When paid social tanks, I can predict the conversation before anyone opens the dashboard. Someone sighs, stares at the numbers, and blames the algorithm. Meta changed something. iOS wrecked everything. TikTok is a chaos machine. The platform is rigged and we’re just along for the ride.

I’ve been running paid social for 10+ years now across entertainment, CPG, B2B, and franchise brands—and managing budgets from “we have $5k to test this” to $10M+ annual programs. From that seat, here’s the uncomfortable truth: the algorithm absolutely matters. It has real power. It shapes what you pay, who you reach, and what gets delivered at all. But most of the time when performance falls apart, the algorithm is not the villain. The strategy is.

And more specifically: it’s the strategy of pretending paid social exists in a vacuum—that if we just “hack” Meta, or outsmart TikTok’s delivery, we can brute-force our way to performance. That’s not how this game works. The algorithm is one input in a much larger system. You don’t beat it. You learn how to work with it and you fix the structural stuff that’s actually holding you back.

1. The Algorithm Is Powerful — But It’s Not Magic

Let’s start with the part everyone wants to argue about: the algorithm itself. Meta’s auction, TikTok’s content graph, Pinterest’s intent signals, LinkedIn’s audience modeling—these are not neutral pipes. They are opinionated systems designed to maximize their own objectives (time on site, ad revenue) while approximating yours (results, relevance).

Meta’s system is ruthlessly math-driven: it matches your bid, your predicted conversion rate, and your estimated value to decide who wins the impression. iOS signal loss changed what signals it can see, which means what it can optimize toward. TikTok doesn’t care about your funnel diagram; it cares whether people watch, rewatch, and share your content, and it will deliver accordingly. If you don’t understand these mechanics, you’re essentially driving with the parking brake on.

Respecting the algorithm means acknowledging what it’s good at—pattern recognition at insane scale, real-time auction decisions, modeling off incomplete data—and what it isn’t built to do for you. It will not fix a broken offer. It will not invent demand in a category no one understands. It will not override a terrible user experience or misaligned KPI. You can’t “optimize” your way out of strategy debt.

2. The Real Performance Killers Are Strategic, Not Technical

When campaigns underperform, the easy story is: “Meta’s delivery is trash right now,” or “TikTok fell off a cliff after the last update.” Sometimes, sure, there are platform wobbles. But the patterns I see inside $10M+ programs are a lot less glamorous: basic strategic and structural problems that no amount of machine learning can solve.

Funnel gaps. You’re running hard at conversion with zero awareness or education in market. The algorithm is scraping the same small pool of in-market users over and over while leadership wonders why scale is painful and expensive.

Siloed channels. Paid social is chasing one narrative, search is running on a different planet, email is blasting random promos, and the site experience is an afterthought. From the customer’s perspective, it feels like five brands, not one. From the algorithm’s perspective, your signals are noisy and inconsistent.

Wrong KPIs. You say the goal is revenue, but you’re optimizing toward cheap clicks and thumbs-up metrics. The algorithm obediently finds you the people most likely to click or watch or comment—not the people most likely to buy. You asked for the wrong thing and then got mad when it delivered it.

3. You’re Giving the Algorithm a Bad Brief

Think of the algorithm like a brutally literal media planner. It will optimize toward exactly what you tell it matters. The problem is, most brands are giving it a bad brief—the wrong objectives, the wrong optimization events, and creative that sends mixed signals about who the ad is actually for.

CPM, CPC, CTR—those are inputs. They matter, but they’re not the destination. If you tell Meta your north star is lowest cost link click, don’t be surprised when you get low-intent traffic that bounces. If you obsess over view-through rates on awareness video but never ask whether those people progress down-funnel, you’re optimizing for entertainment, not outcomes.

Every stage of your funnel needs its own definition of success: awareness should be about qualified reach and recall; consideration should be about depth of engagement and signal collection; conversion should be about actual revenue and efficiency. When you collapse all of that into one number and call it ROAS, you’re not “data driven”—you’re just averaging away the story.

4. What It Looks Like to Work With the Algorithm and Own Your Strategy

I’m not here to sell you some secret bidding hack. The teams I’ve seen win consistently do two things at the same time: they respect how the algorithm works, and they build a strategy that gives it the best possible chance to succeed.

• They respect the machine: Clean account structure. Clear objectives. Enough conversion volume for stable delivery. Creative that aligns to how people actually use the platform (TikToks that feel like TikToks, not TV spots with subtitles). They don’t fight the algorithm’s need for signal.
• They build a real funnel: Top-of-funnel campaigns that actually introduce the brand and create demand; mid-funnel programs that educate, differentiate, and collect first-party data; bottom-of-funnel campaigns that close the loop and keep buyers in the ecosystem.
• They connect channels: Paid social, search, email, and site all telling the same story, in the right order, using shared audiences and consistent messaging. The algorithm sees coherent signals instead of chaos.
• They choose metrics like adults: Each stage has 1–2 primary KPIs that map to business reality, not just what’s easiest to screenshot for a deck.

In that setup, the algorithm becomes a powerful collaborator instead of a convenient excuse. It does what it’s built to do—find patterns, surface likely converters, optimize delivery—inside a system you’ve actually thought through.

That’s when every impression, click, and conversion actually earns its treat.

---

I’m Amanda Antosh, founder of 3 Dog Stare. I’ve spent a decade building and rescuing paid social programs for brands that care about more than cheap clicks. If you’re tired of blaming the algorithm and ready to fix the parts you actually control, let’s talk — and build something that deserves great performance.