You can adjust all of your cookie settings by navigating the tabs in this window.
After three years of building best-in-class software for advertisers to more efficiently manage and optimize their Facebook ad campaigns (including scaling a campaign to spend $1M in 1 day on Facebook), it’s no surprise that we have a tried-and-tested methodology for successfully launching new campaigns.
We’ve continually updated this methodology as both the Facebook advertising ecosystem and our software platform have evolved to meet the needs of large-scale performance advertisers. The recent broad launch and success of unpublished page post ads was the impetus behind the latest update to our recommended launch process, which we’ve decided to share publicly below:
The launch methodology starts with what (or rather, who!) you already know — your core customers. We don’t mean this in terms of general buying personas, nor do we mean this in terms of their average demographics, psychographics and behaviors. We mean the actual individual people in your CRM database!
By uploading your CRM information (be it phone numbers, emails or Facebook user IDs), we can match the information to Facebook profiles containing this same information. This creates a custom audience of known customers to target (learn more about custom audience targeting here).
People similar to your known customers have the highest probability of converting and delivering return. So next in the process is expanding targeting by identifying new audiences on Facebook similar to your known customers. We don’t recommend haphazardly creating these look-alike audiences; instead (in true Nanigans fashion), we recommend you look at the data.
Building affinity models to do this can be an incredibly time-consuming process with the plethora of targeting data on Facebook, so we’ve automated the process by building a look-alike tool within our platform. This tool can examine your known customers uploaded in Step 1, and stack rank new potential audiences based on percentage interest affinity overlap (learn more about the tool here).
Thousands upon thousands of audience possibilities could be stack ranked. Audiences with high affinity overlap with your known customers will be few in number, and those with low affinity overlap will be many in number. It’s key to expand your targeting into only the highest quality look-alike audiences the tool surfaces.
Just how many look-alike audiences should you target? This depends not only on a high affinity overlap, but also your budget. It’s critical to have enough budget to test each segment with data sufficiency. We’ve built a model to make this easier for advertisers, by recommending the exact number of impressions that should be delivered for a given audience on Facebook when considering a click and post-click conversion rate.
With targeting in line, it’s time to turn to creative – unpublished page post ad creative, to be specific. The graphic below highlights why we recommend this ad unit over others in launching a campaign (in fact, here’s 10 more reasons our advertisers are flocking):
… Don’t worry, we of course have the data to back up the recommendation! In January we conducted a study of nearly 1 billion Facebook ad impressions to understand the ad unit’s effectiveness in driving sales, purchase revenue, and ultimately return on ad spend (ROI). The result: page post ads in the News Feed delivered on average 14% higher ROI than standard domain ads delivered in the right-hand side of Facebook (read on here).
Note: If you are a mobile-only advertiser looking to acquire customers for your mobile app, we recommend using mobile app install ads instead of unpublished page post ads.
After deploying unpublished page post ads to your core and look-alike audience segments, it’s time to measure and understand behavior of people who click or engage with your ads. Post-click measurement and attribution is enabled through Nanigans’ pixel technology, which maps to specific events and business goals in your conversion funnel (e.g., registrations, email subscribes, repeat visits, add-to-carts, purchases).
Since people convert and purchase over time after engaging with an ad, it’s critical that you measure behavior over time — not just upon immediate click. We recommend doing so with Ad Engine’s cohort analysis tool. In the example cohort table below, the advertiser has run cohort analysis to understand the ROI of their ad spend in the days following ad delivery.
With ads delivered to the November 26th cohort, the advertiser saw a 103% return on their ad spend in the same day ads were delivered and a 208% return after 7 days. Ads delivered a week later, the December 4th cohort, performed much better – achieving 438% return in the first day and 1,080% return after 7 days.
Notably, this cohort table also reveals the benefits of lifetime ROI optimization on Facebook (which the advertiser behind this cohort table employed). The December 4th cohort achieved the highest ROI and drove the second highest revenues, all despite the fact that it also experienced the highest CPCs (read more in this case study).
Next in the launch process comes remarketing. Leveraging Facebook Exchange (FBX) is a highly effective way to convert missed customers and sales. While up to this point in the launch process you’ve focused on audiences, FBX allows you to bid in real-time at the individual impression level on Facebook.
FBX not only helps convert people you’ve already targeted with unpublished page post ads who may need a “reminder” to fully convert; it also helps convert people who have visited your web properties organically or through other paid channels who then visit and spend time on Facebook.
Why not leverage your standard DSP for remarketing on Facebook? Well, you’ll miss out on 39% additional customers who purchase 89% more (read more here). And there’s plenty more reasons in the below:
With performance learnings reported in our tool from Steps 1-5, next up in the launch process is refreshing your audiences. This includes both your known customers uploaded in Step 1 as well as your affinity audiences identified in Step 2. Rather than refresh these representative audiences manually, it’s possible with Nanigans to do so programmatically with thanks to the deep data integration and automation we’ve built into our software platform. For example, our bidding algorithms will automatically drop spend from poor performing look-alike audiences and shift it to ones performing strongly.
While our recommended launch process is completed after these 6 steps, advertising on Facebook is a continuous and iterative process. As you scale further and test new targeting and ad units, we recommend advertisers continue to keep this process in mind!
Want to learn how the Nanigans launch process can help you successful scale your Facebook advertising campaign? Contact us today!