New Best Practices For Facebook Ad Targeting
Consumers dislike seeing irrelevant ads. This aversion is compounded in the digital landscape where personalization has become the norm. According to McKinsey & Company, nearly three-quarters of consumers expect personalization from companies, and an even higher percentage feel frustrated when it doesn’t occur. Knowing how to optimize targeted ad campaigns is therefore crucial for companies: If you do it poorly, you may repel customers and hurt your ad performance. If you do it well, however, you can expect to attract new customers and elicit brand loyalty.
Social media platforms offer prime opportunities for personalized marketing. While newer social platforms like TikTok are growing in importance, Meta’s Facebook and Instagram are still the most important platforms for companies of all sizes to invest in. After all, Meta Ads Manager taps into a group of approximately 4 billion monthly active users. In an ever-evolving digital marketplace where exponential growth is the rule of play, this is a consumer audience that companies cannot afford to miss.
As a seasoned growth marketer, I’ve witnessed firsthand the transformative power of implementing strategic Facebook and Instagram ad targeting. It’s common for small companies and those without dedicated growth experts to invest in campaigns that are too broad. Although Meta has made it increasingly easy for companies to launch marketing campaigns without any targeting customizations—including with a new artificial-intelligence feature called Advantage+ audience—social media marketers should continue to use data-based testing and treat Meta’s automated targeting recommendations as options. Meta’s objectives may not always align with your own. Plus, companies generally have unique customer insights that Meta does not.
This test-first approach is especially important given that Meta has less access to audience data from outside the app than it did before 2021, when Apple began requiring Facebook, Instagram, and other apps to request tracking permission on iOS devices. Within a year of the change, a majority of users had opted out of tracking. This trend means that first-party customer data is even more valuable as a starting point for finding new customers and reengaging existing ones. In this article, I provide five strategies to help companies maximize their customer data and otherwise drive growth with Meta campaigns.
Leverage High-value Audience Lists for Lookalikes
Smart usage of Meta’s lookalike audiences is key for Facebook and Instagram ad targeting. This option allows businesses to upload a high-value audience list to Meta and build new, larger audiences based on patterns in the source list. Facebook launched lookalikes in 2013, yet many companies still don’t use them effectively (or at all). For example, when I began working with an online luxury home decor retailer, I discovered the company’s target audiences for Facebook and Instagram ads were defined according to demographic data, such as age and location, and didn’t include lookalikes. To advance from this common but basic approach, we uploaded a list of its top 25% customers by spending and then tested different lookalike strategies.
Meta allows two essential optimizations for lookalike audiences. First, marketers can select a geographic region or country for the lookalikes. (Finer-grained geotargeting can later be added while designing ad sets.) Second, you can adjust how similar the lookalikes should be to the existing list—this is crucial for testing. A 1% lookalike audience represents the 1% of Meta users in a specified geographic area most similar to the source list, whereas a 10% lookalike comprises a larger but less similar set of users. The number of users in that percentage depends on the region selected; I recommend testing 3%, 5%, and 10% lookalikes for effectiveness—which is exactly what the luxury home decor client did.
The new lookalikes resulted in a significant increase in conversion rates for the client and also a moderate increase in average order value relative to the baseline ad sets. In my experience, companies can expect to see improvements of 20% to 40% on these metrics when first deploying high-value lookalikes, although it’s important to note that these improvements do not suggest the beginning of continuous hypergrowth. The rates will ultimately even out, requiring additional testing and iteration.
The key to creating successful lookalike audiences is to analyze your customer database for high-spending or highly engaged segments and to test often. In addition to your cohort of top customers, consider using lists of visitors who have spent the longest time on your website or social media users who engage most often with your content. This method zeroes in on high-potential lookalikes, enhancing the efficiency of your acquisition campaigns. In terms of experimentation, remember to test different lookalike percentages and lists on an ongoing basis. For instance, it’s not uncommon for a larger company with a digital marketing budget that exceeds $100,000 per month to take a hands-off approach to testing lookalike audiences—but that is a missed opportunity for growth. Perhaps a 1% lookalike has worked well in the past, but a 10% lookalike today would produce real improvements in ad conversions.
Stack Audiences for Multi-source Lookalikes
Another valuable strategy for Facebook and Instagram lookalike audiences involves an approach that I call audience stacking. This is when lookalike audiences are constructed from multiple source lists, such as customer lists, social engagement data, and website conversion interactions from Meta Pixel (i.e., the JavaScript code that all companies advertising with Meta should install on their websites). Stacking allows you to combine company-collected source data (e.g., high-value customers) with Meta-collected data (e.g., the people who have watched over 75% of an Instagram video in the last 90 days), thus broadening your reach while maintaining relevance.
I stack lookalikes for almost every account I work with, especially when I see one particular lookalike audience performing well. For instance, I used this approach to run targeted Facebook and Instagram ads for an e-commerce apparel brand, combining lookalike audiences from subscribers, customer lists, and social followers, which led to a significant decrease in purchase cost per acquisition (CPA) compared to the broad ad set—again, an improvement of about 20% to 40%. This multifaceted targeting strategy captures a wider audience while ensuring relevance and increasing the likelihood of finding high-potential prospects.
