In late September, Google announced the rollout of their new Customer Match program – enabling advertisers to use their first party data for campaign segmentation. While the customer match program is currently limited to search (similar to RLSA), Gmail and YouTube campaigns, it is a tremendous leap forward for both Google and advertisers alike in terms of targeting capability and opportunities for improving performance, which will likely only grow from here. If you’re not already familiar with the customer match feature, you can learn more about it here.
In this article, I’ll provide you with some ideas for segments which you may want to incorporate with your SEM program. Skip to the bottom for a universal customer match strategy.
RFM (recency, frequency and monetary value) scoring has been a long-trusted method for identifying those customers who are most valuable to the business, and distinguishing them from those who are least valuable. Essentially, all customers in the file are graded on the following criteria:
Recency – how recently did the customer make their last purchase?
Frequency – how often does this customer purchase from us?
Monetary Value – how much money has this customer spent with us?
Traditionally, RFM quintile scoring grades each customer in the file with a value of 1-5 for each recency, frequency and monetary value criteria – where 5 corresponds to the top 20% of performers. The top performing customers will be scored as 555, while the worst performing customers will be scored as 111. While very large retailers may benefit from the number of segments which quintiles will yield (125 customer segments), retailers with a smaller customer file may want to consider utilizing quartiles (1-4) or terciles (1-3) for scoring. Regardless of your method for RFM scoring, you will end up with 27-125+ customer segments which can be tested within AdWords. Some RFM segments which you may want to isolate within your program are:
Top Customers (555) – These customers purchased recently, make frequent purchases, and have spent a lot. Consider excluding this audience from coupon-specific ad copy, as they are likely to purchase anyway. There’s no need to give up margin to this audience (so long as you know they don’t require a coupon to purchase).
Loyal Customers (X5X) – These customers purchase frequently. You may choose to further classify these customers based upon their R and M scores, or combine with other historical data, such as promotion sensitivity. How you act upon this segment will be unique to your business.
Worst/Defected Customers (111) – Your worst customers do not spend much, purchase infrequently, and have not made a purchase recently. These are low value customers, who do not merit a great amount of investment. They may have already defected.
You may want to consider adding this audience as a ‘bid only’ segment, and measure their performance independently of the rest of your campaigns, before determining whether to either decrease your bids for them or to change your messaging (to try to bring them back).
High Value Defected Customers (155) – These customers have historically purchased frequently and spent the most, but they have not made a purchase in a long time and you may have lost them as a customer. It’s possible you have lost these customers to a competitor – consider bidding extremely aggressively on this segment in any search campaigns which target your competitors’ branded terms. Or consider promoting a generous coupon offer to this segment.
Big Spenders (X15 & X25) – These customers have a higher average order value (large spend over few orders) and may be worth bidding more aggressively for. Conversely, you may want to consider bidding less for customers who make smaller purchases i.e. X11/X21.
How you use RFM segments is up to you. I encourage you to test segments as ‘bid only’ without any adjustments in order to determine whether there is a difference in value between the segment and the general population which the campaign targets. You may also want to consider holdout testing, where you exclude segments of your customer file from any paid search spend, instead directing your ad spend towards new customers. You may find through holdout testing that these customers will purchase from you anyway.
If you are not currently applying RFM scores to your customer file, you would likely find value in our RFM project for improved email, direct mail and paid search effectiveness.
If you are a catalog merchant, you may be familiar with Hillstrom’s Judy, Jennifer and Jasmine personas. If you’re not familiar with these personas, you can learn more about them here. Essentially, customers who act like Judy, Jennifer and Jasmine are all identified within the customer file.
Judy – An older customer who tends to purchase from catalogs or through your call center.
Jennifer – A middle-aged customer who is more comfortable purchasing online than Judy, but is likely still receiving, and perhaps being influenced by catalogs.
Jasmine – A younger customer who is very comfortable purchasing online and through mobile. According to Hillstrom, Jasmine will likely purchase regardless of whether or not you mail her a catalog.
How you might act upon these segments:
Judy – You may add a ‘bid only’ segment for Judy across your search campaigns, in order to measure the ad spend associated with those customers. Depending on web purchase behavior, you may choose to reallocate some or all of the SEM expense for Judy back to the catalog. If Judy is purchasing through the catalog, but web is a touch point for her, then it would make sense to reconcile the web expenses with the catalog purchase. You may also consider decreasing bids for Judy, though your performance will be the deciding factor.
Jennifer – In the opposite fashion of Judy, you may choose to reallocate catalog mailing expenses to online channels if you know Judy is purchasing online, and those purchases are influenced by the catalog. This actually isn’t impacted by customer match, though you might find it interesting to segment your campaigns for Jennifer in order to identify how this segment performs differently than Judy and Jasmine in search.
