Be Smart: The Added Value of Third-Party Audiences
Third-party audiences are a vital tool for narrowing down advertising campaigns to specific, highly-relevant users that closely align to a company’s message or goal. Built on a wealth of data across online, offline, purchasing, and additional information, these audiences provide a more complete picture of user bases and can be deployed to reach audiences that will have a much greater resonance with specific branding or products. In theory, the larger the information synthesis in the third-party data set, the more accurate and refined the audiences can be.
If third-party audiences are truly this powerful of a tool, why aren’t they used over standard audiences on a regular basis?
On average, a couple of issues prevent this possibility: limited audience scope and data costs. Third-party audiences, due to their specificity, are on average smaller than broader standard audiences. For example, a standard audience such as “Birdwatchers” would have a much larger audience size and reach than a third-party “Eastern Goldfinch Birdwatchers”. In addition, lookalike and WCA audiences have very limited scope with third-party audiences, as they are inherently drawing from a smaller sample size to create their customized spinoffs. Third-party data is also not cheap due to the amount of disparate data sources that need to be merged together to create accurate and reliable user profiles. This cost is passed along to customers (i.e. advertisers) in order to turn this hard work into positive margins. Think in terms of a S&P 500 ETF versus a managed fund: a managed fund may perform quite well against the S&P 500 benchmark, but once costs and fees are added in, many managed funds struggle to consistently clear the non-managed alternatives.
So, do the inherent strengths of third-party audiences outweigh their costs and limitations? We took a look at our internal data over the running lifetime of HYFN’s Facebook and Instagram client accounts across various verticals and objectives to see how third-party audiences compare to standard audiences. We limited the data to periods where standard and third-party audiences ran concurrently in order to remove standard-only periods from the overall performance of standard audiences. Finally, we ran the data through our internal quantitative spend-on-objective factor model in order to statistically separate out the spend-based performance of standard versus third-party audiences across seven different campaign objectives: Reach, Purchases, Non-Purchase Conversions, Link Clicks, Video Views, Estimated Ad Recallers, Post Engagements, and Store Visits. We then looked at the performance of standard versus third-party audiences against each other on-average, both with and without a fee (our fee was calculated as a $2.5k/month cost). Here are some of the major results we discovered.
Due to the limited scope versus their broad counterparts, scalability isn’t easily achieved within third-party audiences, resulting in high objective rates but low objective volume and high cost per objective. Purchases, Non-Purchase Conversions, and Post Engagements are the most prone to this effect (Figure 1).
HYFN’s recommendation: In these cases, third-party audiences are best used as a supplement to a diverse targeting strategy that includes a range of audiences from broad to specific.
FIG 1: DAILY CONVERSIONS
The Ad Recallers Brand Awareness objective is affected by this as well.
This makes sense, given that Brand Awareness is going to be greatly hampered by a low-user-volume audience size, correlation overlap notwithstanding (Figure 2). While closer than its DR counterparts, the standard audiences are still driving higher volume at a lower cost than third-party on average for this objective.
HYFN’s recommendation: Again, coupling third-party audiences with a spectrum of effective broad audiences is the best route forward, especially if your Ad Recall campaign is part of a large full-funnel strategy that hopes to turn these ad recallers into future purchasers and/or converters.
FIG 2: DAILY AD RECALL
Most other objectives deliver higher volume and lower Costs per Objective…before fees.
While standard audiences outpace third-party audiences at lower spend levels, Brand Awareness objectives like Reach and Video Views and DR’s Store Visits show a higher maximum volume and lower costs per objective at these levels. However, the added costs from fees eliminated any inherent cost efficiencies in the third-party audiences (Figure 3), leading to the customer to make a choice: use lower-volume, lower-cost standard audiences, or use higher-volume, higher-cost third-party audiences. This will clearly be a customer-by-customer decision, based on the customer’s margin and additional business math.
HYFN’s recommendation: Unless it makes sense for a full-bore efficiency or a full-bore volume strategy, we’d consider counterbalancing as an initial play to see whether the standard or third-party audience type responds best to the user base. More often than not a ratio between standard and third-party is achieved with this message, with tilts to one strategy or the other depending on the influx of new creative, product, etc.
FIG 3: DAILY VIDEO VIEWS
Clicks is the sole objective where third-party audiences hold a clear advantage.
While standard audiences are more cost-efficient than fee-included third-party audiences at low spend, third-party audiences show a significant gap in volume at a slightly more-efficient cost at more-optimal spend levels (Figure 4). On average, clients see an improvement for third-party audiences over standard audiences when it comes to the Link Click objective. This is mostly due to Link Click rates being less affected by audience size or correlation and more affected by creative, product, or messaging.
HYFN’s recommendation: If you’re running a Clicks campaign that contains a dynamite creative or message, tilt spend into third-party audiences until you see a clear decrease in cost per click.
FIG 4: DAILY CLICKS
Remember: These results represent performance on average.
This is something that can’t be stressed enough: the results we’re discussing represent an equal-weighted average of our data set. Brand performance will be distributed around these model curves based on a variety of factors. Third-party audiences will greatly outperform expectations for some brands, but third-party audiences will also greatly underperform for other brands. It boils down to a question of risk-reward: can your brand afford to take a loss in efficiency if it doesn’t achieve better-than-average performance after fees? For large Fortune 500 companies the fees may be marginal and worth the risk, but for a new-to-social advertising SMB trying to increase market share through a social ad strategy, this may not have enough utility to be a viable path.
Overall, third-party audiences are an impressive data engineering feat that can blend together a wide variety of consumer information sources to provide a collection of users that fit into quite specific categorizations. While there are inherent strengths in hyperspecificity of audiences, there are also drawbacks: the narrow categorization can potentially lead to difficulties in scale and cost inefficiencies, and the inclusion of fees that are necessitated by the labor needed to construct said audiences can push up overall costs to a inefficient total. Based on our findings, third-party audiences have limited use cases as a central strategy for brand campaigns, but can serve as a supplemental service in a diverse and flexible deployment. Think of it as a powerful tool in a well-stocked tool chest: extremely effective when deployed properly in ideal use cases by someone who has the knowledge and experience to know what tool is best suited for a particular obstacle. Just as you wouldn’t pay a carpenter to nail together 2x4s with a shiny new sledgehammer, you wouldn’t want to deploy a social ad strategy centered around the wrong tool for the goal, no matter how powerful the tool is. Shiny tools are great, but will always be outpaced by an adaptive, flexible, and diverse brand strategy.
Director of Analytics at HYFN