This is in essence the "killer feature" of the analytics world, and it is hardly an easy problem to solve.

Before Mixpanel can even attempt to solve that though, they first need to tackle multi-channel attribution. Right now GA and Adobe are the big players who tackle it, and GA only offers dynamic attribution (ie. the holy grail of attribution) in GA Premium (which costs $150k/yr).

If you want a standalone attribution platform you're left with options like VisualIQ, Convertro, Adometry, etc., all of which have recently been acquired by big players in the space.

So yes, I'd love to see Mixpanel answer the "why." However the first step is to truly answer the "where" question and until they add those tools, their model is still fully last touch, and therefore dated.

Of course even GA doesn't use its attribution modeling capabilities outside of the multi-channel/attribution reports, so they still have a ways to go too. The big difference is, those tools are available today for free.

That said, congrats to Mixpanel. They made huge strides in the event tracking approach that GA has since gone on to borrow, and that is what any modern web analytics platform looks at these days because it simply makes more sense.

Absolutely. It's a really hard problem. I think predicting the future data accurately is also really hard though, and I'm not sure how much value is there for end users. That said, I haven't built a successful company and they seem to know what they're doing.

A plug for a former employer but I'd like to add optimine, http://optimine.com/insight to your list of attribution platforms.

They've got an elegant and, as far as I know, unique modeling approach that avoids the first-touch/last-touch/path-analysis pitfalls.

I've heard of them before. Seems like they are largely for bigger players with offline touchpoints. How do they do with online SaaS companies with sales cycles 30-90 days out and a variety of content efforts, social media, video campaigns, etc.?

Also, would I need an ad server in place to properly attribute display or can they handle that on their own?

I think you are right they are targeting large advertisers with significant spend.

It's a unique approach in that they don't need a lot of granular attribution data, but instead use higher level aggregate data. As a result it's easier to integrate with all kinds of media campaigns.

You need to be able to provide impressions over time for your early funnel ads and conversions over time for your late funnel touchpoints. They use some econometric time series techniques to analyze and estimate the impact so you can focus on whats working.

Interesting, thanks for the details. Right now I'd give anything for a dynamic attribution solution that didn't cost an arm and a leg and could be used at the level of volume a SMB deals with.

GA's attribution tools are fun to play with, but at the end of the day they are still static models that don't evolve over time as they collect data.

Heap [1] actually does a good job of what you're asking for.

They have these flows that chart the user's path through your site, so you can see where, when and why they dropped off where and when they did.

1. https://heapanalytics.com/

Isn't that what you being in your space is about? Understanding the "Why"? If someone else can do that automatically, you're no longer needed in that space.

So, cognitive science? The most effective way to do this is user interviews as this is purely subjective. Hints only tell part of the story.

Localytics does that pretty well. In their funnel view, you can see, for each step in the funnel, it shows a comparison of users who went on to the next step of the funnel vs. users who did not, and highlights a diff of their actions.