Prof. Aleks Farseev is an Entrepreneur, Research Professor, Keynote Speaker and the CEO of SoMin.ai, a generative AI Marketing Suite.
Within the ever-shifting world of advertising giants like Meta and TikTok, the ebb and flow of ad performance week by week is a challenge known all too well. In this dynamic landscape, marketers frequently grapple with a perplexing scenario: one day, a particular ad set or creative dazzles with exceptional results, only to witness a cost surge unexpectedly on another day. This unanticipated volatility serves as a poignant reminder of the intricate and ever-evolving nature of digital advertising ecosystems. Yet, the industry’s tribulations don’t stop here.
While some marketing teams may accept such fluctuations as par for the course, others raise doubts concerning the proficiency of performance managers or the very efficiency of the channels themselves. These uncertainties give rise to frustration among all stakeholders, with the underlying problem often left unresolved.
I firmly believe that the crux of digital advertising success lies in a profound comprehension of the fundamental principles governing ad volatility. In this article, I aim to shed light on the most pivotal principles I’ve learned as a founder in the AdTech space.
AI In Control
For advertisers, the challenge lies in the lack of external control over this process, as it is intricately tied to Meta’s impression distribution AI and a user’s recent profile and browsing history within a specific market, country and interest category.
To put it differently, if an individual was categorized under an interest like “burger” because they took a selfie at McDonald’s a week ago, they could find themselves reclassified under an entirely different interest in the coming week. Consequently, there’s no assurance that the same audience will continue to view the ads, even if they were initially placed in the same ad interest category.
But we also must not forget that the whole process of bidding for a particular user impression with particular content and targeting is an AI probabilistic inference process. Imagine this scenario: I have a captivating creative (let’s call it “Creative A”) featuring an adorable kitten. It’s been crafted to sell a specific product (pet food, for example), and we’ve targeted it toward a particular audience (people who own pets) via a cost-per-purchase optimization goal.
Let’s say the AI tells us that there’s a 47% probability of it performing well. Now, here’s where things get intriguing. For another advertiser using the same exact AI models, TikTok might decide that their creative has a 48% chance of success with the same audience and impression. This suggests that they might secure that impression at a lower cost per thousand impressions (CPM) or that your bid would even not participate in the auction as you are “less likely” to provide the “relevant” ad to a user.
But who’s behind these probabilities? It’s the finely tuned models. Whenever these models change or optimization objectives, products, creatives or even the time of day shift, the probabilities can realign themselves. Suddenly, a once lackluster ad could start delivering outstanding results or vice versa. In this ever-evolving landscape, adaptability and staying ahead of the curve are paramount to success.
You are not the only one.
In the intricate landscape of advertising, a fundamental principle holds true: as you narrow down your target audience for better compatibility with your content, the pool of potential viewers diminishes. However, when you decide to allocate more budget, you signal a willingness to compromise and reach a broader audience, even if they don’t perfectly align with your desired user actions. In response, Meta adjusts its algorithm to gauge user engagement probabilities, expanding your reach but diluting the conversion funnel, resulting in a less responsive and cooler user base. This delicate balance between specificity and reach is a constant challenge for advertisers seeking broader visibility and impact.
It’s a trade-off advertisers must navigate, where a broader audience might mean more eyes on your content, but potentially at the cost of a less targeted and engaged user base. Balancing the scales in this supply-demand tango is the key to optimizing your advertising strategy.
Here we’ve described two of the most common problems of social advertising—price volatility and the inability to scale the ads due to warm audience size limitations. It is always easier to name the problem than to solve it, so what can we do about it?
To mitigate these issues, we must shift our focus beyond primary campaign metrics and delve into secondary indicators. For instance, when optimizing for lead generation, don’t solely track the cost per lead. Be sure to also monitor the click-to-lead conversion ratio. A decline in this ratio signals that your product may be losing relevance to the target audience in the lower funnel. This could prompt a reevaluation of your audience or necessitate changes to your offer, action buttons or product.
Climbing up the funnel, assess the conversion rate from click to landing page view. A decrease in this metric might indicate that your audience is clicking for reasons other than engaging with your content or that your landing page is plagued by loading issues. Swiftly address these concerns by fine-tuning the landing page.
Occasionally, a decline in click-through rates (CTR) may surface, hinting at ad fatigue, a mismatch between creative and audience or subpar content quality. In such cases, reassess your content quality.
Lastly, abrupt spikes in CPMs can signify platform decisions on Ad relevance or backend glitches. In those cases, consider pausing campaigns or duplicating them to prompt a system “reboot.”
To excel in performance social advertising, you don’t need to be a data scientist or a TikTok lead engineer. What truly empowers you is being well-versed in the intricacies of this dynamic landscape. In my view, the secret lies in not regarding the “Metas” of our digital world as mystical AI entities, isolated from reality. Instead, it’s about becoming a connoisseur of the inner workings that guide your decisions at every juncture of your digital advertising odyssey.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?