Latest / Social Media Marketing with Fexingo: Instagram, TikTok, LinkedIn, and Platform Strategy / The Hidden Cost of Social Media Attribution
Transcript
- Lucas: There is a quiet crisis happening in digital marketing right now, and it has nothing to do with algorithm updates or ad fatigue. It is about how we measure success. Specifically, the way most companies track social media performance is fundamentally broken, and they are overpaying for it by tens of millions of dollars annually. Luna: That sounds dramatic until you look at the actual dashboards. Most teams are still trusting the last-click model, which gives all the credit to the final touchpoint and ignores everything that happened before. Lucas: Exactly. And because social platforms like Meta and TikTok are so aggressive with their own reporting tools, they have convinced the market that their view-through conversions are real purchases. But when you strip away those inflated metrics using incrementality testing, the true return on ad spend drops by nearly thirty percent for many mid-market brands. Luna: So we are essentially paying for ghosts? That explains why my clients keep asking if they should just cut their social budgets entirely. Lucas: Not cut them, but reallocate them based on reality rather than vanity. Let me give you a concrete example from a direct to consumer apparel brand we analyzed last month. They were spending forty thousand dollars a month on Instagram reels, and their platform reported a four point five return on ad spend. Luna: Four point five is huge. Why was that not real? Lucas: Because half of those sales came from people who saw the ad but never clicked, then bought three days later through an email newsletter they had already signed up for. The social platform claimed credit for the sale, even though the social content only played a passive role. When we isolated that traffic, the actual incremental lift was closer to one point two. Luna: That is a massive discrepancy. So the platform is taking credit for organic loyalty loops that would have happened anyway. Lucas: Precisely. And this is where the attribution model itself becomes a liability. Most marketers use last-click attribution because it is simple and built into every major dashboard. But it completely erases the upper funnel work that social media actually does. You stop seeing social as a discovery engine and start treating it as a checkout counter. Luna: Right. And since social feeds are interruptive, they are terrible at closing deals but excellent at starting conversations. Does this mean we need to abandon these platforms? Lucas: Absolutely not. It means we need to change how we define a conversion. The industry is slowly moving toward multi-touch attribution, where each step in the customer journey gets a slice of the credit. But implementing that correctly requires first-party data, which is becoming harder to capture as privacy regulations tighten. Luna: Which brings us back to the cookieless future. If we cannot rely on third-party cookies to track users across sites, how do we even know if that Instagram ad led to a website visit? Lucas: You build your own tracking infrastructure. The smartest brands are shifting toward modeled attribution combined with clean rooms. Instead of waiting for a user to click a link and trigger a pixel event, they are using probabilistic models to estimate the likelihood that a viewer converted, based on historical behavior patterns within their own walled garden. Luna: It feels like we are trading transparency for speculation. Isn't that risky for CFOs who want hard numbers? Lucas: It is risky, but staying with last-click is riskier. Consider the cost of customer acquisition. If you ignore the top of funnel impact of social, you will eventually bid too high on search terms that social actually influenced. You end up paying twice for the same customer once on social, once on search. That double counting bleeds margins dry. Luna: I have seen that exact bleed in retail accounts. The search team thinks they did the heavy lifting, but the social team planted the seed weeks earlier. Lucas: And neither team knows it. So they both optimize for their own siloed metrics. The solution is what we call cross-channel incrementality testing. You run controlled experiments where you suppress ads in certain geographies or demographics and compare the results against a control group. Luna: That requires a lot of data volume. Smaller brands might not have enough users to make that statistically significant. Lucas: True. But even small brands can use simpler proxies. Look at the time between ad exposure and purchase. If you see a spike in sales within two days of a major campaign launch, even without clicks, that is likely social influence. You do not need perfect data to spot the pattern. Luna: So it is less about precise attribution and more about recognizing correlation as a signal. But how do you convince a client to trust a correlation over a hard click? Lucas: By showing them the alternative. Show them what happens when you turn off social. Watch the search volume drop. Watch the brand