A quick introduction to Critical Mass, we are a full service digital marketing agency with over 650 employees spread across 8 offices and 3 continents.
First a quick introduction of my department, the Critical Mass Marketing Science practice community spans 4 offices with over 25 specialists and a combined measurement expertise of well over 150 plus years in many diverse areas of measurement including web analytics, data warehousing, multi-channel measurement, neuro-marketing analysis and data mining.First though, I’d like to offer a quick definition to those who may be unfamiliar with the term Marketing Science
First though, I’d like to offer a quick definition to those who may be unfamiliar with the term Marketing Science
Let me explain in a slightly different way. Marketing Science analysts use data from many different sources. These data sources do include web analytics data from systems like Google Analytics, Omniture.. webtrends... Even Testing and marketing systems like the adobe online marketing suite...But they can also include...CRM systems like Salesforce, databases and data warehouses... Even Point of Sale systems...Then we take all that data.... And analyze it... In order to...Measure the performance of websites,microsites, advertising and more...Find opportunities to optimize...Improve marketing campaigns...Delivery highly customized content which increases conversion...Improve the customer...experience and flow...Reduce costs and...Improve salesAnyone who’s been involved in a measurement program knows that the combination of technology, data and people can be very complex and challenging to do right. We here at Critical Mass Marketing Science deal with a variety of clients across many different industries. Clients like Nissan, HP, and addidas are all very different and all with complex measurement needs. As varied as they are, our clients do experience some consistent problems, and we’ve helped many of our clients overcome themToday we’ll cover the three biggest challenges, unreliable data, complex reporting needs, and the lack of action from analytics reports
Measure the performance of websites,microsites, advertising and more...Find opportunities to optimize...Improve marketing campaigns...Delivery highly customized content which increases conversion...Improve the customer...experience and flow...Reduce costs and...Improve salesAnyone who’s been involved in a measurement program knows that the combination of technology, data and people can be very complex and challenging to do right. We here at Critical Mass Marketing Science deal with a variety of clients across many different industries. Clients like Nissan, HP, and addidas are all very different and all with complex measurement needs. As varied as they are, our clients do experience some consistent problems, and we’ve helped many of our clients overcome them.
So what causes web analytics data to be unreliable? There are couple of common reasons, but they both relate to how the coding is done.Most commonly we see analytics code.. often this is commonly called tagging... that is incomplete or wasn’t customized correctly. This can happen for several reasons.In many cases, web developers don’t understand the complexities of tagging and may not customize the tags properly.We also commonly see tracking of flash and AJAX actions being missed, usually because the code gets placed in the main page, but no one takes into account the multi-page form that sits on the page and doesn’t actually change URLs.Often applications are developed by other groups and don’t get tagged, or new domains are created without consideration to tracking implicationsThe end result is that specific metrics and KPIs may not be being captured.The other common problem we see is that web analytics code WAS properly installed, but then stopped working.On complex sites with multiple groups responsible, often a new page or section gets released without codeA personal favourite of mine, a code rollback is done and the tagging breaksEven something as minor as an incorrect file getting released, which may be totally unrelated to web analytics, causes tracking issues.Complex websites demand complex measurement solutions and unfortunately the web analytics systems are often the last thing on the minds of those involved in website changes. Particularly if your agency isn't doing development but internal or off shore parties are doing the work, we’ve noticed that the likelihood of tagging problems occurring increases significantly. We’ve also seen many instances where analytics is not a priority in the IT department and there is no direct accountability in that Tech department for marketing performance, so errors can sometimes go uncorrected for far too long.
Options:Never, it’s been perfectI’m not sure, we don’t really check it regularlyA couple of times, but nothing majorIt seems to go wrong every other week
So, let’s look at our first part of this, when we have web analytics code on our site that hasn’t being installed correctly.A recent technology has now proven it’s effectiveness in solving this issue, it’s called Tag Management Systems, or Universal Tagging....Complex online ecosystems can result in a variety of tags, not just web analytics tags, but advertising beacons, testing tags, affiliate tracking and more. It’s no surprise that in such a complex environment that even slight changes to the website can cause tagging errors.
Tag management systems takes the complexity of tagging out of the developers hands and places it back with the measurement teams. Essentially, one simple tag is placed on the site, usually in some universal component so new site additions won’t be missing it.Then, the measurement teams can control the more complex tagging requirements of the site without involving Tech personnel. Multiple tags can be added, removed, changed or updated without redeploying the site.
