Here at WordStream we have access to a ton of ad data thanks to our free tools and client base. We’ve analyzed over $60 million in Google ads spend. And we love to play around with that data, giving you performance insights like average click-through rate, conversion rate, cost per click, and cost per conversion across all different industries.
But one thing we had never looked at before was the ad text itself. Obviously, all of us folks on the content team are obsessed with words. So we got to thinking: What are the actual WORDS that go into really great, super-high-CTR ads in Google Ads? And while we’re at it, what about stuff like numbers and punctuation marks? What are the best CTA’s? Are the best ads in Google Ads positive or negative? Are they keyword-stuffed or more creative?
In this article, we’ll cover nine, often surprising things we learned by running text analyses on some of the best ads in Google Ads from the past year.
This analysis was a group effort. Shout-out to everyone who worked on the project including our data analyst Josh Brackett, our web team leader Meg Lister, our designer Kate Lindsay, and content team leader Elisa Gabbert.
Here’s how we arrived at our set of top Google Ads ads. Our free Google Ads Performance Grader checks every account that runs a report to determine that account’s best ad (along with its worst ad)
We pulled all the data collected from Grader reports between August 1, 2016 and April 30, 2017, then looked at the top 15% of ads seen during that time period (judged by a combination of impressions, clicks, and CTR), which gave us 612 total ads to analyze.
We only looked at advertisers bidding in USD, AUD, CAD, and ZAR, and eliminated non-English ads and duplicates from that set. We also split the set into branded and non-branded ads (335 brand ads, 277 non-branded).
Note: We separated out branded ads (brand ads are the ads that show up when users search for a business’s brand name) because these ads naturally tend to have very high performance, including high click-through rates and Quality Scores. Why? Because people searching for a brand name are often just trying to navigate directly to your site; plus, they’re already heard of you, so there’s natural brand affinity (which can increase CTR by 2 or 3x). Therefore, we excluded brand ads from most of our analysis, so we could see what tricks and tactics lead to super high performance even when businesses don’t have the benefit of high brand affinity.
With that said, let’s get into the data!
Back in 2016, Google released Expanded Text Ads, which basically double your ads’ max character counts. Compared to their predecessors, Expanded Text Ads are a sprawling canvas. To some, having more characters to fill with ad copy is akin to torture; to others, it represents opportunity.
Regardless of which camp you’re in, it is a universal truth that more opportunities to use more words means that choosing the right ones is more important than ever. But which words are the “right” ones?
To find out, we looked at our set of top performing text ads and broke them into individual words, stripping out conjunctions (and, or) and articles (a, the) until we were left with the ten most popular words:
According to our data, in 335 high-performing non-branded text ads, we determined that the words appearing most often were, in this order:
“Your,” “Free,” and “Now” seem obvious, but “Shipping” (which, I’d assume, is typically accompanied by “Free”) making an appearance surprises me; it leads me to believe that our top performers lean more towards eCommerce than lead generation. Speaking of, the fact that “Get” and “Save” are the only verbs on the list makes total sense since they have application across both B2B and B2C text ads.
With both “Free” and “Save” on the list, we can assume that shoppers are as cost-conscious as ever. And “Now” and “Get” imply they want stuff fast.
Also really important: the second-person pronouns “You” and “Your.” Make sure your ad copy speaks to your customer; don’t make it all about you!
TL;DR: When you’re struggling to write your best ad copy, remember: “Your Free Now Get!”
The call to action, or CTA, is perhaps the most important component of your ad copy. If it doesn’t live in one of your headlines, you’re doing it wrong. If you can squeeze a reminder into the description or URL paths, do so. You’ll thank me later.
What’s a Call to Action?
In this context, it’s the bit of language you append to an ad that tells a searcher which action you’d like them to take (and how/when to take it). Knowing which is the right CTA for a given ad is very much vertical, audience, and offering specific. You’d never tell someone who’s about to score your ballin’ mid-funnel whitepaper to “Buy Now,” would you?
Now, we’ve excluded branded ads in our analysis of calls to action because oftentimes advertisers will fail to explicitly drive action (or actively choose not to) in branded ad copy. Instead, many advertisers use them more like text-based billboards atop the SERP; the goal is to tell prospects that you exist, not necessarily incite a click. While this is certainly useful to some degree, it doesn’t help us understand what makes non-branded (often more commercially relevant) text ads tick.