mentions vanish. Watch the email open rates decline because the audience is no longer being reminded of the product. The absence of activity proves the presence of value. Luna: That is a powerful argument. It shifts the conversation from what social does directly to what it enables indirectly. Lucas: Exactly. And this is where the technology landscape is shifting again. With the death of third-party cookies, platforms are building their own identity graphs. Apple’s App Tracking Transparency and Google’s Privacy Sandbox are forcing everyone to adapt. We are moving toward a world where consent-based first-party data is the only reliable currency. Luna: Which means brands that ignore social community building are losing their ability to collect that consent. It is a strategic loss, not just a tactical one. Lucas: Spot on. If you are not engaging on social, you are not building the relationship required to get permission to market to people later. You become dependent on rented land and paid distribution forever. Luna: So the metric that matters most is not ROAS anymore. It is relationship equity. How much goodwill have we built that allows us to ask for a sale later? Lucas: Relationship equity is the new ROAS. And measuring it requires a different set of tools. We need to look at sentiment analysis, share of voice, and comment quality, not just click-through rates. These are softer metrics, but they predict long-term retention better than any hard conversion number ever could. Luna: It sounds like we need to hire anthropologists instead of data analysts. Or at least train our analysts to speak human. Lucas: Something like that. The data is there, but it is buried under layers of platform bias and outdated reporting standards. The brave marketers are digging it out. Luna: Before we go deeper into the tech stack side of this, I want to pause for a moment. If these marketing conversations have sparked something you've actually used in your own workflow, or if you find value in keeping this network independent, we deliberately don't run ads on these shows. Lucas: If you want to support that choice and help us keep producing detailed breakdowns like this, the link is buy me a coffee dot com slash fexingo. It keeps the lights on and the research deep. Luna: Thank you for listening. Now, let's get back to the attribution problem and how to fix it. Lucas: Right. So assuming you have decided to move away from last-click, what is the practical first step for a marketing director today? Do they rewrite their entire analytics setup overnight? Luna: No, that would be chaos. Start with a single channel audit. Pick your highest spend platform and run a ninety-day holdout test. Don't show ads to ten percent of your target audience and measure the difference in overall site traffic and branded search. Lucas: That is smart. It isolates the variable without breaking the rest of the machine. And once you have that baseline, you can adjust your internal reporting to reflect a blended view. Stop sending monthly reports that isolate channels. Send reports that show the customer journey. Luna: But how do you present that to stakeholders who are addicted to simple dashboards? Simple is easy to digest, even if it is wrong. Lucas: You present the cost of being wrong. Show them the money lost to double-counting. Show them the missed opportunities from underfunding top of funnel activities. Frame it as efficiency gains, not just accuracy improvements. CFOs care about efficiency. Luna: Efficiency is the key word. If we can prove that reallocating five percent of the budget to awareness increases overall conversion rates by two percent, the conversation changes instantly. Lucas: It does. And that is the shift we are seeing across the industry. From chasing clicks to optimizing for lifetime value. It is slower, harder, and much more accurate. Luna: Do you think smaller agencies can survive this transition? They rely on quick wins and simple reports to sell their services. Lucas: Some will struggle. But those that learn to offer incrementality testing and holistic strategy will thrive. The low-hanging fruit is gone. The value is in the complexity. Luna: Complexity is where the expertise lives. Okay, one last thought on this. If a brand is truly stuck in last-click mode, what is the smallest tweak they can make tomorrow to start thinking differently? Lucas: Add a UTM parameter that tracks the source of organic referrals. Often, social posts drive direct traffic that looks like organic, but it started with a social interaction. Capturing that nuance starts the journey toward better attribution. Luna: Small tweaks that reveal hidden truths. I like that approach. It is less about revolution and more about refinement. Lucas: Refinement is how you survive the noise. What do you think is the next big shift in how we will measure attention? Luna: I think we will see a move toward outcome-based pricing. Platforms charging for engaged time rather than impressions. It aligns their incentives with ours finally. Lucas: Aligned incentives. That is the holy grail. Until then, we keep digging through the data.