Why is tag management such a good thing?These systems use a single line of code in a universal site wide element, and as a result the Tech effort required for installation, QA and ongoing maintenance drops almost to zero. Tag management systems can be used to manage multiple tags, tags like web analytics but also voice of customer systems, ad servers, multi-variant testing systems and more. They take the responsibility of customizing the tagging off the hands of developers and place it back in the hands of the people who understand the complexities the best, the analytics team. These systems work with even the complex web sites, AJAX, Flash, Flex and Air interactions can all be monitored They also allow tags to be removed and replaced without recoding or redeployment. Imagine replacing your entire web analytics deployment in a few days, instead of 6 months
At a high level, here’s how it works.After development has placed the one bit of universal javascript code, the tag management system takes over.The analytics teams uses the tag management system to create the tagging rules that govern what tags will fire with what variables and when.Then, when a visitor comes to your website, the tagging management system detects the page and any actions within and fires the correct tag which then goes on to populate the web analytics systemCritical Mass uses a custom tool like this on one of our biggest and most data-driven clients to effectively manage their analytics tagging installation. Also, we are currently conducting tests on a new vendor tools, including one called “Ensighten” which promises even more effective tag management.
The second part of this tagging problem is that on complex websites it’s very easy to “break” tagging.Fortunately, another new technology is now available. These are called analytics audit and monitoring tools. It seems like a fairly generic name but complex web environments means that more robust quality assurance tools are needed.These systems scan your website and are capable of detecting a variety of different tags. They then record what tags are firing where and with what variables.More importantly, they alert you to pages which aren’t tagged or where tagging has changed. They can even fill out forms and go through shopping carts to ensure some of the most important actions on your website are tagged and reporting properly.There are a few vendors in this space that we have evaluated. One of the leaders is a tool called “ObservePoint” created by one of the founders of Omniture, which we have leveraged for many of our largest clients to provide constant monitoring of their installations.
Why would you use an analytics audit and monitoring tool?Imagine it takes 30 seconds for a human to check a tag on a website and verify the correct variables are firing. Now multiply that by 5000 pages. We’re talking about more than 40 hours of work, not to mention the human error that would creep up. Now imagine having to do that every time your website releases a new version. The time and costs of manual verification can become enormous on large and complex sites. In fact, it generally just doesn’t get done, or becomes quick spot checks which can miss mistakes. Analytics monitoring tools can scan a website in minutes. These monitoring tools can provide you with an overall view of how “healthy” you tagging is and they can also automatically alert you to any specific pages where tagging is missing or has changed in even the smallest way. There isn’t a lot of set up work either. The systems crawl automatically, so as long every page is linked to another page, these monitors can find and log them. Once they are set up, the systems can also run on their own. Audits, reports and alerts can be scheduled to run at whatever intervals are best for your business If your worried that frequent testing might actually pollute your data, not to worry. The systems are designed NOT to fire your tags, so they don’t artificially load up your analytics systems.And of course these monitors work with flash and secure sites, and can also be set up to deal with complex forms and ajax actions
How auditing and monitoring works is fairly straightforward:First you select a main URL or sitemap page that ensures the monitoring tools crawler will be able to find every page. Next the system crawls ever page on the site and records the various tags and their variables For parts of your site not easily crawled, the system can record a series of keystrokes, for filling out forms and shopping carts, and then recording those tags An analyst can then view what pages don’t have tags, and review the tags to make sure they match the information that they are supposed to be sending. Alerts and scheduled scans are then set up so that key stakeholders will be notified quickly when there is an problem.Now that our data is accurate, this brings us to our second problem.
As businesses begin to leverage web analytics data, we’ve noticed a couple of key problems come up regarding reporting needs.Firstly, multiple reports which go to multiple stakeholders and need to be delivered quickly. One of our clients required over 30 reports, each one totally unique, to be delivered within two days! Each of these reports required different kinds of data, and had to be distributed to stakeholders all over the world. Actionability is directly related to on-time report delivery.We’ve also noticed a second problem around data and dashboards being easily accessible at anytime from anywhere, and not just in an email that would quickly be deleted or forgotten. Often, our clients would train several stakeholders in how to use the analytics system, but because it wasn’t often used they forgot how to access the data they needed. Emails with large dashboard attachments often get deleted or filed without even being opened, and executives who are seldom at their desk had no immediate access to important data when they needed it.
Options:One report goes to one stakeholderOne report goes to a lot of stakeholdersA few reports go to different stakeholdersA lot of customized reporting going to specific stakeholders
First, let’s have a look at solving the problem of multiple custom reports that require rapid delivery.Our solution? Deliver the reports in a new and innovative way using web based reporting and data warehousing.We worked with a vendor who specialized in data warehousing and online reporting to develop individually customized reports. These reports were populated with data that was automatically imported from the web analytics systems as well as several other sources. Next, a simple-to-use series of custom interfaces were created where various world-wide stakeholders could access their individually customized reports at any time from anywhere.
