 times rather than clicking "month." They made a UI change to the calendar and saw their booking rates go up.
This is also incredibly useful for providing stellar customer support. As an example, Dollar Shave Club is using Activity Feed to see exactly where a user is, which browser and OS, and what they've done recently. For example, if a user claims to have changed plans, the customer support rep can see what buttons the user clicked, what the old plan was and what the new plan is. They say it has made them much more efficient and effective when dealing with customers.
Sampling. You don't have to follow every individual path but if you look at a few you can draw out common themes that you can then potentially verify with a statistical analysis of all the users (or a bigger sample at least).
Besides tailored customer support interactions, this is useful for characterizing cohorts. You would use aggregates to identify customer groups and interesting business insights -- 'one geographic region produces more order completions than another' or 'low purchase price customers initiate more support interactions' -- and then use something like this activity feed to discover the narrative behind samples from those clusters. The aggregate stats give you the high-level business logic, while this use-case view tells the story behind how those overall traits arise from single user actions.
It allows you to build intuition of the actual singular customer profiles that are using your application. It's ineffective to design for a group. You have to design for a person. So this allows designers and developers to say "Ok - given what we know about Sarah - what is she trying to accomplish and how can we make it easier?". Currently the best way to do this is to hire a qualitative user researcher or be good at it yourself. And it's really hard to get good honest data and enough data to matter.
This is great for lead scoring. Makes it really easy to see who might be on the verge of buying The Expensive Plan and lets you personally contact them for follow up.
I think Mixpanel just shot to the top of list of analytics I'm willing and wanting to integrate, even before Google Analytics (of course it's needed regardless).
A backup and comparison. And I can imagine Google uses it partially to determine value of a website in ratings. I never researched this to confirm it or not.
I'm not sure it's that expensive if you consider the value it can add. I would say you should be actively using it, and having a large amount of experimenting going on while using it. Also, you could discontinue using it after you've done testing that shows the positive results you want - and while not experimenting, then not using it. You won't have long-term or cohort results then necessarily, but I imagine they are targeting high revenue or high revenue potential businesses.
There's a lot of value when a company tries to make it simple to understand, and more importantly to show what's capable and why it's good to do certain things. This is not only for my own understanding, but for when I pass the work off to someone else, I know they'll be able to learn quickly and communicate back well.
Has mixpanel made it easier to keep track of a customer when they transition from anonymous browser to registered user? When I last looked into it, you had to capture the random cookie token associated with that user when they were anonymous, store it in your database when they register, and then present it to mixpanel as identification forever after.
We went with KISSmetrics for a couple reasons, but one of them was that whole process was much easier: just a simple identify() call with an email address, and KISSmetrics will associate the previous anonymous token with that email address automatically, so you have a complete history of that user.
Suhail, I am new to mixpanel but dont you identify a user already via a cookie of yours and all I need to do is create an event that the user has created an account for it to show up in the now new activity feed?
Unfortunately you'd have to not use a majority of the web services you probably use. If its not Mixpanel, its their own internal solution, an alternative etc.
This is essentially just visualizing the logs most companies are already keeping. To be hyper vigilant:
1) Sign up for a unique email address for each service
2) Don't upload Gravatars
3) Read the privacy policy and TOS of the various services you use to see if you are comfortable with what they are measuring
The truth is, this kind of information is easily overwhelming, but will be useful for customer service and product teams trying to identify where users get lost using a feature, signing up, purchasing a product etc.
This data is crucial to determining who is using your product, how they're using, and where users are bailing out. A very simple use case is when you're tracking users that hit your landing page, then hit your pricing page, then leave. If that's too big a percentage of your users, then it might signal that your pricing is off. If it's in response to a certain campaign, then it might be a signal that ad campaign isn't level-setting correctly.
This data tends to be uber-private, and would never be shared outside of the company, but it's also information that could be pieced together from server logs natively anyway (though with significantly more effort). Basically speaking, almost every company collects this data already, this just makes it more actionable.
It's also perhaps worth noting that following an individual user around the site is low value -- what you want to see are aggregates and percentages. Of all the users that make it to a funnel, how many complete? Of those that do complete, what's different about them vs. the others.
For a company the size of AirBNB to track each user individually is prohibitive, but being able to isolate those metrics down into measurable factors is huge.
If your startup wants to do something like this, it's worth noting that you could very easily semi-anonymize the data by attaching tokens to the users instead of tracking by username or email address. You'd still get the same traceability, but perhaps be slightly more able to be comfortable with doing it.
Fact: I'm not going to write an app that you can use (perhaps for free) without getting a little something-something in return. Like knowing what features you use in it the most, for example.
Airbnb has the information that is shown in those examples because that user has given them the info (by registering). If someone is using a website it's reasonable to expect that analytics is tracking every move in order to make proper business decisions. If that user has registered for the application, it makes even more sense to attach their actions to their account so you can track customer lifecycle data in order to improve your funnel and decrease churn.
If you don't want a site to gather info about you, then don't use the site. Or use incognito mode I suppose.
I think by showing the individual being tracked and their path it feels creepier than just aggregated data of all user paths but obviously this needs to be know to generate the aggregates and provide data to refine the site or application.
If you are really bothered browse with NoScript and don't allow MixPanel/Google Analytics etc. Check unknown domains to see if they are advertising or analytics services. This won't stop everything but will make it harder.
Note that anything you are doing server side can leave a trace in the logs if nothing else. Also phone apps can easily track this sort of thing too. I've made opting out of anonymous logging a (cheap) paid option in my app (http://itunes.com/apps/fastlists) because I think the data is valuable so I want to allow but discourage opt-out and it is also their to justify the full package price.
Web servers have kept access logs from pretty much the inception of the web. They're not gathering any new information, tools like MixPanel are just organizing and presenting it more effectively.
The issue is cross-site analytics. MixPanel promises not to do it, but if multiple sites all used MixPanel it would be well within their capabilities to gather data on you as you move about their customers' websites.
They're quite different. Intercom activity streams shows you when how often a user has visited your site and the conversations you've had with them. Mixpanel shows what actions a user has taken.
Got to love the pace of feature rollout at mixpanel. The design seems miles ahead of other analytics tools as well, or any webapp for that matter. tripl's ui also comes to mind.
