Updated: Feb 26, 2013
I'm frequently asked about the big three web analytics tools: Google Analytics, Omniture, and Coremetrics. While the question is frequently posed as "which one is the best?" I also hear "how are they different?" So in that spirit I'm putting them head to head and giving my honest and quite comprehensive review of how they do things differently, in the hope that it helps people both choose the right tool and, for the curious, understand how they vary under the hood. And, so I don't disappoint, at the end I'll pick a winner 🙂 I've used each of these three tools extensively so all thoughts come from my own experience.
These are the categories I'll be evaluating:
- Admin and Implementation
- Data Warehouse Integration
- Traffic and Campaign Tracking
- User Interface
- User Community
Note: I am primarily comparing GA (free version), Omniture SiteCatalyst, and Coremetrics Analytics. Where applicable, I've noted paid add-ons like GA Premium, Omniture Discover, and Coremetrics Explore with a dollar ($) sign. Note that, according to IBM, access to Coremetrics Explore is now built into the standard Analytics package, though there are still strict usage limits unless you pay.
ADMIN AND IMPLEMENTATION
Note: I'm putting implementation in this category just to give it a brief overview. It actually permeates many of the other categories.
Coremetrics uses several tags that are preset to collect different types of information (e.g. a "registration" tag that has slots for user info). Can add element tracking (similar to events) or extra attributes (similar to custom variables). Has special real estate tags specifically for tracking clicks on a page, which is a neat feature that is unique to Coremetrics.
Implementation difficulty out of 5: 3
Omniture uses different types of traffic (sprop) , ecommerce variables (evars), link tracking, and event tracking to pass in information. Nothing is assumed and everything can (and must) be configured. Code plugins can be added for extra functionality.
Implementation difficulty out of 5: 5
GA provides different preset tags (page view, ecommerce, event) to collect different information, though event tracking is a much more flexible bucket that replaces several of either Coremetrics and Omniture's tracking methods. A lot can be set up in the admin interface without implementing code, e.g. onsite search tracking, which is a great feature not present in Omniture or Coremetrics.
Implementation difficulty out of 5: 2
User management has 2 roles, user and admin. Can assign user groups to limit access.
User management at different levels. Can assign report suites to limit access.
User management has 2 roles, user and admin. Can assign profiles to limit access.
No concept of rules
Can apply IP filters in interface or implement VISTA rules
Enables users to implement rules (called filters) from the admin interface to transform data before it hits the reports
For better or worse, menu layouts can all be customized. It's a bit like if Excel had a different layout and options for every company...
DATA WAREHOUSE INTEGRATION
Coremetrics doesn't surface a user ID in the regular reports though it can be added into a variable. Can view user ID in the Explore tool and their own user ID gets passed into data feed.
Omniture doesn't surface a user ID in the regular reports but there is no limit on data collection granularity. Can capture a user-level identifier in an evar for use in reports. Can view Omniture's own user ID in the Discover tool ($) and it gets passed into data feed.
No limit on data collection granularity. Can capture a user-level identifier in a visitor-scoped custom variable for use in reports. Note that this is not against GA's terms of service unless it explicitly contains identifying info (e.g. an email address or username). An ID is fine.
Coremetrics provides clickstream data feeds via SFTP, called the Digital Data Feed. These tables don't contain site promotions or real estate click tracking data, though this could possibly be produced via a custom feed.
Omniture produces clickstream data feeds via SFTP ($). I think these clickstream tables contain all the data provided in the interface, though not 100% sure.
According to commenter Billy below, "the Data Warehouse tool allows you to easily export large volumes of data via email or ftp. The exports can be scheduled via email or FTP to allow for easy integration with other systems."
GA doesn't provide any clickstream tables. By default all data is aggregated and must be extracted via the API, which has some limits on what can be extracted. It is possible to de-aggregate the data and recreate a clickstream by passing in timestamps and other data into custom variables, or by using some of their leftover Urchin functionality to send directly to local servers, but I consider these to be workarounds.
Provides Category Definition File (CDF) functionality that allows for categorizing data and adding supplementary attributes after data collection occurs. This can be pretty useful for grouping content together or categorizing merchandise, and for passing additional attributes that can't be collected from the page tag.
Provides SAINT functionality which allows for categorizing data and adding supplementary attributes after data collection occurs. This is mostly used for campaign processing (unnecessary in Coremetrics and GA since campaign data is collected at the source) but can be used to group any kind of information. Especially useful for passing additional attributes that can't be collected from the page tag.
Data can't be added once it has been collected, aside from a late 2012 update to allow uploading marketing cost data. Some similar functionality can be done via filters (rules).
TRAFFIC AND CAMPAIGN TRACKING
Provides 4 pre-set query strings that are appended to the end of marketing campaigns (no extra code needed)
Additional marketing meta data can be appended as parameters for use in Explore reports. Categorization can also be uploaded via IBM Digital Analytics Multichannel ($)
Provides any number of query string for identifying marketing campaigns but they must be configured and assigned in the JS tracking code.
