These days, search is often treated by marketing departments as old news — a mature, well-defined, even staid marketing channel that lacks the excitement of emerging opportunities in social, video and native. The reality, however, is that paid search marketing is constantly evolving, and it is absolutely crucial for marketers to remain on the leading edge.
Never has this been truer than over the last couple of years, with the following shifts taking place before our eyes: desktop to mobile, text-based to image-based ads and keyword targeting to user targeting. These emerging capabilities, combined with stronger competition and media fragmentation, means that search is continuing to get more complicated, requiring a more sophisticated and scientific approach than ever before. In fact, in Criteo’s recent Forrester study on search marketing trends, 75 percent of retail marketers polled said they rely more on agencies and technology partners to navigate paid search than they did just two years ago.
Until now, generating a strong ROI from search could largely be driven by individuals who could process the data coming back from search engines in Excel spreadsheets and make informed decisions on what to do next. In the performance display market, that type of approach was never really possible, due to the sheer scale of impressions delivered and the sheer number of variables to be considered before placing a bid.
As a result, the display industry developed a set of capabilities to compete for search dollars and become more relevant to performance marketers. Per eMarketer data, the display market in the US is now larger and growing faster than search. The opportunity now exists for search marketers to embrace these same capabilities to re-energize their own performance. We call these the 4Ps of Display:
As it becomes easier and easier to store and access large amounts of data, our ability to predict the future by studying the past increases. We are now able to determine what variables influence campaign performance, as opposed to merely reacting to a performance change in a particular element of a campaign (like a keyword or ad group). To be able to predict performance accurately, marketers need the ability to collect and store data cleanly and consistently.
For instance, when running a Google Shopping campaign, if you are able to pull down performance data from AdWords and combine it with information from your product feed keyed on the product_id or gtin value, then you will be able to extract very interesting information about your program that can have far-reaching effects on your long-term performance.
Prediction allows you to take the guesswork out of search marketing and answer questions such as:
• Are there specific types of products that perform better on certain days of the week?
• Which user behaviors on my site indicate that someone is in market for my product?
• What happens when I put a product on sale discounted by more than 15 percent? Does the volume gain justify the promotional expense?
• Am I more competitive in online marketplaces on certain brands than others?
• Which product types deliver the most loyal customers to my business?
These types of analysis often do not require massive data processing or machine learning capabilities, but this type of intelligence will allow search marketers to predict the present confidently, as opposed to constantly reacting to the past.
One of the biggest changes ever to search marketing was Google’s introduction of re-marketing lists into search (RLSA) in 2013. This enables marketers to use behavioral information such as onsite browsing and purchase information to inform their bid strategy and take a different approach based on customer profiles. Google has since added Customer Match capabilities to allow for CRM data matching, further enabling rich targeting strategies within search.
With these new capabilities comes opportunity, but also significant complexity. Our recent Forrester study tells us that 58 percent of retail marketers believe they don’t currently have the resources required to manage RLSA effectively and will be investing further next year. When we evaluate existing programs, we also see that marketers are leaving a lot on the table by not taking full advantage of these capabilities.
For instance, one of the most common behavioral segments we see in search campaigns is “Visited Last 30 Days.” In essence, these marketers are bucketing everyone who visited the site recently into the same segment and applying the same bid modifier to those customers without regard to engagement and recency of visit within that 30 days.
In our evaluation of shopping behavior across our 12,000 advertiser relationships globally, we see that there is up to a 10X difference in value among customers who have visited in the last 30 days. Segments for RLSA need to be much more granular in order to value visitors appropriately and bid efficiently in an ultra-competitive market. In our display business, we are able to value each customer individually and can bid for inventory accordingly. While in search there are limitations to how small your segments can be, a more granular approach is always superior.
On the SERP, images are quickly taking the place of text ads, and with the explosive growth in mobile, search is fast becoming a truly visual environment. Shoppers respond very well to images, and that is a key reason that PLAs are scaling so quickly. In her recent 2016 Internet Trends report, Mary Meeker cited that visual shopping experiences on sites like Houzz, Pinterest, OfferUp and others are driving online monetization across the industry.
One thing that we have known for years in display is the importance of image quality and layout to consumer engagement. The ability to introduce multiple ad layouts and product images significantly increases conversion potential.
In search, product listing ads populated with images selected from an advertiser’s product feed are becoming much more commonplace. These ads are showing up on the core SERP, in image search, on local queries and off-network. Those advertisers who are set up to test multiple images to arrive at the best ones for these new ad environments are going to have a leg up on the competition.
This word gets thrown around a lot these days, but the concept is relatively simple. If you are truly programmatic, you are responding to everything that impacts the value of an impression in real time. While it is challenging to optimize all aspects of your search campaign in real time (due to API/publisher restrictions), marketers need to think and operate in near-real time to be truly efficient.
Over the past few years, the number of actionable pieces of information that impact the value of a search impression has increased significantly. Some examples:
• What is the query being executed, and who is executing it?
• At what point are we in the business cycle?
• Has this searcher been on my site before? Have they bought before?
• What device are they on when conducting their search?
• What is their current location? Is that different from their typical location?
• Which of my products are currently discounted? By how much?
In order to respond effectively and at the right price, a programmatic approach — one that is constantly considering the value of each piece of information — is required for success. This is the basis of programmatic advertising in display, and it is increasingly important for marketers that wish to excel in search.
To respond to the rapid changes in search that we have observed and help marketers take advantage of these 4Ps of display, today we are publicly releasing Criteo Predictive Search, our first product designed specifically for search marketers. Leveraging 600 TB of consumer shopping behavioral data daily, Criteo Predictive Search takes the guesswork out of managing Shopping Campaigns to deliver maximum ROI.
Built from the ground up to drive performance on Google Shopping, Criteo Predictive Search is a fully automated solution that uses proven technology to continually apply precise, predictive optimization to your campaigns. We strongly believe Criteo Predictive Search can help retailers consistently and confidently increase results on Google Shopping.
Early adopters of this solution who have taken part in our beta tests have seen as much as a 22-percent to 49-percent lift in revenue at constant cost. These clients include 30 leading US retailers, including Revolve Clothing, Teleflora and Camping World. For more detailed product information, check us out at Criteo Predictive Search.