While phrasing will differ slightly on a case-by-case basis, the verbs that you leverage are pretty limited based on the action you’d like a prospect to take.The top non-branded text ads we analyzed revealed that the most popular CTAs were (by number of ads each verb appeared in):
If you sell a product, the obvious CTAs are “Buy” or “Shop” (perhaps “Get” if you’d prefer to downplay the required exchange of currency); conversely, if your goal is to accrue mad leads, it’s likely your CTA deploys the likes of “Learn” or “Discover” or “Sign Up.” If you look at frequency, though, you’ll notice that the most-used CTA verb in our data set is “Get.”
Nobody with a top ad is using the verb “Click,” as in “Click here.” (In fact, trying that will probably get your ad disapproved.
From schadenfreude to merriment, the sentiment of your ad copy plays an important role in how prospects consume and interact with your ads. This isn’t news. In fact, over the years we’ve talked about using emotional triggers to bolster your ad copy and landing pages ad nauseum: how many of you were listening?
We used the Vader sentiment analysis tool in concert with Python’s NLTK library to determine the sentiment (positive, negative, or neutral) evident in our set of top ads. If an ad returned a compound score of greater than .5, we called it a positive ad. Ads with a compound score of between -.5 and .5 were considered neutral. And finally, ads with a score less than -.5 were considered negative
According to our findings, ya’ll are all smiles, which made me frown.
45% of the text ads conveyed a positive sentiment; 53% conveyed a neutral sentiment; and just 2% conveyed a negative sentiment.
I am shocked by these figures, especially when you take into account that they only represent non-branded text ads.
It’d make total sense if the same analysis, conducted on branded ad copy, yielded the same results; after all, no brand wants to be seen as negative (see: dumb Pepsi commercial, every airline ever, etc.). But isn’t negativity — a measurable degree of sentiment — better than milquetoast copy?
Personally, I’d take ad copy that can incite a response (even a negative one) over that which resembles wet cardboard stuffed with keywords. Causing a prospect to experience sadness or fear or a sardonic giggle with your ad copy is better than going unnoticed, contributing to the SERP’s ever-growing white noise.
Now, I know what you’re thinking. “What the heck does lexical diversity mean?!” It means it’s time to get a little academic.
It refers to the proportion of unique words in a given corpus (set of textual data) to all of the words in the texts. When used as a lens through which to analyze and optimize text ads, measuring lexical diversity is a key indicator of keyword stuffing. It’s important to have a handle on how often you’re using keywords in your ad copy. Too little and you’ll derail your Quality Scores; too many, and you’ll piss people off.
Our data reveals that top advertisers tend to repeat just 2 words per ad. Like so:
Which is a helluva lot better than, say…
The moral of the story here? Maintaining lexical diversity in your ad copy is essential. Even if stuffing your text ads with the keywords you’re bidding on might please an algorithm (for now), it’s wholly unappealing to human eyes.
This is a problem you should try like the dickens to avoid since, you know, humans are the ones who will click your ads and purchase your wares.
“Let’s eat grandma!” Eh.
“Let’s eat, grandma!” Sounds lovely!
Punctuation is important, right? It doesn’t matter if you’re attempting to distinguish between supping with or on a geriatric relative or trying to sell products to prospects, it’s pivotal to getting your message across. In fact, I can’t think of a way outside of punctuation that a single character can so profoundly impact the meaning and tone of a text ad.
As such, the way punctuation is used in top performing ad copy can offer you some particularly salient advice on which types of punctuation to use in your copy and, more importantly, how often.
Shock of the century: exclamation points win out in a landslide. I’ve got a feeling that if we were to look at the location of exclamation points in text ads (branded or non-branded) the majority of them would come at the very end of the description line. The general idea here is pretty straightforward. Basically, “You just read my ad! I hope you’re jacked up! BUY BUY BUY!”
The relatively low usage of dollar signs in branded ad copy makes contextual sense because it seems likelier that branded ads would attempt to hammer home, you know, brand over, say, pricing. Ditto for the percentile.
The place where punctuation use in branded ad copy really interests me, though, is question marks. Just 2% of the 335 branded ads surveyed asked a question (or did so knowingly at least).
Test different punctuation marks in your ad copy – even though most top advertisers aren’t using questions to get high CTR’s, you might find that doing so makes your ad stand out. Pattern breaking can lead to high performance.