Why would you use web based reporting and data warehousing?It’s capable of merging data from almost any source, including web analytics data It can be done without adding additional workload or large projects to you Business Intelligence department, or a Business Intelligence department at all! Like any data warehouse, it can connect to a variety of data sources. Web analytics systems can connect directly to online data warehouse solutions directly. Once the system is set up, it can automatically extract the correct data and populate the various custom reports which can then either be delivered or accessed by stakeholders from a web interface at anytime.
How these systems work is straightforward at a high level. Take the data from wherever it happens to be, extract, transform and load it into the data warehouse, which then populates the customized portal and reports.
The second reporting problem we often see our clients face is that the dashboards and data are not as accessible as they would like them.Even web based reports are often not convenient enough for busy executives as they need relevant web data at their fingertips to help manage their business wherever they happen to be.One thing that stakeholders do tend to find convenient is their mobile phones. Hence this innovation of mobile-delivered dashboards. Web analytics and other data can now be delivered directly to smartphones in a customized app that is intuitive and easy to use. Not only do these tools deliver dashboards, but they also offer some forms of interaction with the data and the ability to drill further in as need be.
Why would you use mobile dashboards apps?- Simply put, busy executives find this method of delivery convenient. When a decision needs to be made, having the relevant data at you fingertips makes all the difference in the world.- Data can be sent to phones from a variety of sources, anything from a simple Excel file to data warehouses and CRM systems.- These tools have been pre-built with mobile users in mind, so tools that allow users to visualize and interact with the data are very intuitive and require little or no training.- In addition, the data is secure and the services are hosted and scalable for any size of organization.There is a variety of vendors which offer mobile applications. Generally though, they fall into two categories, data warehousing generalists, and web analytics specialists. Which way to go is a matter of deciding upon the complexity of the data you want to use, and the power of the interface required. Generally data warehouse apps have more robust tools, but web analytics only versions are simpler to run on a day-to-day basis and may have no additional licensing costs. Web analytics vendors like Omniture are now offering far more sophisticated mobile apps, like the SiteCatalystiPhone app.
In fact, mobile dashboards work in much the same way that the web based reporting and data warehousing does.Data moves from various systems into a centralized data warehouse which then gets transformed into a user accessible dashboard on a smartphone. Generally though, the dashboards are less customizable then their web-based cousins.Now that we have tagging and reporting problems out of the way, let’s focus in on our third measurement challenge...
The focus on web measurement tends to be directly related to the focus of your company culture. Here is the general spectrum of cultures we’ve noticed as they apply to measurement programs.If your company has a culture of advertising, the only metrics that people tend to pay attention to are media related, things like impressions and reach. In the end the company is often just trying to generate awareness and may do little if any measurement beyond the data your media company supplies. We’ve generally noticed this in companies that have low-consideration low-cost products like beverages or chewing gum.If you’re selling a product or service that’s high-consideration but still potentially low cost, like baby food for example, we’ve seen many of our clients with a culture of research. These companies are trying to start to understand web behaviour and how it may relate to product sales. Often there is a need to communicate more complex product benefits or features online and provide comparisons to competing products. These products may or may not be purchased online. These companies may not doing detailed web measurement and conversion analysis, but they may be using the data to understand general trends and to try to inform their bigger web projects. Finally, the culture of accountability is usually the most effective user of web analytics, or at least the most interested in leveraging measurement data. Typically we’ve noticed that these companies offer high cost high consideration products, like cars for example. Their websites may be heavily used by consumers to research their products and at least provide leads to a sales team if not sell the product directly.In our opinion, all companies should be striving for at least some form of accountability using web measurement, even if your product is low cost low consideration. However, if you have a high-cost high-consideration product and you don’t have a culture of accountability along with an advanced measurement and optimization program, you may need to rethink your strategy. Chances are if you generating a lot of reports that no one is doing anything about or no one is held accountable to, you may need to evolve your culture to the next level.
So how do you evolve your measurement culture and start getting some action out of your data? There are a couple of key building blocks that we’ve noticed can be missing.Perhaps you don’t have clear objectives, or KPIs. No one really knows what the website is supposed to be accomplishing, or decisions about your website are being made by the highest paid persons opinion.Perhaps you don’t have the resources. Your tech department may be managing your analytics implementation, but no one really understands it or is leveraging it. Maybe your BI department added web analytics to their practice, but they treat it like any other data source. You have to know what to ask for, and by the time you get a report it may be too late to take action, or the data might not be relevant to the business problem you’re trying to solve.In fact only 1 of our clients has what we consider a proper size analytics team with 28 people working on website, email, social, mobile and paid media analysis. Of course this is proportional to their overall marketing budget, but we often see other accounts having disproportionately low resourcing including their own internal teams. In short, you may need to triple or quadruple your analytics team in order to run an effective measurement program.Most importantly, if no one is watching the site data on a regular basis, then likely no one is looking for ways to improve it.
