Search Marketing Insights (Search Analytics)
Current Considerations for Future Tracking
By: Pat Stroh, VP Analysis and Decision Support
While there is a widespread recognition that online tracking is important for the operation of optimized marketing campaigns, I believe there is a widespread lack of understanding about how online tracking impacts actual campaign analysis and management. First and foremost, it is important to recognize if and how conversions are tracked BACK to actual clicks (from Search Ads, etc.). The reason why this is so important is that it is very difficult to understand the impact of bid changes, ad copy tests, etc. if the conversions are not attributed to the actual management activities. Imagine a situation where conversions occur days after the initial click (e.g., loan applications). If you don’t synchronize that conversion with the click management tactics, you will often be mislead about the success or failure of your campaigns. So, as a foundation of an Online Search Campaign, it is very important to establish the link between the conversion and the click-date. This can be complicated, because the tracking codes must be “sticky” throughout the conversion process. And there are other complications, such as cookie length, “conversion attribution to which click?” questions (e.g., the first click or the last click?), and whether “intermediate conversions” such as the online application itself can actually predict a lagged conversion days later. Without a firm understanding of that process, however, it is very difficult to make informed decisions regarding your online campaigns.
That being said – that your online tracking foundation must be strong and well understood – the connection between online activity (clicks, online conversions, etc.) and offline conversions (sales) is probably the foremost area for additional tracking, analysis and optimization. Many studies demonstrate at a high level that offline conversions are strongly influenced by online activity. Yet, it is very difficult to extrapolate specific tactics from those studies, or to make keyword-level ROI calculations. The most sophisticated system for tracking is, of course, the creation of persistent tracking codes that “follow” the conversion process from the online expression of interest to the offline conversion/sale. This “hard-wired” solution requires substantial technological integration across your online Analytics package and other systems (lead management, etc.). Certainly, there are several “direct tracking solutions” which also must be considered, including coupon and quote printing and online only telesales numbers (unique to campaigns). In some circumstances, these can provide a fairly clear picture of how offline conversions are connected to online activity. But in many cases, they tend to be incomplete (an estimated “floor” of how much online impacts offline) and perhaps misleading (coupon clippers, etc.). They also tend to be very time consuming and difficult to manage. Offline surveys fall also in the “get a general idea” area; the results are typically directional only, and do not provide the kind of tactical precision that online marketing offers.
The final method for tracking online activity to offline conversions, I think, is not getting enough attention by most online marketing departments – econometric modeling of the media mix. Of course, many marketers will recognize that econometric modeling forms the backbone for many different strategic and tactical applications, including brand valuation, pricing analysis, promotional measurement, direct mail targeting, etc. Additionally, many marketers know that sophisticated econometric modeling can “tear apart” the influences of different media – TV, direct mail, ONLINE, etc. – on overall sales (often down to the retail branch level) and, thus, provide an estimate of the return on marketing efforts. But what has not gained widespread recognition is that it is possible to estimate the impact of specific online campaigns on offline conversions via econometric models. There are several ways to do this, including historical data analysis, testing (campaign pulsing), and geo-targeting, which can illuminate (often at a very tactical level) the impact of online campaign management on offline conversions. Of course, there are many considerations when exploring such an option: Do you have a lot of historical data, including other elements of the media mix (TV, etc.)? What percentage of your overall marketing budget is devoted to online? In the grand scheme of things, does your site garner a substantial amount of traffic? To what extent are online conversions, such as particular page views, store locators, etc., a “leading indicator” of offline conversions? How open are you to testing (optimization over the long-run, as opposed to the short-run)? In many cases, larger and/or more sophisticated online marketers will actually be positioned very well to push an indirect, econometric method forward. And, in doing so, make a fairly strong connection between their online efforts and offline conversions. That, I believe, is the next wave in online tracking – supplementing and strengthening “hard tracking methods” with analytical approaches.
Let me give an example of how IMPAQT is pushing forward on the connection of offline conversions to online activity for one of our national retail clients. First, we want as much Search traffic data as possible, including impressions, clicks and conversions (that presumably might drive offline transactions). Is there a lot of fluctuation in those metrics (econometric models like “variance”)? No? Then we do a test; pulse your Search campaigns by “going dark” and “doubling up” on alternating days or weeks = create mountains and valleys of Search traffic. Now, we pull in extraneous (“uncontrolled” / planned long ago) sources of both online and store activity, including offers/price cuts, TV and online display. (These can have gigantic effects, and we want to “control” for those sources of variation in the offline transactions.) Once those pieces are in place, we’re ready to answer the core question: Did Search traffic (store locator conversions) correlate with store traffic, above and beyond our “normal” expectation (seasonality, TV, display, etc.)? If the Search-to-offline effect is large, the econometric model should pick it up (even without many control variables). If the effect is small, but you have enough historical data, variation (via testing), and control data, then the econometric model has a decent chance of picking up the effect. In the latter case, even a small percentage contribution of Search to overall offline transactions can translate into major dollars. In either case – large or small effects – the goal is still the same … measure the full contribution of Paid Search on your company’s revenues and consequently, get a firm grasp of your Search ROI. This is something IMPAQT believes that other methods of tracking and offline conversion attribution (e.g., coupons, surveys) struggle with, because the Search activity and offline transactions are directly and mathematically correlated with one another. So, if you want to “track” the contribution of Search to your enterprise; don’t just think about “hard tracking”, etc., but also an econometric approach.
