Latest / Social Media Marketing with Fexingo: Instagram, TikTok, LinkedIn, and Platform Strategy / How Social Platforms Changed The Ad Auction Model
Transcript
- Lucas: Look, if you have been running paid social ads for the last five years, you know that the game has changed more than it has in the previous twenty combined. As of October 18, 2026, the old playbook of manual bid caps and simple cost per click targets is effectively dead. We are no longer buying inventory; we are participating in a predictive value auction where the platform's algorithm decides your worth based on who you are targeting, not just what you are paying. Luna: Right, and most brands are still trying to use a key from 2021 to unlock a door that was welded shut three years ago. They are confused why their ROAS keeps dropping even though they doubled their budget. Lucas: Exactly. The core issue is that platforms like Meta and Google have shifted from being passive distributors of ad space to active partners in your conversion funnel. They want you to hand over control of your bidding strategy because they trust their machine learning models to find high-intent users better than you can. It is a surrender of tactical control for strategic scale, but only if you feed them the right signals. Luna: So the friction isn't about reaching people anymore, it is about proving to the algorithm that your product actually solves a problem for those people once they arrive. Lucas: Precisely. Let us look at the mechanics. In a traditional auction, you might bid ten cents for a click. If you get ten thousand clicks, you pay one hundred thousand dollars. Simple. But today, the auction includes a value component. The platform estimates the probability that a user will convert, purchase, or engage deeply. Your effective bid becomes your maximum bid multiplied by that predicted conversion rate. So a brand with a low bid but a high conversion prediction often wins against a brand with a higher bid but poor historical performance data. Luna: That makes sense, but it feels risky for smaller advertisers who do not have massive datasets to train those initial models. How do they compete without that historical weight? Lucas: They cannot compete on volume, so they compete on specificity. The winners in 2026 are using hyper-specific audience segments with clear, single-goal objectives. Instead of a broad awareness campaign, a DTC apparel brand might target a narrow segment of past purchasers looking for winter gear, optimizing strictly for add to cart events. By narrowing the funnel, they give the algorithm fewer variables to guess, which actually increases confidence in the prediction. It is less about shouting louder and more about whispering directly to the right ear. Luna: And that requires a level of creative discipline that most teams lack. You cannot just throw generic lifestyle imagery at a pixel and hope for the best anymore. Lucas: Correct. Creative is now your primary lever for signaling quality. Platforms analyze the visual and textual elements of your ad to predict its appeal before it even hits the feed. If your creative looks like a native organic post, the algorithm flags it as high-quality content. If it looks like a polished, salesy banner, it gets penalized in the auction ranking. We are seeing case studies where brands improved their cost per acquisition by forty percent simply by changing their creative format to match the aesthetic of the surrounding feed. Luna: It is almost like you are hacking the algorithm by respecting the user experience rather than interrupting it. Lucas: That is the perfect way to put it. User experience is now a direct economic factor in your ad spend. When a user scrolls past an ad quickly, that negative signal drags down the predicted conversion rate for that creative asset. The platform learns that your ad is annoying, so it stops showing it to high-value users. You are essentially being taxed for bad creativity. It sounds harsh, but the data supports it. Brands that invest in creator-style content see lower customer acquisition costs because the platform rewards the engagement. Luna: So the incentive structure has completely flipped. Previously, the goal was to grab attention. Now, the goal is to blend in so well that the user doesn't mind engaging. Lucas: And blending in requires authenticity. This brings us to another critical shift: the decline of third-party cookies and the rise of first-party data. Since iOS updates and privacy regulations have limited tracking, platforms are struggling to attribute conversions accurately. To compensate, they are asking brands to share more of their own customer data. This is where value-based optimization really shines. When you upload your customer list to Meta or Google, the platform can look at those existing customers, find their profiles, and then go out to find lookalikes who behave similarly. Luna: Which is basically building a digital twin of your best customers and telling the platform to hunt exclusively for clones of them. Lucas: Yes, but it is more nuanced than just cloning. You need to