Provides 5 pre-set query strings that are appended to the end of marketing campaigns (no extra code needed). Also, AdWords is tagged automatically so no query strings needed there.
Campaigns must be categorized (in the interface) into a marketing channel to show up in the marketing report. Doesn't let you drill down to actual pages on referring sites that sent traffic without going to Explore.
According to Coremetrics, this can also be done by uploading campaign data.
Omniture has 3 major methods for creating marketing reports: reports based on the tracking code, out of the box reports for search/referring sites, and reports set up via marketing channel rules assigned in the admin interface. Confusing and highly unlikely any of these match.
Traffic Sources report that automatically populates based on the referral strings and query strings.
In standard Digital Analytics package, there is an array of options. Each metric lets you choose between 1st click, last click, shared, custom attribution, or same session combined with 1 day window, 14 day window, 30 day window, or 90 day or other window lengths Confusing setup and rarely used.
IBM Digital Analytics Multichannel ($) provides some more intuitive attribution modeling capabilities.
GA really shines here. The free version of GA provides several Multi-Channel Funnel reports that give a very intuitive way of understanding how users pass between different marketing touchpoints. The paid version of GA provides an attribution modeling tool that lets you quickly view and compare the results of different attribution models ($), which is far and away better than what either Coremetrics or Omniture provides.
3RD PARTY INTEGRATIONS
Several tools available to integrate with 3rd parties like email/ PPC vendors and retargeting partners ($)
"Genesis" integrations available for integrating with 3rd parties ($). This is not automatic and may require considerable configuration depending on the tool.
Very easy (you can do it yourself in the interface) integration with Google's own products: AdWords, AdSense, WebMaster Tools, and Remarketing. GA Premium enables integration with Doubleclick ($)
REAL TIME REPORTING
|Tracks real-time data in Monitor add-on ($). Although it says "real time" it actually includes historical data from the day as well -- this is nice as you can see a trend, but a bit confusing as it fuzzies the distinction between the regular reports and the real-time reports.||Omniture has begun producing real-time reports, called Current Data.||Tracks real-time data in Real-Time reports|
|Availability: Coremetrics lets you create report segments in either the main tool Analytics or in the add-on Explore.|
Time Limits: Segments in Analytics can be either for one-time only in the past (limited to 35 days) or ongoing, as in starting today without retroactive data (for up to 93 days ata time). Segments in Explore are similarly confined, though it can be extended for a fee ($).
Other Limits: Both types of segments are limited by the # of segments and reports available to the account as they are shared across all users. Coremetrics allows up to 2 simultaneous segments in the Analytics, and up to 10 simultaneous segments in Explore. However, when you use Explore, you can only view a limited number of metrics rather than all the reports.
|Availability: Starting from version 15, Omniture SiteCatalyst allows segmentation in the main tool (prior versions allowed no segmentation at all). Segments can also be created in the add-on Discover ($)|
Time Limits: No time limits -- segments in SiteCatalyst are available for any time period.
Other Limits: No limit on # of segments that can be created. Can only view one segment at a time in SiteCatalyst. Can view unlimited simultaneous segments in Discover ($) but then you can only view a limited number of metrics rather than all the reports.
|Availability: Google Analytics has fully integrated segmentation capabilities.|
Time Limits: No time limits -- segments in GA are available for any time period.
Other Limits: No limit on # of segments that can be created. Can view up to 4 segments at a time.
