Are you feeling the Q4 crunch yet? It seems the ecommerce gods have decided to ride the 2020 wave of ‘too much of everything.’ Instead of giving us a normal October (as normal as it can be in the current situation), they have gifted us with yet another big sales event. No, marketers, it’s not just Singles Day followed by Black Friday followed by Cyber Monday followed by Christmas and Hanukkah in 2020! Welcome, Amazon Prime Day, to the Q4 insanity. I mean, I’ve always advocated for starting your Q4 campaigns early, but that was to give marketers some lead time, not to have a whole other thing added. Careful what you wish for.

The good news here is that the flood of ecommerce and affiliate content campaigns coming into Keywee over the last few weeks can offer us a peek into the kinds of campaign creatives that work, detached from your run-of-the-mill holiday promotional texts.

To that end, I took a look at what’s been working for ecommerce marketers, added in a bit of pumpkin spice in the form of our Marketing Language Platform, and tried to figure out what words and phrases have been driving performance. Let’s dive in.


Settings the Stage

To put this list together, the first thing I did was look for top ecommerce campaigns and identify common themes and words that perform well across different brands and products. Because CTRs and conversion rates can differ greatly between different brands and products, I used campaign spend as the key metric here. Campaigns that spend a lot are, clearly, campaigns that are working for marketers. 

But I didn’t stop there. I really wanted to put these approaches to the test, so I ran an experiment on our platform to see if our AI agrees with my assumptions. Some background: our Marketing Language Platform, apart from helping marketers ideate, optimize and generate text variations, also gives those variations a score. The score is a predictor of how well each variation will do in terms of engagement. The AI was trained on millions upon millions of social posts. 

The score given ranges from 0 – 100, or, as I like to think about it, brussels sprouts to Kim Kardashian. So when you’re working on a message, you’re not necessarily trying to hit 100, because hitting 100 usually involves some sort of breaking of the internet. What you’re actually trying to do is see how high you can score when creating text to promote your product. 

So when I saw a certain approach doing well, I checked that approach on our platform to see if it actually worked. I picked a nondescript product — headphones — and saw how the system graded every variation. My baseline text was so-so with a score of 43:

From there, I put the trends to the test to see if they improved the score. 

Confused yet? Don’t worry – if you’re not following you’ll see what I mean in a sec. 

On to the Results!

Here are the approaches that are doing well, and how they hold up to our AI’s scrutiny:

  1. Emoji. Ok, I know. Emojis are an easy ploy to get attention. The interesting thing I found was that not all of them worked. In fact, emojis worked best when they were related to the product sold. Winter coat = snowflake emoji. Jeans = pants emoji. So how effective was that type of related emoji for our headphone copy?
    Score! 6 points up from the original copy.
  2.  Calling out reviews or reviewers. In ecommerce, social proof is everything. So calling out a huge number of positive reviews, or quoting a review was incredibly common on our top performing campaigns. Let’s see how the AI measures it up:
    Mic drop. 18 points up!
  3. Your actions. It’s common knowledge that a call to action helps, well, drive action. The twist here is that among our top performers, this tactic worked specifically when using the word “your”. “Treat yourself,” “transform your wardrobe,” “make your kitchen brighter” and so on. Let’s see what our Marketing Language Platform had to say about it:
    Look at that, our AI concurs to the tune of a 14 point improvement.
  4.  Variety. Calling out choice seemed to be very effective amongst our top spenders. “A range of sizes,” “8 different colors,” etc. This approach tells the user that there’s more to explore. “You may not like this shirt in blue, but check out the 37 other colors we have”. So what’s the AI got to say about it?
  5. Numbered lists. Yes, yes — it’s a motif that we’ve seen over and over, but cliches are cliches because there’s usually a shred of truth behind them. I think people like numbered lists because they help tell them what experience they’re going to encounter post-click. If you come across an article titled “Why Justin Bieber is the Harbinger of the Apocalypse” then you may be wary about what you’ll encounter post click. On the other hand, “4 Reasons Justin Bieber is the Harbinger of the Apocalypse” tells you clearly – you can come in, read the subheaders, and figure out fairly quickly if 2020 is about to unleash some Bieber-sized horror upon us. Heck – I even used it in this subject line. And you’re here!  Let’s see what the AI says:
    Yep – 17 point improvement. Some things never change.

So there you have it – 5 data-backed, AI-confirmed approaches to help you amp up your ad copy as we head into what will be an unprecedented (sorry) Q4. 

As always, I’d love to hear what’s happening in your neck of the woods. And if you feel like taking Keywee out for a spin and seeing how it can help you craft and optimize your copy, drop me a line and I’ll set up a trial for you.