We Fact-Checked 5 Viral Meta Ads 'Hacks' Against Our Own Client Accounts

"Half the stuff I read about Facebook ads contradicts the other half. How am I meant to know what's actually true?"

A founder said that to me on a call last month, half-laughing, half-exhausted. And honestly, fair. The Meta ads internet is a firehose of confident men in nice studios telling you the one trick that changed everything. Some of it is real. A lot of it is noise dressed up as signal.

So we did the boring thing. We took five of the most viral "hacks" doing the rounds and checked each one against what we actually see in client accounts. Real budgets, real spend, across a spread of brands including one homewares account we've ridden from roughly $24k to ~$190k a month.

I went in expecting to debunk most of them. That's not quite how it turned out, and the surprises are the interesting bit. Here's each, with the verdict.

1. "Farm comments to crush your CPMs"

The claim: Get people commenting on your ads, even with staged arguments and dog photos, and Meta reads it as engagement, bounces the post around, and your CPMs collapse. Some versions of this promise a 160x reduction in cost. Yes, 160x.

What we found: Nothing. We've never once seen comment volume drive performance, and the maths on those claims falls apart the second you look at it. A 160x drop on a $10 CPM is a six-cent CPM. That's not a strategy, that's a typo.

Here's the thing people get backwards. They see an ad that wasn't doing much suddenly pick up comments and then take off, and they conclude the comments caused it. The causality runs the other way. The ad got good, so it earned comments and conversions at the same time. Manufacturing the comments doesn't reverse-engineer the result.

The old "PPE wrapper" trick, running a tiny engagement campaign alongside every conversion campaign to juice the comments, made the rounds years ago for exactly this reason. If it worked, every serious buyer would still be doing it. None of us are. And worse, garbage engagement teaches Meta the wrong signal. You don't want comments. You want engagement that ladders to a purchase, which is a completely different thing.

The call: Ignore it. Make a better ad. The comments are a symptom of a good ad, not the cause.

2. "You need a dedicated testing campaign"

The claim: Run a separate "testing" campaign to trial new creative away from your scaling campaign, so new ads can warm up without competing against your winners.

What we found: This one is messier than the gurus on either side admit, and I'll be honest about where I've landed.

The harm is real and it's cultural, not technical. The moment a campaign is labelled "testing", everyone gives themselves permission to waste money in it. Nobody does the cost maths. I watched one debate where the punchline was perfect: if your test campaign is quietly spending $30k, the right question is "was I willing to pay $30k to learn that?" Almost nobody asks it. They just shrug because it was "only testing".

But here's where I won't fully debunk it. There's a defensible version. If your testing campaign runs the exact same bid as your scaling campaign, all you've really done is create a separate environment for new ads to gather data before you move them. That's not nothing, and a few people at Meta have suggested the learning a new ad accrues in a quieter environment can travel with it. We've played with both. The version that bleeds money on a loose bid is a genuine mistake. The version with the same disciplined bid is closer to "harmless, possibly mildly useful".

The call: If you run one, run it on the same bid as your scaling campaign and put a cost cap on it. If "testing" is your account's word for "this is allowed to lose money", kill it. This is one I half-expected to throw out completely, and I couldn't. The honest verdict is "fine if disciplined, dangerous as an excuse".

3. "Never turn off an ad. Stop judging at the ad level."

The claim: Don't switch off an ad just because its own ROAS looks ugly. Meta optimises for the whole ad set or campaign, so a single ad with a terrible cost per purchase might be doing useful work that shows up elsewhere.

What we found: This is the one I most expected to be guru nonsense, and it's the one that's held up best. That surprised me, so I want to be careful with it.

There's a real mechanism underneath. Meta isn't trying to win at the individual ad level. It's optimising the ad set or campaign toward the aggregate result. So it can pour most of a budget into one ad with an unimpressive CPM precisely because doing so lets the rest of the ads in the set close conversions. Judge that one ad in isolation, switch it off, and you can knock over the whole structure for reasons you'll never see in the columns.

We've watched exactly this. An ad with a CPA that looks like it should be cut is sometimes load-bearing, and accounts that twitch and pause at the ad level tend to be more volatile, not less.

That said, "never" is too strong, and the gurus oversell it. You still cut an ad that's genuinely going haywire, eating spend and driving nothing. And the related "just relaunch losing ads with a new thumbnail to reset them" claim is a trap. The ad is the ad. A new thumbnail can technically register as a new unit, but a weak ad stays weak.

The call: Stop reacting to ad-level ROAS. Judge at the ad set and campaign level, give the algorithm room, and only intervene when an ad is clearly broken, not just unflattering. I went in sceptical and came out converted on the principle, wary of the absolutism.

4. "Lowest-cost bidding is fine. Throw all your products in one campaign."

The claim: Just set everything to lowest cost, the default, drop all your products into one campaign, and let Meta sort it out.

What we found: This is the most common setup we see, and it's quietly costing brands money.

Two problems. First, lowest-cost bidding optimises for cheap acquisition, which is not the same as optimising for value. Value-optimised bidding reaches a different customer at a different price, and when we test the two side by side, the audience overlap is routinely under 10%. A Meta rep will often tell you to consolidate them to avoid overlap. We've checked. The overlap isn't there. Running both reaches people you'd otherwise miss.

Second, mixing products with very different margins in one lowest-cost campaign makes your ROAS lurch around. Meta spends to your target cost per purchase, but if a $100 item and a $50 item sit in the same campaign, spend piles onto one and your reported ROAS bounces day to day. People then blame bid caps for being "volatile" when the real cause is the mismatch they built.

The nuance the simple version misses: separate by margin profile, not just by price. AOV is a clue, but margin is what you're really steering.

The call: Don't default to lowest cost. Run value optimisation, and split campaigns by margin profile so your bids and your ROAS hold steady. This one debunked cleanly.

5. "The learning phase forces you to scale in 10% steps"

The claim: Respect the learning phase, so only ever raise budgets by 10 to 15% at a time or you'll reset everything and tank performance.

What we found: Partly a hoax, and the part that's a hoax has a hidden cost almost nobody prices in.

The trap isn't the 10% rule itself, it's what it hides: the cost of inaction. When you only nudge budgets up 10% at a time, your best ads are still slowly fatiguing in the background, just slowly enough that you don't notice. Every cautious week is a week a good ad quietly decays while you sit on it.

A chunk of the learning-phase fear is really about how long impressions take to convert. Jump your budget meaningfully and yes, the new impressions need time to mature, so the next stretch feels rough before it settles. That patience gap is what scares people back into baby steps.

So the honest position is in the middle. If your purchase window is long, ease up more carefully. But if your incremental attribution says you can scale, and especially if your purchase window is short, you can move far more aggressively than the 10% gospel allows, you just have to hold your nerve through the lag.

The call: Don't treat 10% as a law. Let your purchase window and your incremental numbers set the step size. Half-true, and the cost of being too cautious is bigger than the cost of being too bold here.

The scorecard

Tallying it up: two debunked outright (comment farming, lowest-cost-everything), one that's situational (testing campaigns), and two I walked in ready to dismiss and couldn't, the ad-level "never turn off" principle and the learning-phase pushback. The honest takeaway isn't "gurus are always wrong". It's that the confident absolute is almost always where the danger lives. The truth usually has a condition attached.

If you want to run this exercise on your own account, it's genuinely worth an afternoon: pick the three "rules" you currently obey without questioning, and check each against what your own numbers actually say. You'll likely find one you've been following religiously that's costing you. And if you'd rather talk one through, reply and tell me. I'm curious whether your accounts say the same as ours, so which of the five surprised you most?