Remarketing List for Search Ads
If you are unfamiliar with Google’s Remarketing Lists for Search Ads (RLSA) here are the basics. RLSA allows advertisers to retarget users who have previously visited their website via standard keywords and audiences. Unlike standard display retargeting, remarketing ads serve on Google’s search engine results page. Audiences are created in the shared library of an account and are used to group visitors based on previous behavior. You can make audiences based on which pages people visited or how long users stayed on the website. Sky is the limit when it comes to possible audiences. This, in addition to the keywords, determines whether or not a user is served an RLSA ad.
Why is RLSA awesome? Retargeting is already a tried and true digital marketing tactic. Being able to segment customers based on behavior and bid accordingly increases your odds of winning the second chance. For our client Cheap Caribbean remarketing is extremely important. On average it takes customers 45 days to make a decision and purchase a vacation package. Most of our customers need to do a lot of research and Googling before they find the best place to dig their sand in the toes, hold a tropical drink in their hand, and unwind with a perfect view right outside their window. RLSA gives us the opportunity to bring back the people who left the site to do more research before they completed a transaction.
Our Winning RLSA Strategy
We started by creating audiences using Google AdWords and Google Analytics to gather the most precise audience insights and retargeting metrics. We tested audiences in 3 separate categories:
- All Site
- Checkout Process
- Custom Parameters
1. The All Site audiences include: any person who visited the site, all visors excluding bouncers, sitewide remarketing pixel within a 5 day window and past purchasers within 15 months. We created the All Site audiences in GA to utilize the richer user data GA can accumulate based on customer behavior on site.
2. The Checkout Process audiences were created for every step of the checkout process. We included date selection, flight selection, room choice, excursion selection, trip insurance, login/checkout as guest option, and final payment audiences. Additionally to prevent overlap, each audience excludes the checkout stage before it.
We bid differently on each audience (ranging from 15% – 75% bid adjustments) according to where the user was in the checkout process. For example, we bid less on the earlier stages of the checkout process, and we increased bids as the user reached step 5 and 6 of the checkout process.
3. Lastly, we created audiences based on custom parameters, specifically with price and destination intent. We created these audiences using the Data Layer, Google Tag Manager and the AdWords Remarketing Pixel to specifically re-target individuals based on the price point they searched on the site and the specific destinations they browsed.
The following is an analysis of RLSA campaign performance from 10/26/15 to 12/20/15.
The Checkout Process audiences perform the best in terms of efficiency and booking volume. In addition to having the most bookings at 470, the Checkout Audiences had a CPA 2.5 times lower than both the Custom and All Site audiences. The conversion rate for these audiences is also 3 times that of the other two audience groups. As seen in the graph below, the payment and trip insurance selection outperform audiences that are based on stages earlier in the checkout process. As a result, we have paused or decreased our bid modifiers on audiences in earlier stages in the funnel such as the date and room selection. The exception to this is the flight selection audience with 25% of bookings. This is likely due to the volatile nature of flight costs. A flight that originally costs $800 one day could easily decrease to $500 a week later. We believe that our bid adjustments help effectively target people at different stages of the checkout process which has played a large role in our success.
All Site Audiences
The All Site audiences are not far behind in terms of performance with 373 bookings. While the all site visitor audience excluding bouncers is about half the size of the all site visitors audience (680K vs 1.2 million), the all visitors excluding bouncers audience performs better and is live on all of our remarketing campaigns. Pausing the general all site visitors audience has led to a steady decrease in total RLSA CPA by approximately 22%.
Custom Parameter Audiences
The Custom Parameter audiences, while beneficial, do not currently contribute to many RLSA bookings. However, these audiences do have a lot of potential in terms of granularity by allowing us to serve more relevant ad copy with the addition of new AdGroups using specific custom audiences. We can also combine them with checkout and all site audiences to scale the RLSA programs. They have played a huge role in our success with dynamic remarketing campaigns on the GDN.
RLSA Account-Wide Impact
RLSA is currently trending 26% under our target CPA for NonBrand campaigns. Additionally, the RLSA programs account for 38% of bookings that come from all NonBrand programs, a number we see increasing week over week. It continues to be the strongest booking contributor out of all NonBrand campaigns. As we continue to scale these programs we expect to see their performance continue to improve over time. Bing recently introduced RLSA and will likely be our next expansion early 2016.