The complexity of any advertising campaign rarely correlates with money spent but usually hinges on the number of channels, products, audiences and product sales cycle covered by the effort. Where you allocate your advertising spend will determine your success or failure. You can allocate spend in many different ways, but one vital way to assure you spend wisely and efficiently is through channel segmentation. In a series of articles on juggling today’s digital advertising challenges, we consider balancing many channels, handling accounts with thousands of products, reaching diverse audiences and making sense of conversion tracking in long sales cycles. In this article, we discuss why segmenting and comparing your advertising channels will help you properly distribute your advertising spend.
The opportunities in the digital advertising world change seemingly every day. The endless new extensions on Google AdWords, changing targeting options on Facebook, Sponsored updates on LinkedIn and Yahoo Stream Ads just to name a few of the options that create a complicate your advertising mix. Before attempting to figure out how to manage the vast array of options crying out for your advertising dollars, you first must understand each of your channels and their relative value to your overall results.
Challenges in e-Commerce and Lead Generation
On an e-Commerce situation, the relative value of the conversions is somewhat easier to gauge. Conversions coming from Bing Search might be of a higher average order value than those of Google Content. The complexity steps up a notch when you start comparing segments of channels like: Google Search, Google Display, Google Placement, Remarketing (from many different sources), LinkedIn, Facebook, Yahoo Stream, Bing Search, Bing Content, Twitter, and on and on. For now we won’t even discuss adding in demographic targeting on each channel segment, tracking interaction between channel segments through attribution, device types, time of day, day of week, location or weather. Still, e-Commerce is significantly easier to compare since you often see in near real-time how effective a channel segment performs relative to others. However, when you add in the element of lifetime value of a customer received from a channel segment, your allocation of spend could shift. In the case of lifetime value considerations in e-Commerce, you find yourself in the same situation as lead generation by needing to weigh the quality of a transaction coming from one segment verses another.
With lead generation, you add in an element of quality related to the segment that extends beyond the cost of the lead and into the lead’s quality. It hardly matters if I generate 3x the leads from Bing than Google with the same spend, when the additional leads don’t turn into more sales in the long run. Adding the element of quality involves a lot of tracking and integration. You must merge data from search engines into back-end systems to know where leads originated and pull from back-end system to help make decisions about relative quality sometimes six-months to a year after the original lead. Obviously you can’t wait a year to make changes in your advertising mix, but keeping a constantly updating model that estimates overall quality makes sure you put your dollars in the right place with the best information possible. If a final sale is way too far off in the future for managing spend allocation, then use some middle step like qualified leads or proposals use for assigning value.
Segmenting has its limits and without the force of a really good management system to take all of these elements into account, we need to balance the value of the information gathered from segmenting with the time needed to analyze and make decisions based on the results. Although your management platform, if you use one, might take into account some or even many of these segmentation possibilities, the platform likely can’t keep up with the new sources and their influence on the mix. You’ll likely need to augment the reporting of your management system with your own custom built reporting to get the whole picture without too much manual work. Pick your segments wisely, or the sheer amount of variables will cause you to curl up in a ball on the floor of your office. We have taken a somewhat grand view of the options, but now we need to take a breath and scale back to something a lot more manageable.
Pick Channel Segmentation That Makes Sense for You
The first step in reigning in the possibilities comes from understanding the past. Look at the segments you have used historically: Google Search, Google Display, Google Remarketing, Bing Search, Facebook, etc. How have they performed based on your overall goals? Everyone should already segment based on engine at the very least and most likely add some more granular option such as Display, Search, Sponsored Stories, In-Feed, Remarketing, etc. Take into consideration the historical differences between the channels and your overall volume of leads or sales. Much like A/B or multivariate testing, with lots of volume comes quicker decision making.
Relative Value of Each Channel Segment
In next step, you assign a relative value to each channel segment. Pull all of the historical data from the native engines or your management system. Your best involves looking at the segments on a monthly basis to make sure that performance over the life of your segments haven’t dramatically changed. Aggregated data, especially over the life of a segment, can mask changes in segment performance hiding recent negative or positive trends. We want to use this information to allocate future spend, so we need to understand segment quality, but only the recent trends should be used to allocate spend. Analyze dramatic changes in channel quality to see if negative changes can be reversed or positive changes replicated in other segments.
Once you have segmented and pulled the performance metrics together, you can begin making judgments about the relative performance of the segments. The judgments you make are no different than the ones you already make at the keyword, ad group or campaign level. Your KPIs remain the same, you merely use them to help you decide where to allocate potential new spend, where to focus your limited testing attention and to judge new segments relative to your existing segments. The same exercise happens every day from within a channel through automated keywords bidding algorithms and shifting budgets from one campaign to another. But what we want to accomplish with channel segmentation influences projections and spending allocation across all segments rather than just within Google or Facebook. Additionally, if a segment begins to under-perform, you know where the next best option resides and can quickly estimate how the shift in spend will influence your KPIs.
As an example, let’s pick a simple lead generation case to analyze. Assume that you have two channel segments: Google Search and Google Placement. Google Search averages a CPL of $20 and Google Placement averages $35. Placement leads close at 2x the rate of Search historically. The obvious answer is in spite of the higher CPL we would allocate dollars to Placement leads before Search leads. In the real world, the total number of leads from any segment is finite, so we would need to estimate the total number of leads possible from each channel, apply funds to the best segment, second best and etc. till you either run out of funds or out of estimated leads. No matter the KPIs for your advertising efforts, the same principle applies. Of course, things change quickly in digital advertising; new sources arrive, competitors change the game and your campaign changes will shift performance. You must keep updating your segment quality values or risk allocating spend to the wrong channels.
| Source | Leads | Cost | CPL | % Qualified | Q Leads | Q CPL |
| Google Search | 35 | $ 700.00 | $ 20.00 | 15% | 5 | $ 140.00 |
| Google Placement | 23 | $ 805.00 | $ 35.00 | 30% | 7 | $ 115.00 |
In today’s article, we built a foundation for understanding how segmenting and estimating performance allows you to project and allocate spend efficiently. Using a historical sense of quality related to a segment gives you some guidelines on where to spend budget and where to shift should performance vary from estimates. In our next article, we will discuss the challenge of working on campaigns for products with very long sales cycles.
