Despite the strain on resources and energy, the holidays are the most wonderful time of year for data-driven eCommerce professionals. High traffic and large purchase sizes offer rich customer data for robust analysis.
However, three challenges retailers face during the holiday season (and, in reality, all year round) include:
- How to extract meaningful, actionable insights from analytics data.
- How to ensure data is accurate for decision-making.
- How to meet deadlines for development work to capture data.
Consumer insights don’t just jump out of data. To extract insights from data, you need to apply a methodical process to coax out data insights—and verify them for quality. The digital analytics framework, a series of steps designed to guide the process of pulling actionable insights from digital data.
Use the following steps to keep your analytics running as efficiently as possible when getting ready for those data-heavy holiday events—and all year round.
- Define business goals and questions
- Determine KPIs
- Conduct a site audit
- Develop a tagging solution
- Implement QA and deploy the code
- Create dashboards and reports
- Analyze the data
- Make recommendations
- Test and optimize
- Repeat the analysis, recommendation, and testing steps
1. Define business goals and questions
You will have a hard time enacting a measurement strategy if you don’t know what you’re looking for. Start by outlining the purpose of your website, then building business requirements and KPIs.
The Purpose of Your Website
The whole reason that you have a website is that you want to sell more stuff. That may sound a bit crass, but it’s the truth: for retailers, analytics is sales optimization.
As such, revenue will be your most prominent KPI, or key performance indicator. In addition to revenue, however, you will have other KPIs that map back to your business requirements. Before you can define these KPIs, you will need to outline your business requirements.
At a 20,000-foot level, your business requirements are the answer to the questions “How are we going to accomplish our purpose?”, “What features are we including on our site?” or “How are we building our site?”
When you hypothesize about and then adopt new methods to improve site performance, then you are creating business requirements.
For example, as a retailer one of your business requirements might be to use an algorithm to recommend appropriate content based on search history. This requirement is meant to increase revenue by grouping purchases.
2. Determine KPIs
Once you have outlined and implemented your business requirements, you will be able to build a measurement strategy to measure the success of those requirements via KPIs.
KPIs for the above requirement would be “purchases of recommended products” or “revenue generated via recommended products.” These metrics will help you see how effective you are at meeting your business requirements.
A performance indicator is some sort of measure; it’s a metric. It is a number.
The concept is rather simple, but sometimes people don’t really understand what a key performance indicator is, or the power it holds. Here is an example from the Olympics to show you the power of KPIs.
Historically, the British cycling team had won only a single gold medal in seventy years. In 2008 in Beijing and again in 2012 in London, they won an astonishing seven out of ten gold medals in their events.
How did they make such a drastic improvement? Instead of focusing solely on measuring the principal KPI (speed), the team looked to create incremental improvements in other areas.
These improvements included increasing aerodynamics by testing and measuring in wind tunnels.They worked to reduce friction, even painting the floor of the team maintenance trailer white to make dust particles more visible so they could avoid getting dust in their gears.
The goal was to improve all the contributing metrics by 1%. In all, the aggregate improvement in the secondary goals helped to contribute to the key performance indicator, their speed, and made all the difference.
In the digital world, our top key performance indicator is, of course, revenue. However, measuring and analyzing other metrics—like number of conversions, number of conversions per visitor, number of product views, or number of items in the basket per purchase—can help companies make incremental improvements to win the gold.
If you need to come up with some good ideas on how to change your KPIs or how to make KPIs out of your business requirements, you might look at the Digital Analytics Association’s analytics cookbook.
3. Conduct a site audit
After you’ve defined your business goals and determined applicable KPIs, the next step is to conduct a site audit.
The goal of a site audit is to assess the overall state of your site’s analytics with an eye towards improvement. During a site audit, you want to ensure key user information and interactions are tracked, especially if they align with one of your KPIs.
Often, even if data is being collected, it’s not as organized as it could be, making clear reporting difficult.
For eCommerce, you want to pay particular attention to campaign tracking, engagement metrics, and conversion data.
Taking the time to identify opportunities to improve data collection processes is worth it in the long run. Every transaction is a moment of truth—if you’re not there to capture the data, you missed a key opportunity to collect visitor data.
