If you want a Meta Ad to be successful, it needs to accomplish something very specific. It needs to generate:

Engagement that converts.

"Engagement" is the first key, because Meta's machine learning reads customer engagement with an ad in two important ways... 1. The thing most people understand: engagement signals to Meta that it's content that the customer likes. Meta likes this. Even more broadly, capturing someone's attention with your ad is good for your ad's performance.

2. The part people often miss:

People really hate my Meta Ads creative testing methodology.

But I remain convinced that bad testing methodology is one of the biggest ways ecom brands have operational inefficiency and financial waste.

Let me explain... 🧵 The core of my method:

Don't build creative testing campaigns. Instead, launch all new creative directly in the same campaigns as your evergreen content using bid caps (or other Cost Controls).

I know DTC is a struggle in this particular moment. And there is a new "ecom is on the brink of collapse" tweet every day.

But the best evidence is that the future of ecom is VERY bright.

Here are 7 arguments for why you should be optimistic about the future of ecommerce: 1. The world is quickly getting a lot richer.

Here's McKinsey showing global net worth and GDP growth (the gap between them is for another convo).

And U.S. Millennials are about to be the richest generation in history as they inherit Boomer wealth + enter peak earning years.

You can still build your SMS list before BFCM.

It's not too late, and the strategy is simple, low-cost, and low-effort.

Make no mistake: your SMS list matters for BFCM because SMS is the best channel for cutting through noise and reaching your customer.

So here's how to do it: The basic concept works like this: before any promotional moment (including but not limited to BFCM), email your customer file and tell them:

1. X big promotional event is coming.
2. Sign up for SMS so you don't miss out.

Here's an example from @bambuearth.

Want to get Black Friday Cyber Monday right for your brand?

Then there is nothing you need to be thinking better about than your offer. That's the foundation of BFCM success.

So here are my 4 keys to getting your BFCM offer right: 1. Know what you’re trying to accomplish.

There are basically three approaches to BFCM:

A. BFCM is not a big deal for my brand. We’ll leave these aside.
B. Maximize volume.
C. Maximize profit per order.

Facebook spend is decreasing this year because performance is down.

Right?

Maybe.

Or maybe for some brands performance hasn’t declined at all. It’s just LOOKED like it.

That was the case for a business I looked at recently. And the consequences were massive. 🧵 First, let’s be clear: FB spend IS down for most brands. @commnthreadco updates a data set from ~220 stores comprising $5B every week.

Look at FB spend this year. The pullback is real.

(Btw you can get this sent to you free every week: bit.ly/3s1qG7k).

(1/6) I'm against A/B testing.

Sort of.

I'm against most iterative A/B testing. I'm wildly pro "make giant changes" testing.

The first reason for this is that small changes create small impact. There are exceptions, but they're outliers. Which means you shouldn't target them. 2/ This is one reason @TaylorHoliday beats this drum a lot: most of us are simply not trying to squeeze out another 3%. We're trying to, say, double our revenue YoY.

Align your strategy with the outcome you're pursuing. Want big changes in outcomes? Then test big changes.

1/7 Test. The. Offer.

This is @FCGOODS. Click-only attribution.

The difference between these two campaigns:

"Evergreen" is regular pricing, where monogramming costs $25.

"Testing" is free monogramming. There's a margin hit here, but with this performance gap it's worth it. 2/ It's a slow time of year for FCG and performance has stagnated. So I thought, "Why not try free monogramming?" ~75% of our orders are gifts (I know this because of my @enquirelabs post-purchase survey), and gifts have a higher AOV than non-gifts because of monogramming.

1/13: How To Create Product Photography That Actually Sells On Facebook Ads

It's early, but this ad has taken the top spot as @FCGOODS' best performer. I'll share data as the sample grows, but on the first $1k spent, it's winning and the algorithm is prioritizing its delivery. 2// That's significant because it's even beating this ad, which has spent >$600k at a 2 ROAS (mostly prospecting) and has been unbeatable for more than a year for FCG: instagram.com/p/B_8CLpXD6i8/…

That the algo is prioritizing it over that ad is a really strong signal of success.

1/9: Don't Misunderstand Your Conversion Rates

Consider the results from these two emails @FCGOODS sent:

Email 1: Nearly Full List Send
Send Time: 6/11/20 (yesterday), Noon PDT
46k Recipients
12.5% Opens
2.1% CTR
19 Purchases
$4,649 Revenue 2/

Email 2: Tiny Fraction Of List
Send Time: 6/12/20 (this morning), 7:50 am PDT
220 Recipients
38.2% Opens
27.2% CTR
36 Purchases
$5,380 Revenue

What is happening here? Why did Email 2 perform SO MUCH BETTER than Email 1?

1/26 How We Improved @BambuEarth’s Facebook Ads Performance 250% In Less Than Two Weeks: A Thread

I was looking over the results from a recent win my team had and realized there were some replicable principles for other ecom folks.

So I wrote it up.

commonthreadco.com/blogs/coachs-c… 2/ Let’s start with the problem: the other day I tweeted a graph showing volatile (and recently declining) customer acquisition for Bambu Earth– and also showing that my team appears to have fixed it.

https://twitter.com/andrewjfaris/status/1247931294681419776

(1/22) How To Create Winning Facebook Ad Creative: A Thread

The below includes:

- my basic theory of how to win on Facebook Ads
- inspiration from the all-time great advertisers
- summary of why those ads work
- examples of how we're doing this 2/ Media-buying tactics are no longer the major differentiator on FB Ad performance. Advertisers win now by relying on algorithmic decision-making, which analyzes past data and projects future outcomes far better than a human can.

1/8 There is constant chatter about organic reach on FB & IG being inversely related to Page Likes. This observation from @TheBabylonBee is a classic example.

So what's going on? Does FB just want bigger pages to pay to play? Is this a big tech company cash grab?

I doubt it.

https://twitter.com/SethDillon/status/1171903142574854146

2/ My guess is that the "growing your page results in less total reach" theory (or any suggestion that FB actively forces an inverse relationship between those two) is a red herring for explaining this.

I'd bet organic reach has very little inherent relationship with Page Likes.

1/15 Using Influencer Giveaways To Generate Revenue With Conversion-Optimized FB Ads: My Guruiest Ad Strategy

One of my big ecom questions right now is how to REALLY get value out of digital influencer partnerships. Pay-for-post doesn't work (and I, like you, hate that content). 2/ Organic social reach is so low that I've come to think of influencers less as content distributors and more as content creators. 100k followers is nice, but if I'm only going to reach 5% of them, the CPM on pay-for-post sucks.

So what if I handle the distribution?

1/ 5 Steps To Solving Every FB Ad Account Problem

(NB: this actually applies to any paid media, but in this case, we're dealing with FB):

STEP 1: Isolate the timeline where the problem is. Recent ROAS decline, for example, like you'll see for one of my brands in #8 here. 2/ STEP 2: Pull your CPC, CR, and AOV. These are the only 3 variables in your revenue!

Inexplicably, @FBBusiness Ads Manager still does not have CR or AOV as columns on the dashboard. So you have to export cost, purchases, and revenue to a spreadsheet and calculate those.

1/ Another musing on this: to be truly data-driven, I think, means to fundamentally believe that data reflects reality, and to treat it with according interest, curiosity, and humility. 2/ I've found that lots of people compile and cite lots of data (sometimes because it is a job requirement), but they aren't actually submitted to it. They don't ask it any questions because they don't think it provides real answers.