Anyone who’s built a home or renovated one knows not having a thorough plan can lead to major cost overruns or compromises later. Let’s say you’re planning to add a suite for your in-laws. Not thought about adding an extra bathroom close by? If you haven’t roughed in the plumbing, slotting this extra bathroom in at the latter stages of the project will be expensive or impossible.
Although no walls or floors are involved on the Web, if the first time you walk through your analysis process is post-launch, you could run into walls of impenetrable data or compromised insight when you finally sit down to measure and assess your Web site or campaigns after implementation. It’s best to think through what you need to measure and analyze when you design your Web site. Build measurement capabilities into the site the first time, not as an afterthought.
To make this conundrum clearer, here are three examples describing potholes that one might fall into if measurement and analysis aren’t considered during design:
1. Analyzing Web site traffic after launching your Web site.
- Scenario: Your Web site has been launched, on time and on budget. You’ve used a content management tool that conveniently dynamically generates page names. Your site has a very flat file structure. Your page TITLE metatags have been optimized for search engines (as they should be) rather than uniquely describing page content.
- You look at your first set of Web analytics reports. They’re utter gibberish! The page titles don’t describe the content and because the page names are generated by the content tool, it’s not clear without calling up each page what’s on that page. Other than total page views and visitors, how are you going to analyze the site?
- What’s the solution? Tag each page distinctly using an extra page id tag (if your Web analytics tool allows this) or override the content tool page names manually (may not be easily done, depending on the content tool).
- What about historical data? If you’re using a logfile analysis tool, you may be able to reprocess the data. If you’re using a page tagging script, you’ll probably have to download and cross reference historical data offline.
- How could this have been avoided? Before you build your site, think through how you’re going to analyze your site and make sure you’ll be able to do so. During the site architecture stage, bucket your content into content groups. Try to integrate grouping into the file structure. Tag the pages as you build. See if you can override or supplement the dynamically generated page names with logical names, logical to business people at a glance.
2. Tracking search engine marketing campaigns, for lead generation.
- Scenario: Visitors click on paid/sponsored search engine ads featuring specials. There’s more than one search marketing ad campaign, each with a different keyword basket but visitors are directed to the same landing page. The landing page has more information about the products and contains a call to action to download a discount coupon and take it to a physical store.
- In-store redemptions for the discount coupon are high. But which ads triggered the most redemptions? Which search engines? How soon after the campaign launched were these coupons first printed? The opportunity to measure and collect this information is lost and cannot be re-created. You can’t find out which referral source was most effective, which special created the strongest pull or which creative was most compelling.
- What’s the solution? Creating separate landing pages for each source and campaign. Create downloadable coupons with a different reference code for each product specials. More work is needed to make things measurable, but granularity will allow you link redemptions with ads, attribute success appropriately, and optimize your ad placement next time around. Spend money on the things that are working best.
3. Tracking and troubleshooting shopping cart conversion.
- Scenario: Shoppers add items to a cart and checkout. The checkout process is a single long page containing billing and shipping data entry boxes, including an email confirmation id field. All terms and conditions, such as product guarantees, return policy, credit card privacy and email privacy are included on another single large separate pop-up Web page.
- Once the page is submitted, a thank-you page confirms checkout is complete. Yes, you can measure the start of checkout and the end, but if abandonment is high, such a design makes it difficult to pinpoint the problem.
- What’s the solution? Chunking allows usability problems to be pinpointed sooner. Split shipping and billing data entry onto two separate pages.
- Split the terms and conditions into four separate pop-ups, with hyperlinks closer to each activity. Label each pop-up uniquely, so that your Web analytics tool will differentiate each easily. Viewing frequency will show you which terms shoppers are more concerned with. Also, if more shoppers abandoned their carts after viewing a particular policy (eg. returns), this quickly raises a flag that there might be something unclear or unsatisfactory about the policy. You’ll be able to target the problem quickly and decisively execute a solution.
- Indeed, you can fix problems like the one described after the shopping cart is up and running. But by then, you’ve probably exhausted your initial budget, and you’ll need to fund an “enhancement”, which is often more difficult to justify. If your cart is experiencing a very high abandonment rate, you might have lost credibility internally. Even worse, you’ve annoyed the many shoppers who abandoned their shopping carts, likely damaging your brand and painting your company as one that’s difficult to deal with.