Options:Not currently doing much with analytics at allAdvertising basedResearch basedAccountable and heavily data driven
We’ll have a look at the Critical Mass Marketing Science approach to solving this set of problems more holistically utilizing industry best practices and our own analytics process expertise. We’ll have a look at these individually, but here’s a quick overview.First, a holistic analytics approach is required. Using data to help improve specific projects is a great start, but in the end measurement affects your entire online ecosystem as well as your organization as a whole. Secondly, we recommend and use a web analytics maturity model to help understand where your company is with digital measurement now, and then create a roadmap for where it needs to go. At the site and project level we use as best practice a process model that ensures that goals and key performance indicators for a project are defined, data is being collected, and an optimization program is in place.We’ll also have a look at how important it is that any analytics team be able to collaborate and integrate with an entire organization in order to be effective.
CM Marketing Science recommends a multi-tier approach to analytics that encompasses not only a particular project, but the entire digital ecosystem and the analytical state of the business as a whole. Being able to compete online means understanding analytics at every level.When initiating specific web projects, its important to leverage web analytics to understand the current situation to help inform the project, clarify exactly what the project is supposed to do, then measure and optimize once the project is in place.At the digital ecosystem level, where your website, your site, social media efforts and your mobile apps all exist, an ongoing measurement and optimization program must be in place. Engaging in a continuous improvement process will allow you to compete.Everything starts at the enterprise level. The first thing to do is understand where your company is at now with analytics and where it’ needs to go. Defining and following this roadmap is critical for success. Critical Mass uses a web analytics maturity model to help lay out this roadmap.
Here’s an example of a maturity model. Full credit to Stephane Hamel, a web analytics guru and evangelist, for developing this model.We here at Critical Mass take advantage of Stephane’s great work as a best practice to help our clients understand how advanced they are with analytics in several large categories: Management adoption, program objectives, the scope of analytics within the organization, the analytics team, the use of a continuous improvement process, and tools and technology integration.As you can see maturing in analytics requires a significant amount of skills, budget, time and talent. Maturing your measurement program requires an investment some executives aren't yet aware they'll need to make. Partially funded, and partially owned analytics programs fail time after time. We can help educate and build these programs plugging in the parts where you are mature, and then working together to develop solutions in areas that need help. Once we have an understanding of where you are now, we can help lay out a roadmap in each of these areas to reach the point where your organization is effectively competing using data and analytics.For websites and online projects, Critical Mass has developed our own best practice process which we leverage on all sites and projects we work on.
While each one of these stages has many individual steps, here is a high level view of the Critical Mass Marketing Science Best Practice process.The first stage is to identify specific goals, key performance indicators and measurement strategies. At a basic level, we need to clarify what business objectives the project or site is supposed to accomplish. Then that information can be used to identify specific KPIs. This can be as simple as driving online sales or as complex as improving online sentiment, but in the end any objective can and needs to be quantified into a set of measurable KPIs.At this point, we will then look at the Current State of performance on your digital properties focused on the identified objectives and KPIs, in order to inform the projects overall and creative strategy. This is a critical part of the discovery process, after which we can complete the measurement strategy. These first steps must be conducted on every project regardless of whether you have your own internal analytics team and are the key to properly informing the rest of the project.The next stage involves designing and implementing the systems that will collect the data. This can include web analytics but also many other systems.Next is design of the reporting systems. While the end result can vary wildly, the one consistent factor is that dashboards and reports are not enough. Analysis is the key factor as understanding where the data comes from and what it means is critical to successful data-based decision making.Finally, training, education and socialization can help our clients develop self-sufficiency if that’s desired.Once implementation is complete, then our goal becomes to enter into a continuous improvement optimization cycle. Ongoing analysis will not only be able to answer specific business questions, but also find opportunities for optimization. This is a cycle of pre-analysis, setting goals and targets, launching and measuring, analyzing and optimizing. Following this cycle using data driven decision making will ultimately improve your digital efforts and your business.
If lack of resources is a problem, of course I’ll take a moment for a shameless plug and suggest you consider the Critical Mass Marketing Science department. We can help evolve not only your projects, but your entire online ecosystem and the enterprise wide adoption of a culture of accountability.As a best practice here at Critical Mass, our marketing science teams are in constant contact with virtually all of our other departments. Analysis is greatly improved when as much background information as possible is available. On it’s own, web analytics rarely tells the entire story of any situation. By leveraging other information from a variety of groups, this helps colour and guide the recommendations for improvement.Analytics teams are most effective when they know almost everyone in the company and should become a nexus of information in order to enable informed decision making within your organization.Critical Mass Marketing Science has helped many clients set up and evolve their measurement programs, regardless of their current state. In our experience, to be effective at measurement, organizations truly need to embrace not only technology, but skilled teams and a culture of data driven decision making.