Ads API Profile: XA.net’s Optim.al Provides Cheap Multivariate Testing
Facebook’s performance advertising system lets buyers target users based on a wealth of demographic and interest traits, but it doesn’t provide the depth of control to test creative and targeting parameters on a large scale. To allow companies to build more powerful tools for designing and managing ads, Facebook released an Ads API last fall. One company offering a self-service tool for both small and large advertisers which utilizes fractional-factorial multivariate testing to quickly determine ad components with the most significant impact is XA.net.
Having already focused on display ads, XA.net designed a system for testing ads with different creative and audience targeting elements without testing every single permutation. This helped the company build a base of clients interested in getting actionable results without spending as much time and money. XA.net was an early tester of Facebook’s bulk uploader and now-defunct conversion tracking tool. It developed its new Facebook ads tool in-house with its managed spend team, and is now ready to offer the tool in private beta so advertisers and agencies can manipulate Facebook ad campaigns themselves.
The company is in private testing now, but the first 20 interested advertisers can get access through a special invite code, available at the bottom of this article.
Originally named CPM Advisors, XA.net was founded in 2008 with a focus on building audience-based advertising tools, primarily for online display ads. In 2009 it released a self-serve ads tools so clients could buy media, generate tracking tags, and use retargeting pixels to manage campaigns. The company’s 25 person, technology driven team operates out of downtown San Francisco and Los Angeles. Advertising technology investors including Jonah Goodhart, Noah Goodhart, Jerry Neumann and Mike Walrath provided $600,000 in funding for XA.net in 2008.
Clients helped to buy Facebook ads by XA.net include both direct advertisers and agencies representing industries including financial services, online education, retail, and consumer goods. XA.net also helps publishers, developers and media companies who want to drive traffic to their Pages and apps by using their permission to target ads to the Page and App’s existing audiences. Client average spend varies greatly, down to small advertisers spending just a few hundred dollars a day. Many Ads API tool providers only license to enterprise class advertisers, making XA.net an option for those looking to spend less. The company charges a percentage of buy on top of cost, with smaller, short term clients paying up to 20%, with that percentage dropping as clients spend more and commit to longer buys.
XA.net is taking a “Silicon Valley 2.0 approach to releasing the product” said founder and CEO Rob Leathern. “Put the product out, see how they use and respond to it, how they like it, make changes and enhancements, and find the right way to charge users to incentive use.” Currently these is still some interaction with humans at XA.net when using optim.al, but the company is building towards creating a no-touch technology platform.
XA.net’s optim.al Ads API Tool
Optim.al will not only be for Facebook advertising, but will also integrate other text and image advertising channels over time. The product currently only runs Facebook ads, but users will eventually be able to take the Facebook ad content that’s proven to work and use it in display ads.
Since Facebook’s CPC estimation and optimization algorithms tend to bias traffic towards ads in a campaign which perform well early on, optim.al helps users run ads on a CPM model to get relatively even traffic. This helps clients perform accurate multivariate testing, then translate the results into CPC campaigns.
XA.net believes some Ads API tools pull back inaccurate segment sizes when targeting keywords and cities. Therefore, the current release of optim.al only allows clients to target by region, state, country, age, gender, sexual preference, relationship status, education, and Page connections if optim.al has permission to target or negatively target a Page’s audience.
Clients can dump images into an image library where they are automatically re-sized and made available for use. Once ads are completed, clients can view previews of ads, or see their images in-line with targeting and other creative info in the campaign creator. Unique to optim.al is that ad body copy is split into two lines in the ad builder. This allows clients to break out and separately test an explanation and a call to action within a single ad, or keep the first line of the body copy the same while testing variations in the second line.
What differentiates optim.al from many other Ads API tools is the ability to use different creative element multivariate testing models. Along with standard full factorial testing where every permutation of targeting parameters and creative elements is tested as its own ad, optim.al offers fractional factorial multivariate testing. This allows the tool to only test variations where a meaningful variable is different, producing optimization insights without having to spend as much time and money to reach a sample size of test data on insignificant variations. in some cases clients can get comparable data at 25% of the cost, the company says.
Optim.al provides optimization suggestions based on the tests, and in the future clients will be able to set conditions such as “if this ad performs 30% better than this one, shift additional budget to it”. Currently clients can only use the multivariate test builder on creative elements, not targeting parameters.
To allow for accurate performance reporting, ads are dynamically fitted with campaign or ad IDs for tracking, clients can install a conversion tracking pixel on their landing pages, and clients an upload conversion data from their own analytics systems. Data from optim.al can be exported as .csv or Excel files, and eventually the tool will include native data visualization and graphs. Optim.al’s reports can interface with XA.net’s display ads self-serve tool CPMatic for advanced reporting across channels.
To assist clients with writing high performing ad headlines and copy, XA.net has partnered with BoostCTR, a company which optimizes text in ads. Clients can give BoostCTR access to their account, and the company will go in and provide suggestions for better copy.
Looking For More Data
XA.net is building their tool according to what its clients want, which so far has been testing efficiency. Leathern says there’s lots for Facebook to improve on the Ads API, but XA.net have been pleased with the responsiveness of Facebook team’s to feedback. It hopes Facebook will expose more of the data users add to their profiles to advertisers for targeting. Leathern says he’d “also like to see more defined control around frequency, as frequency is a big driver of performance.” Currently, audience exhaustion and creative rotation is a big concern, and better tools for monitoring frequency would help.
While still evolving, optim.al is a powerful tool for smaller advertisers who can’t afford large full factorial tests. By concentrating on offering a cheap and fast way to gain more data with less tests, and eventually integrate that data across channels, XA.net can find its niche in the growing Ads API industry.
The first 20 advertisers interested in trying out the private beta of optim.al can input the code ISFB01 on the optim.al sign up web site.