How to Start a Site Audit
If available, use a template when conducting a site audit. There’s no reason to reinvent the wheel. If your organization has documentation of the site’s implementation—which it should—this documentation will be a good starting point for your template and for your audit.
During a site audit, make sure that you document what you find with each technology, variable or tracking tag. Having a summary and notes for each area of concern is very helpful in thinking through areas of improvement, and sets you up with an outline for the next step in the framework. Some of the main topics that you should always review during a site audit are:
- Which code version is being used on the site?
- Is the site using a tag manager?
- How is the account set up?
- Is there cross-domain tracking in place?
- Is there a page naming or product grouping strategy that makes sense?
- How are campaign tracking, internal search, and site conversions being tracked?
- Is there a segmentation strategy?
- Is eCommerce tracking or attribution modeling set up?
What to Audit
For eCommerce sites, the most important types of data to review thoroughly during an audit are acquisition data, engagement data, and conversion data.
For acquisition data, you want to understand how campaign tracking is set up and if marketing channels are coming through consistently. You’ll also want to understand who manages the campaign governance plan.
For measuring engagement on your site, you will want to ensure that features like newsletter signups, store locator clicks, add to cart buttons, wish list buttons, social media buttons, and product information are all functioning properly.
For conversion data, specifically review the product and order data captured at the time of conversion. Make sure data is captured accurately and that you’re able to segment for key data points, like revenue and repeat customers.
Auditing your site is very important, especially for the holiday season. To paraphrase Susan Vertrees, Principal Adobe Analytics Consultant at Adobe, decisions based on bad data are worse than decisions based on no data at all, including for automated decisions (such as for a remarketing technology automatically targeting past visitors based on what they viewed).
Special Note About Page Speed
During your site audit, you should also pay attention to page speed:
- How long does it take for your pages to load?
- Do they have a lot of heft?
- Can you measure how many bytes and kilobytes your pages are?
Three seconds should be your max load time. If your content is not there in three seconds, a lot of people are going to leave. The expectation of consumers—perhaps especially holiday shoppers—is that they will have a quick website experience.
Ensuring fast pageload speed can become more difficult during heavy traffic events, like Cyber Weekend, so you may need to check this more often than during just a site audit.
Check your responsiveness on mobile devices as well. People are impatient with a mobile device—maybe even more impatient. They will give up very quickly if things don’t load.
4. Develop a tagging solution
The next step is to develop a plan of action, a.k.a. a tagging solution, for any fixes or improvements you discovered during your audit that should be resolved (especially before the holiday shopping season begins).
Documentation from the audit template will serve as a jumping off point for determining a list of what needs to be done. From there, you should determine the priority level of each task and develop a realistic level of effort each task will require.
The amount of time and resources available to you prior to your Cyber Monday deadline will determine which items you can address or which should be prioritized.
Create a Solution Design Reference
To be most efficient with your limited time and resources, start by creating and relying on a solution design reference (SDR). An SDR includes all technical requirements for the development team and any other important stakeholders, as well as information on what should be tracked and how.
After building your SDR, build out your tagging solution, explaining exactly what needs to happen to get your implementation in order.
Building Your Tagging Solution
Depending on your list of tasks to accomplish, you could create one larger tagging solution document that outlines all pending tasks, or you could choose to create a set of documents, one for each major implementation task. This document will be sent to the development team to implement.
Note the difference between a tagging solution and an SDR. A tagging solution is a plan to implement new business requirements, whereas an SDR is documentation of all tracking technologies on your site.
Keep in mind that development teams may not be as familiar with the items on your list, so be sure to provide specific and detailed examples of what you’d like to accomplish.
Again, use your site audit documentation to outline how the data should be structured and exactly what should be collected.
Be specific. Even seemingly small things like using upper and lower case text interchangeably or slight structural differences will have an effect on how data is captured and can make reporting much more difficult down the line.
To give an eCommerce example, if you discovered during your site audit that the add to cart buttons were tracked inconsistently across the site, you would need to determine the best strategy to fix this issue and outline what needs to be done in the solution design reference.
Use your notes from your site audit to determine how to capture this data more consistently. Check the site to see if there is a common ID for add to cart buttons across the site that you can use for tracking. If so, the solution could be to modify the existing implementation based on that common ID.