Refine Attribution Models for Improved Ad Targeting
Another crucial aspect of targeting Facebook and Instagram ads is choosing the best attribution window. The default model for Meta campaigns is known as the “seven-day click and one-day view.” To understand this, imagine the following scenario: You’re in the market for a new pair of sneakers and happen to scroll past an Instagram ad for a shoe company. You don’t open the ad or even pay much attention; it just appears in your feed. The next morning, you search for shoes on Google and see an ad for the same company. You click on it and purchase the sneakers. Because the purchase happened within 24 hours of viewing the Instagram ad, but the purchase was made through Google, both Meta and Google will take credit for the conversion. This overrepresents the effectiveness of Meta ads.
In my experience, an alternative model known just as the “seven-day click” tends to be a more balanced and accurate measure. In the sneaker scenario, this means that the conversion would only be attributed to the Meta ad if you clicked on the ad and made the purchase—via Meta, Google, or any other site—within a week. Just having seen the ad wouldn’t be enough to attribute the conversion to Meta; you would’ve had to click.
Opting out of the one-day-view attribution may seem unappealing at first. After all, the in-platform reporting will initially look worse. But the performance metrics will be a better match for what’s happening, closing the reporting gap between Meta and any third-party tracking software your company uses, such as Amplitude or Triple Whale. (These third-party tools are better at avoiding double attributions.) Even more important, Meta’s system will focus on learning more about the kind of users who actually click on your ads and then go on to make a purchase. By accepting lower performance numbers up front, you will ultimately train the system to target a higher-value audience, thereby improving your conversions and lowering CPA. Regularly testing and comparing different models can help determine which option provides the most accurate reflection of your campaign’s impact and aligns with your specific business goals.
Compare Broad and Interest-based Ad Targeting
By default, Meta encourages marketers to use broad audience targeting because it allows Meta more control over how to spend your marketing budget. Yet it’s always good to test whether Meta’s recommendations are actually right for your business. Broad interest targeting on Facebook and Instagram offers extensive reach but isn’t always the most effective investment. After all, companies often have a detailed sense of their audience’s interests, which may provide a better reference point when building a new campaign. I recommend comparing broad audiences against more defined interest-based audiences; this will provide insights into which approach garners better conversion rates for your business.
I often refer to three categories for interest-based audiences: direct, wide, and psychographic. Direct interests are aligned exactly with the product. For a beauty brand that sells skincare products, a direct interest could be people whom Meta determines are interested in moisturizers or face washes based on engagement or online shopping behavior. Wide interests are a more general match for the product and could include people who have demonstrated a broader interest in beauty products, such as makeup and other cosmetics. Psychographic interests, meanwhile, are the broadest category and could include an audience interested in fashion magazines or fashion accessories.
Meta Ads Manager allows marketers to select interests like this when setting up a campaign or ad set. It’s as simple as studying the list of interests available in the tool and checking the desired boxes. I recommend regularly evaluating and comparing different interest-based targeting methods on otherwise identical ad sets. The side-by-side comparison lets you adjust your strategy based on performance data and optimize for reach and engagement.
Enrich Data for Precise Ad Targeting
Data enrichment is another essential optimization for growth marketing on Meta platforms. This is the process of combining audience lists or adding more data so that the platform can improve targeting. Suppose a company has uploaded a list of customers to Meta, providing only their names and email addresses to create a lookalike audience. An enriched list could include revenue associated with each customer and their ZIP code, providing a clearer picture of a high-value customer.
Data enrichment is often a manual process of cross-referencing and combining lists, but you can boost and automate it with a tool like Retention.com, which can identify anonymous website visitors and associate email addresses and other data for more specific retargeting. Because Meta is no longer as effective at tracking off-platform web traffic, especially on Apple devices, this third-party tool can be especially powerful in identifying a significant segment of website visitors who would otherwise slip past you. Retention.com also uses its enrichment algorithm to associate visitors with additional attributes, such as online purchasing behavior. Marketers can add these lists to Meta Ads Manager through a direct integration that protects user data. In my experience, data enrichment and integrations with tools like Retention.com once again provide an improved cost per acquisition in the 20% to 40% range.
Implementing these targeting strategies can be complex, requiring a deep understanding of digital marketing nuances and continuous adaptation to the changing digital landscape. Once you identify targeting optimizations that meet your business objectives, it’s time to scale the approaches. As you do so, maintain your testing volume by allocating 10% to 30% of your paid media budget to experimental concepts. The tests can run as separate campaigns with independent budgets. As winners emerge from your testing, migrate those into your evergreen campaigns.
Over time, personalization with advanced targeting is a surefire way to boost return on investment for Meta campaigns. Key metrics, such as CPA and conversion rate, will improve. Remember to continue testing new targeting strategies, even as you scale your winning campaigns. Don’t rely on strategies from previous campaigns to continue resonating indefinitely. Personalization isn’t a once-and-done operation. After all, most consumers expect personalization from companies—and even more importantly, they expect personalization to be done well.