Social Proof Might Not Matter After All

Does social proof not matter after all?

So this is one of those questions that I never would have given a second thought before. But I saw something that made me rethink everything I believed about social proof and Meta ads.

So, first up, what do I mean by social proof?

Social proof means a post, and in this context an ad, has received lots of comments, reactions, shares, and any other signals that prove value. Social proof is essentially saying that this post or ad or product is important or deserves attention.

We often talk about leveraging social proof, and I’ve certainly discussed it before. If you have an organic post that has a ton of engagement, it may be smart to promote it. Or if you have two identical ads, one that has social proof and one that doesn’t, the one with social proof will likely do better.

We even go so far as being scared of losing that social proof. So, don’t make edits to an ad with a ton of social proof because all of that engagement will disappear.

The assumption is that Meta’s delivery algorithm prefers ads with social proof, keeping costs down for such ads. Or it could mean that people respond more positively to an ad with social proof, which is what ultimately lowers costs.

But maybe we’re overvaluing the importance of this engagement.

I recently wrote a blog post about a weird drop in performance I encountered. I won’t go through the whole thing because it’s not relevant, but you can read about it at jonloomer.com/weird.

Anyway, there was a single ad that had accumulated a great deal of social proof. It had more than 6,000 comments.

Then, for no apparent reason at all, performance tanked. The reason was the subject of that blog post and it was an anomaly, but again, it’s not relevant here.

What I did was duplicate that original ad and made very minor edits. So I didn’t promote the existing post, and that ad would start from scratch.

The assumption was that this ad, if it fixed the weird drop in performance, would still take a while to return to the levels that the original ad was at. It may never get there.

The original ad had run for over a year, promoting a professional service. Before the unexplained drop in performance, it had actually peaked at $50 per lead. A year ago, it was over $60, and those costs were at their lowest in January and February.

So when I started the new ad, which looked exactly like the old ad but without the social proof, I had limited expectations.

And then, something interesting happened.

That new ad not only performed at the same level as the old ad, but it’s surpassed it.

The original ad, with more than 6,000 comments, peaked at a $50 cost per lead. But for the first two weeks of this new ad, it’s running at $46 per lead.

Not dramatically better. Probably within a range of randomness that doesn’t matter.

But what’s unavoidable is this: The new ad, starting from scratch, didn’t need that social proof after all.

There was no build up period, where the ad performed below expectations before gaining momentum. Despite the fact that it looked almost exactly like the old ad, but without the social proof, it immediately performed well.

Look, I’m not suggesting that social proof is never valuable. It absolutely could depend on the situation, the product, or the ad. And if given the choice, I’d still prefer the ad with social proof over the one that starts from scratch.

But this experience made me rethink everything.

So here’s the bottom of the glass.

Is social proof important? Maybe, but it’s also quite possible that we overvalue it.

Don’t assume that the loss of social proof will cripple results. Don’t allow the fear of losing social proof limit your options.

In the end, it’s the content of the ad that’s most important, not the number of reactions, comments, and shares.

And this experience is also a reminder to question our assumptions, particularly when it comes to things that are a bit vague.

Meta never told me that social proof mattered for my ads, I just assumed it. Years of conditioning led me to assume it to be true.

But sometimes, our real life experiences can surprise us.

If you want to read more about this experience and how a weird solution fixed an unexplained drop in performance, read my blog post. Go to jonloomer.com/weird.