Jasmine – You may find Jasmine converts at a greater rate than Jennifer and Judy on mobile (or in general), and may warrant an increased bid. If you are currently promoting coupons in your ad copy, you might conduct holdout testing on Jasmine (or Jennifer) to determine whether there is incremental benefit in serving a promotion to them in your ad copy.
Before the customer match program, catalog and eCommerce retailers were unable to measure their paid search spend dedicated to Judy/Jennifer/Jasmine independently of each other. You may find value by initially applying each of these customer segments to your existing campaigns, with no bid adjustments or variance in ad messaging – just to measure how their performance varies from each other in paid search.
If you have seasonal merchandise, or have shoppers who tend to make cyclical purchases around select product categories, you may want to consider segmenting for these customers within your paid search channel. For example, let’s say you sell winter apparel. As winter approaches, you may want to consider creating a list of all the customers who purchased winter apparel last year, and isolating them within search, Gmail or YouTube campaigns.
Check out some of RevZilla’s video content (below)– they do an excellent job creating informative YouTube content which makes purchase decisions easier for the shopper. This is the type of creative which works well in YouTube TrueView.
Also consider targeting lapsed buyers. If you expect a newly acquired customer to make their second purchase within 200 days of their first order, target anyone who misses this hurdle. Same goes for time between 2nd and 3rd purchases, 3rd and 4th purchases, etc. These customers may need a push in order to come back – don’t lose them! Sure, hit them with your email campaigns, but also consider increasing bids for them in search, or engaging them with a YouTube TrueView ad.
If customer demand is inelastic for select users i.e. they do not require a coupon in order to purchase, you may want to consider excluding this segment from some or all coupon offers in order to preserve margin.
Connecting Offline Buyers with Online Spend
Similar to how a catalog brand might reallocate online expense associated with the “Judys” in Hillstrom’s model, you might find value in scoring and isolating the customers who are most apt to purchase offline, such as in-store, in order to measure them independently. I recommend setting these users as a bid only audience with no adjustment applied – all you need to do is measure the ad spend attributed to them. Then, consider re-attributing some, or all of the online ad spend associated with these offline purchases back to the offline channels.
Customers Who Leave Positive or Negative Reviews
Have you considered measuring the performance of customers who leave positive, or negative reviews for your business or products independently of customers who have not left a review? Export the list of customers who left a review from your product or business review platform (Trustpilot can do this), marry it to your customer file, and measure. If you find that customers who leave positive reviews are more likely to repurchase, or more valuable, you may consider bidding more highly for this segment in non-branded search. If you find customers who leave a negative review almost always defect, you might consider reducing bids or even excluding these customers from search.
Customers Who Receive Emails From Your Competitors
You may consider creating a Gmail campaign with the goal of retaining customers who may be at risk of churning by targeting those who are receiving emails from your competitors. If your competitors are threatening your ability to retain your customers by engaging with them through email, you now have an opportunity to combat this threat. This is incredibly easy to execute within AdWords. Just setup your Gmail campaign with a target and bid audience for your entire customer file, and utilize the contextual targeting option to target the brand names and domains of your competitors. Bonus points if you combine with recency scores (RFM) to isolate customers who are most likely to defect – if they have not already.
The Universal Customer Match Strategy
If the previously mentioned strategies sound like too much work (or too little return for a small business) then at the very least do one thing. The single, universal customer match strategy which every eCommerce company under the sun can benefit from is to simply measure the ad spend distributed to existing customers. All you need to do is set your entire customer file as a ‘bid only’ audience, with no adjustment applied.
Right now, it is not possible to differentiate between the cost of a new customer and the cost of an existing customer through SEM. The retailer is instead forced to measure a blended average customer acquisition cost (CAC) through SEM. With the release of the customer match program, all retailers now have the ability to produce a much more accurate estimation of how their CAC is split between new and existing customers.
It’s very likely some portion of your SEM spend is dedicated to re-acquiring existing customers – while the customer match program won’t allow you to pinpoint exactly what that cost is (as it only matches to Google accounts) it will enable you to do a better job gauging the amount of media spend dedicated to re-acquiring existing customers, and the amount truly dedicated to acquiring new customers.
While utilizing customer data to influence paid search growth strategies has always been a possibility, the AdWords customer match program, along with similar Facebook and Twitter programs, are steps in the right direction in terms of making this data actionable at a granular level. Nonetheless, every retailer, regardless of size, should begin tapping into their customer file today in order to better understand how to invest in marketing tomorrow.