Integrating Marketo lead database and Google Analytics

Did you tag the leads in some way that GA would know about them?

I'd be curious to know how to do that.  Would that be a backend mapping between munchkin and GA tag?

@Ronald R You'd probably want to use a GA Custom Dimension to separate known from anonymous leads.

ga('send', 'pageview', {  'dimension0':  'known'});

However, there is a inherent problem.  How are you going to know to set this dimension to "known" or "anonymous"?  That is, how do you know at a given point in time whether a Munchkin cookie is associated with a named lead?

Hi Sanford,

Thanks for jumping in.

I don't, but don't web activities recorded in the activity log of each lead come from munchkin?  If so, munchkin can tell whether the clicks or views were performed by any leads in the database, so can this information be mapped to GA?

Regards,
Ronald

Ah, there's the rub.

Yes, the Marketo servers know whether or not a Munchkin cookie is associated with a named lead, and the identity of the named lead if so.  (Web activities are always mapped to a lead "in the database" -- the difference is whether that lead is anonymous or named.)

But on the browser side, there is no published way of simply interrogating whether the current cookie is associated or not. You can record when an entry page contains a tracking token from an email link... you can record whether a form was successfully submitted... and you can log when the Munchkin associateLead() function was called (three major ways a cookie becomes associated) but even if you make the technical effort to record those actions and set your Custom Dimension accordingly, you aren't getting a true/false result but at best making a good guess. (The reason it's still a guess is that those methods will if appear to "succeed" even if they don't really succeed in associating: for example, as we ran into at our own company, associateLead() won't kick back an error if the email hashing algorithm is implemented incorrectly.)

However, there may be an undocumented way to quickly get a true/false on this that I have been working on. I would rather not get into it in the public forums.  If you would like to help me with testing it please contact me directly.
 

This might not be the most tech savvy solution, but you could tag every link you use in Marketo with Google URL Campaign tags and then analyze them in GA using those tags. This would also allow you to create a view in GA that would exclusively show you traffic from Marketo.

Here is how to use the Google URL campaign tags:
https://support.google.com/analytics/answer/1033867?hl=en
 

@Yoav G You don't need to tokenize URLs from Marketo emails as those links will already come in with the distinct _mkt_tok string iin the URL. That is the first note I made about how you could identify probably-associated traffic ("You can record when an entry page contains a tracking token from an email link").

Even if you accepted all _mkt_token-ized links as starting or continuing a named session, that is only one of multiple ways a session can become named. There are also form fillouts and API associations. Depending on the environment, the leakage (missed named sessions) could be huge if you only consider one re/start point.

Still, the defining problem (pun intended) with this kind of reporting isn't technical, it's semantic: What exactly is "Traffic for a Marketo lead?" Is it traffic after a session becomes named but not before -- even through the same human user made those initial pageviews -- or should anonymous traffic from an eventually named session be backfilled as if it were also named?  If your gut answer is "only after explicit naming," what about unnamed traffic in browser B that occurs chronologically after named traffic in browser A, and later in browser B's session you discover it's the same lead from browser A?  (This is the exact kind of behavior that Kenshoo tries to track, and which I'm sure most of us have created ourselves by opening a marketing link in one browser, getting interested in a product, and then "organically" reopening the site on another device or browser a little later.)  It's not a simple formula.  And you know how much it sucks to pull up a finished report and then have an exec ask why a certain data point is/is not included!

@All - If you want to make such cases easier to solve please upvote my idea: https://community.marketo.com/MarketoIdeaDetail?id=08750000000KMBiAAO