If the IDs are inconsistent across add to cart buttons, then the tagging solution will need to include creating new code to track the buttons. Make sure that the data for these add to cart buttons matches the existing data. Once you have a solution in mind, create an SDR entry that includes screenshots of the add to cart buttons across the site, a detailed explanation of what should be tracked and how, including the variables or event data to capture.
5. Implement QA and deploy the code
Once you have your plan of action, and you’ve created the solution design reference for the development team, the next step in the digital analytics framework is to implement QA and deploy the code.
You will need to be as efficient as possible in order to meet your Cyber Weekend deadline. The solution design reference will provide a good start.
Test Your Tags in a Staging Environment
Use a staging site for the initial implementation and testing. If you don’t have a staging site to use, it’s still possible to test the code using a tag manager feature or an extension, but it is best practice to use a staging site when possible.
Use Tag Validation Tools for Quality Assurance
Using automated or semi-automated tools, like ObservePoint, ObservePoint TagDebugger™ or Charles, can be very helpful. Be sure to document your testing results clearly, so that you can tap any bugs back to the development team and it will be very clear where the issue lies.
You should have a plan in place for how code will be published to the live production environment. This process can be as simple as deciding who on the development team will be in charge of publishing tags, but it’s good to nail that detail down, so there’s no question of who to contact when the time comes to publish.
Identify Unique Testing Use Cases
Lastly, keep in mind any special circumstances that will need to be considered for testing. In order to accurately QA the code, you will need to recreate the circumstances where that code fires.
For example, in eCommerce, you should be able to test your analytics are firing on purchase. You will likely need a dummy credit card to test on the staging site and a way to make real purchases on the production site.
Implementing, Testing and Deploying Tags
Now, let’s take a look at the process for implementing, testing, and deploying tags.
- The development team creates the code and communicates with the QA team that the code is ready for testing in a staging environment.
- The QA team tests the code and documents the results.
- If the code passes:
- The QA team communicates back to the development team that the code is ready to be published to production.
- The development team can publish, and the code should be tested again in the production environment.
- If the code fails:
- The QA team must add what needs to be done to fix the bug and let the development team know that there are fixes to be made before the code can be pushed to production.
- The developer then makes updates to the code, which starts that process over again.
Being able to communicate clearly both in the solution design reference and the QA documentation goes a long way in eliminating back-and-forth between these steps. Take the time to document thoroughly; it is definitely worth it.
6. Create dashboards and reports
Once your implementation is in place and your data collection machine is running smoothly, create dashboards and reports for your most relevant KPIs. Focus on the KPIs you came up with at the beginning of the framework to provide actionable answers to pressing business questions.
7. Analyze the data
Data only has value when used for informed decision-making. Dig into the story behind the numbers to find real insights from the data.
If you’ve followed the preceding steps, you will be equipped with all the tools to see whether or not your initiatives pushed the needle on your KPIs. Use the data to check the validity of the hypotheses your business requirements sprouted from.
8. Make recommendations
Use the insight discovered during the analysis to make recommendations to appropriate business stakeholders.
9. Test and optimize
Build in time for testing and optimization before rolling out any major site changes.
10. Repeat analysis, recommendations, and testing steps
Repeat steps seven through nine, which are to analyze, provide recommendations, and test/optimize. These last steps of the framework are meant to be an ongoing process of using data from your website to provide actionable insights for your business.
Are You Ready?
You should apply the digital analytics framework frequently throughout the year as you release major site updates. Some of the steps may be difficult and tedious, which we understand, but there are tools to help you out.
By combining automated data governance and tag validation with the right processes and personnel, you will be able to make the most of your holiday season data.
As Senior Analyst at MaassMedia, Rebecca proactively contributes to the continuous development and enhancement of client tracking, data collection and reporting capabilities. She oversees custom web analytics assessments, implementation and reporting plans across a full range of client engagements for website, mobile app and campaign projects. Rebecca studied neuroscience, earning both BS and MS degrees from Tulane University, while working in a learning and memory lab studying the effects of estrogen on cognition.
About the AuthorLinkedIn More Content by Matthew Maddox