|Over a dozen types of modules, several pre-built dashboards||Many types of modules, can combine modules across report suites which is nice||Fewer types of modules, though a few more became available as of Jan, 2013. Can't combine modules from different web properties or profiles.|
|Coremetrics has built-in "Insights" reports that indicate when metrics are more than a standard deviation away. There are also some automated suggestions for action based on marketing sources with high engagement, products with high sales, products with high abandonment rate, etc.||No intelligence/insights reporting||GA has an "Intelligence Events" report that indicates statistically unusual activity on the site.|
|Coremetrics doesn't have much sampling except for a few pathing reports. In Explore you can also choose sampled reports to speed up query time and avoid using up unsampled reports, which typically have a monthly limit defined by the contract.||I believe SC samples data when users apply segments, but I'm not sure where the threshold is or how to tell that the data has been sampled. It seems like the sampling threshold is very high, which is great (though again I'm not sure how to know this for sure).||GA samples data regularly when users request a non-canned report that forces GA to scan through more than 500k visits. GA's sampling can produce extremely misleading results and is one of its worst features, though they always clearly indicate when sampling has been applied. GA Premium ($) allows for users to request an unsampled report.|
|Out of the box, Coremetrics has pre-built purchase funnels. Clickstream (pathing) reports can be configured that show how users reach and depart specific pages. Funnel reports can also be configured to show how users path through different pages, but these are not retroactive.||Can create clickstream (pathing) reports on any traffic variable, making it possible to path on categories, site sections, etc. Can add events to create a product purchase funnel, which is retroactive.||Can create clickstream (pathing) reports for any page, or groups of pages by using the Visitor Flow reports. Can create funnel reports to show how users path through different pages, but these are not retroactive. Can't add events into a funnel.|
|Reports can be downloaded into Excel format. Coremetrics also provides an Excel API tool that lets you access your reports directly from Excel. Basically you download their workbook, enable macros, and then select the report you want to access. It works on a report level, meaning you can extract any reports that exist in the interface but can't combine data points from multiple reports together or customize them beyond editing a date range. There is also no built-in ability to email these reports.||Reports can be downloaded into Excel format. Omniture SiteCatalyst also has a ReportBuilder tool that lets you access reports directly from Excel. This goes far beyond report-level as you can select any dimension or metric from within SiteCatalyst, apply segments, combine data points together, and schedule reports to be sent to specific people at a specific time each day/week.||Reports can be downloaded into Excel format. Beyond this there is no tool from Google to access reports in Excel, though there are many free and paid 3rd party plugins (e.g. Next Analytics, Excellent Analytics ($)) that provide similar functionality to Omniture's ReportBuilder by hooking into the Google Analytics API. These tools typically let you choose any dimension or metric from within GA, apply segments, and combine data points together. Admin capabilities like scheduling emails vary depending on the tool.|
It's a bit hard to describe the interfaces, so I will show a screenshot of the "location" report for each with some notes. That might be interesting for people who haven't seen it before.
unique feature: Tabs open up at the top to let you toggle between different reports while you work.
unique feature: only Omniture lets users tinker with the menu layout, so menu names may be different in any given Omniture instance.
unique feature: Simplicity. All reports are grouped into just 5 main reporting categories (as opposed to 10-15 for Coremetrics and Omniture). The bulk of the interface is made up of canned reports with pre-set metrics while Coremetrics and Omniture request that users select the reports and then add in the metrics they want.
Marketingland reports that Google Analytics is on more than 10 million websites, according to Google's last earning call.
These numbers may not be totally accurate since I just pulled them off the web, and obviously large websites that use Omniture or Coremetrics will also have a lot more people using the tool than someone's personal blog with the basic GA code snippet slapped on. However, whether the multiplier is 50x or just 5x, I think it's safe to say that GA has a much larger user community than Omniture, and Omniture has a much larger user community than Coremetrics.
Why does this matter? These tools are complex. Having a large community of users is extremely valuable in terms of solving implementation questions and getting help for interpreting reports. Frequently other users are more helpful than the company's own customer support since they know about unsupported "hacks" and workarounds, and can discuss more open-ended questions. GA is the hands down winner on this one. Just check the quantity of posts on https://forums.adobe.com/community/experience-cloud/analytics-cloud/analytics vs. https://www.en.advertisercommunity.com/t5/Google-Analytics/ct-p/analytics_category
Another way to look at this is that there are 63.9 million search results for "Google Analytics", 4.6 million search results for "Omniture", and 200k search results for "Coremetrics". With which tool do you think you're most likely to find that someone already has a solution to your question?
In web analytics, a good user interface and deep, flexible analysis options can lead to a company culture of curiosity and fact finding. The reverse will lead to users who complain, avoid, and/or funnel all tasks through some kind of reporting analyst rather than figuring things out for themselves. This is actually the case for all software but I believe that, unlike other areas, analytics should belong to everyone and not just the "experts". I don't really want everyone updating the website or sending out emails to the customers, but I encourage everyone to jump in, find information about their slice of the business, and act on it!
So, coming from the above perspective, I do think Google Analytics "wins" since it is the most user-friendly of the group, easiest to maintain (least likely to break!), and the most likely to create the above scenario of mass data usage. GA has legitimate weaknesses, at this point primarily around sampling and data warehouse integration, but the majority of its criticisms are, in my opinion, put forth by people who aren't experienced in pushing its limits. It has incredible strengths in terms of segmentation, marketing attribution, and enabling users to investigate ideas on a whim -- in contrast, both Omniture and Coremetrics are primarily geared towards producing reports (and not analysis) without the added functionality of their paid Discover and Explore add-ons.
Obviously, if you're in an organization that only sends data through well-trained analysts rather than the wild masses, your opinion may differ. In that case, still need help figuring out what tool is right for you? One way to look at it is that Omniture operates from the perspective that there are infinite types of websites that are all completely different and unique. Coremetrics operates from the perspective that websites are online retailers. And Google Analytics operates from the perspective that there are content sites and ecommerce sites, and they should all track the same general success metrics. Which perspective matches your own the best?