segment that data. Sending a list of your highest lifetime value customers tells the algorithm to seek out whales. Sending a list of your frequent small buyers tells it to seek out volume. The mistake many brands make is uploading a mixed bag of data and wondering why the algorithm goes crazy. It needs clean, segmented inputs to generate clean, targeted outputs. Luna: I see. So data hygiene is just as important as data volume. A messy list is worse than no list at all because it trains the bot incorrectly. Lucas: Absolutely. A messy list introduces noise. If you include churned customers or inactive users in your seed audience, the algorithm might start finding people who look like former customers but don't have the same purchasing habits. You end up paying for attention from people who will never buy. It is a subtle but devastating leak in your marketing budget. We see agencies fixing this exact issue by cleaning seed audiences quarterly, which alone improves ROAS by fifteen to twenty percent. Luna: That is a huge number for something as mundane as cleaning a spreadsheet. It really highlights how much the backend infrastructure matters now. Lucas: It does. And while we are talking about backend shifts, we have to mention the rise of conversational commerce. Platforms like WhatsApp and Instagram Direct are becoming major channels for closing sales. The ad auction now extends into these private messaging spaces. If your ad drives a message conversation, the platform values that interaction highly because it indicates strong intent. Brands that set up automated chat flows to answer questions and close deals within the app are seeing higher conversion rates than those driving traffic to a website. Luna: Because you remove the friction of the checkout process entirely. No page loads, no form filling, just a conversation that ends in a purchase. Lucas: Exactly. And the algorithm knows this. When you optimize for messages, you are competing in a different subset of the auction. It is less crowded than search ads but highly competitive for premium placements. The key is to have a human-in-the-loop response system. Pure bots can frustrate users, but pure humans cannot scale. The sweet spot is automated triage followed by human intervention for complex queries. This hybrid approach keeps the cost per lead low while maintaining the high-touch feel that drives sales. Luna: It feels like we are returning to the golden age of sales, where personal connection mattered, but with the efficiency of modern technology handling the heavy lifting. Lucas: That is a great analogy. It is high-tech, high-touch. And this leads to my final point on adaptation: measuring success differently. Traditional metrics like clicks and impressions are largely vanity metrics in this new environment. You need to focus on return on ad spend, but also on contribution margin. Because the cost of acquiring a customer is rising due to the complexity of the auction, your profit margins must be protected. If you are selling a low-margin item, the new ad models may not work for you unless you can dramatically increase average order value through bundling or upselling. Luna: So the business model itself has to change to fit the marketing model. You cannot run a thin-margin, high-volume play with expensive, value-based ads. Lucas: Correct. You either need high margins to absorb the rising customer acquisition costs, or you need a subscription model that recoups the loss upfront. The era of losing money on the first sale and hoping for repeat purchases is fading because the cost to acquire that first sale is now too high. Brands are having to rethink their unit economics fundamentally. It is a brutal filter that separates sustainable businesses from cash-burning ones. Luna: It really forces a clarity of purpose. You have to know exactly what your customer is worth to you, not just what they click on. Lucas: And that clarity allows you to bid confidently. When you know your customer lifetime value, you can afford to bid higher in these value-based auctions because you know you will make it back over time. It shifts the mindset from short-term ROI to long-term LTV. That is the real secret weapon of successful brands in 2026. They are playing chess, not checkers. Luna: Chess, not checkers. I love that. It implies patience, strategy, and seeing several moves ahead, which is exactly what the current landscape demands. Lucas: Before we wrap up, I want to mention that if these marketing conversations have sparked something you've actually used, like maybe a new ad template or a data cleaning workflow, listener support helps keep this show ad-free and independent. Honestly, if today was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo. Luna: Yeah, we appreciate the support. It allows us to dig deeper into these technical shifts without chasing sponsorships that might compromise our analysis. Lucas: Thanks for listening. As you look at your next campaign, ask yourself: are you feeding the algorithm good data, or are you just throwing money at the wall? The difference is usually in the details.