PPC-Advice.com

Despite the recession and tough economic times, there still seems to be a huge demand for analytics talent in the industry, especially when it comes to individuals well-versed in implementing and supporting Omniture.  You have to hand it to them, they’ve become the defacto standard in web analytics through incremental in-house development of SiteCatalyst, and also through strategic acquisition of names such as HBX and Visual Sciences.  But before you go rushing into signing a long-term contract with Omniture, here’s a few things you should know.

First a disclaimer, before folks in Utah start calling me:  This is meant as a guide to companies interested in signing with Omniture.  In no way, or by no means is it a knock to Omniture sales personnel or consulting group, it’s just practical advice for those companies that want to be diligent.

Understanding the Products

The primary product that all other products stand upon is SiteCatalyst.  Omniture SiteCatalyst is currently in version 14.3, and from a usability standpoint, it’s possibly the easiest to understand thus far.  For the most part, there are very few gotchas with SiteCatalyst, as it provides everything you’d come to expect from an analytics suite, pageviews, unique visitors, time spent on site, content reports, success events, pathing, geosegmentation, and conversion analytics.

Keep in mind, companies can choose to implement as much or as little of SiteCatalyst’s functions as they choose.  Should you have already begun the sales engagement with Omniture, it’s important to clearly outline your current requirements, understand their solutions to your asks, and plot any future expansion of capabilities early in the process.

Due diligence is key.  For instance, one of my clients that implemented Omniture SiteCatalyst and SearchCenter years ago started increasing their spend substantially.  New questions arose such as what cities and countries clicks from Google Adwords were originating from, and how those visitors converted.  This tie-in between geosegmentation and paid search visitors required extra processing on Omniture’s server-side that wasn’t originally outlined in their solutions document.

Another example.  Previous HBX customers didn’t know how good they had it.  Prior to Omniture’s acquisition of HBX, it was possible to find deduplicated unique visitors (by default) across many of the reports HBX provided.  Many customers that are migrating to Omniture SiteCatalyst are shocked to find that if they had not originally requested daily, weekly and monthly unique visitors across specific props and eVars, it simply would not be available.

What does that mean?  It’s important to know what you need from Omniture ahead of time.  If you need advanced correlations or subrelations available in the future, you need to implement that right away, and be sure that your solution design reflects the full functionality required.

Ongoing Support and Consulting

The consulting group at Omniture’s HQ in Salt Lake (Orem, actually) are a smart bunch of people.  They will be your key resource during the implementation, testing, and validation phases.  The more time and resources you dedicate to Omniture Consulting the better.  As is often the case, companies may tend to believe that more chefs in the kitchen is a bad thing.  Quite the OPPOSITE!

Get everyone involved, from executives to front-end developers, so that no rock goes unturned.  Issues that seem trivial after the consulting engagement can quickly evolve into obvious blindspots six months down the road as serious load starts pushing through your solution design.  If you don’t sign up for continued consulting hours, make sure that the solution provided is easily scalable and takes into account all the off-the-wall requests your analytics folks may encounter.

For example, should you find your original design solution is just fine right now, but you’re struggling to figure out weird glitches in data that’s collected, there might be a number of things going wrong:

1. Your collection code isn’t working the way you thought it would.
2. You’re pulling the wrong report for the data you want to retrieve.
3. Data just doesn’t look right, in a very wrong way.

Unless you have a team of javascript and front-end developers on your team, finding collection code problems can be pretty difficult.  Especially if you’re relying on a third-party content management system that is designed to automatically populate props and eVars.  Omniture ClientCare is the first point of contact in diagnosing these types of issues, but Omniture Consulting should’ve seen these problems long ago.

Pulling data the wrong way is also prevalent with SiteCatalyst.  It’s important that your analysts understand SiteCatalyst and how correlations and subrelations work, if they’re enabled in your solution design.  Obviously, training is one solution, but support and consulting is another.

And finally, data that just looks ultimately wrong in a very wrong way, such as negative pageviews (as an extreme example) is pretty tough to diagnose.  A lot of developers will be pretty comfortable diagnosing collection problems using FireBug for FireFox or Charles for Internet Explorer, but what if that collection problem occurs on Omniture’s end?  Consulting is really the only team that can help diagnose very granular problems such as these.

Omniture ClientCare aren’t a group of slouches either, but there is a limit as to what they can diagnose and resolve.

Advanced Reporting Requirements

As is often the case, no matter how much due diligence is dedicated to implementing software, there are always gotchas in the end.  There will always be some wise-guy in the company that will ask for impossible insight that just isn’t possible in the interface – by default.  That’s when advanced tools come into play.

First up, the Data Extract. Data Extracts are useful when digging deeper into reports.  Instead of downloading an AJAX chart and exporting to Excel or PDF, just configure what you need in one screen without loading time for pretty graphics or data pulls, get what you need e-mailed to you in a CSV, and do what you want with it in Excel.  You can even schedule data extracts for delivery on a routine basis.

Limitation:  Data Extracts only process for 4 hours, so should your report be more process-intensive, you may have to rely on another tool.

Omniture Excel Client. A fantastic tool that I reviewed in greater detail in a past post.  Pull all your data directly into Excel, manipulate it as you wish, calculate new metrics based on standard success events and props, and build custom graphs that lend a lot more insight than the web interface would ever afford.

Limitation:  Omniture seems to have halted development of the Excel Client for Excel 2003, as many functions that are still available for use within the plugin appear functional, but really do not work.  Make sure your analytics folks are setup with Excel 2007, and you should have little to no issues.  Advanced correlations and subrelations can time out after 4 minutes, so filter requests as much as possible, or schedule delivery of “Bookmarked” datablocks that process ahead of time (4 hour timeout).

Data Warehouse.  It’s probably one of my favorite tools in SiteCatalyst because it does away with all the fancy (and sometimes useless) user interface, gets down to brass tacks, and gets serious.  Want to correlate and subrelate to your heart’s content?  Want to pull hourly granularity with ridiculous amounts of detail?  Want to investigate geosegmentation versus success metrics versus time of day?  Almost anything is possible with Data Warehouse.  Set up advanced segmentation, pick your props and eVars to investigate, subrelate and correlate them on a whim, and wait for an answer.

Limitation:  Answers aren’t always immediate, as DW requests can take up to 3 days to process and deliver, or you might get an empty report depending on your segmenting rules.  These are better for long-term ad hoc reporting insights, rather than routine data pulls.

Discover. Discover is an entirely different beast.  It is designed to be a realtime processing giant of epic proportions.  Think of it as a realtime reporting interface that sits on top of Data Warehouse.  Whereas Data Warehouse requests can take up to 3 days to deliver, Discover is near realtime.

Limitations:  Discover relies on the same pre-processing engine as SiteCatalyst, therefore the same latency that SiteCatalyst experiences between recieving image request server calls and processing those calls into invidual reports still exists.  Typically, this isn’t a huge deal-breaker, but it’s something to be aware of.  It’s also really expensive for what I think is still a product in beta.

Conclusion

Dancing with the devil on a pale moonlight is easier when it’s the devil you know.  Okay, that’s not fair, but it sounds cool.  Omniture isn’t the devil.  As in any software engagement, it’s important to know who you’re dealing with and what you’re getting out of the solution.  I hope this post has gone above and beyond in some respects, so that executives out there can make informed decisions and feel better about the prospect of signing on the dotted line.