Punctuation frequency for non-branded ad copy parallels pretty closely what we noted in the branded ads: lots of exclamation marks, oodles of commas.
Interestingly, though, the rate at which dollar signs are used is about 3% lower in non-branded ad copy than branded. While the relatively low percentage of ads using the almighty $ makes sense on the brand side (the objective is likely awareness vs. a direct path to conversion), the fact that it’s not only low but lower for non-branded text ads feels counterintuitive. I’d have assumed that there would have been dollar signs galore in the non-brand ads, and maybe with a larger sample of all ads (not just the cream of the crop) we’d see that play out.
How will knowing about punctuation frequency help you write better text ads?
Based on these figures it’s pretty easy to see that many successful advertisers aren’t asking questions in their ad copy, nor are they leveraging the almighty dollar. If you can find a way to do so in a way that speaks to your prospects, your ads are sure to stand out! However, when in doubt, the path is clear: Use an exclamation point!(!!!)
In a word: no.
If you’re unfamiliar with DKI, it’s a tool that allows you to customize your ad to match a user’s search query. This affords you the ability to create more specific, highly targeted ads. All you need to do is create a generic ad containing the formula for dynamic keyword insertion and voila, you’ve got yourself an ad with perfect intent-match.
While it does have useful applications (especially for online retailers with diverse inventory), our data says that top performing text ads– branded and non-branded alike– tend to avoid DKI like the plague.
As you can see, just 10% of the non-branded ads and 6% of the branded ads we reviewed employed DKI.
While this is a bit surprising (since so many accounts use DKI to some degree or another), the data affirms my belief that DKI is an ad copy stopgap, not a solution. Relying on dynamic keyword insertion probably stifles your creativity, wich is why ads using DKI aren’t among the cream of the crop.
Simply observing whether a top performing ad was branded or not wouldn’t have revealed many profound insights; we already know that advertisers bidding on their branded terms leverage those same terms in their ad copy, and this makes a ton of sense. Branded search campaigns are no longer an option: they’re a necessity.
After all, there’s nothing worse than having a stranglehold on the SERP for your branded organic keywords only to be shoved below the fold by some neerdowell competitor with junk Quality Scores bidding on those very same keywords.
To thwart your competitors it’s important to convey authority, and one of the simplest ways to do this is through the use of the Trademark sign (™ ) or language that otherwise conveys officiality. In the past, we even found that using symbols in your ad could increase your conversion rate. Interestingly enough, though, only 28% of the Branded ads and 6% of the non-branded we analyzed did so.
I’d wager that this speaks to the fact that many SMBs lack the brand equity of a multinational corporation like Nike or Coke or what have you. Case in point…
Search for any yuge brand and you’ll see some attempt (whether through ™ or words like “Official”) by the advertiser to let you know that they are, in fact, the real Slim Shady. If the big guys are willing to use every tool on the ‘ol utility belt, you should be, too.
TM and other symbols aren’t a requirement for high performance, but test out using them to see if they have an effect on your CTR and conversion rates.
About a third of the time! Which seems low, right? Think about how often the ads dotting the SERP you see on a daily basis have numbers in them…
Most of those folks are doing it wrong.
According to our data, just 40% of top-performing branded ads and 37% of non-branded ads include numbers.
Before digging into this data, if you’d asked me how many text ads feature numbers, I’d have wagered on something close to 50%: particularly for non-branded ads. But after seeing the relatively low usage rates for the dollar and percent signs in our pool of top-performers, it makes sense that numbers play less of a role in great text ads than I’d initially thought.
This is not to say that numbers are useless; rather, it’s a solid reminder that your prospects care about more than the price of your goods or services. The way in which you frame (and phrase) your offer matters as much as the offer itself. It’s completely possible to convey value in a text ad without numbers.
Top ads are using numbers in their ad copy (not just to showcase better prices and discounts), but more than 60% of the superlative text ads we studied only used words. Arithmophobic advertisers rejoice (sorry Mark!)
How complex are the best-performing Google ads? We ran our ad set through the Readability Calculator using the Fleish-Kincaid scale (a method of determining how easy or difficult a text passage is to read).
What we found is that the average reading level of the copy in our set of best-performing ads is geared at a 14-year-old. So, if you want great CTR, write for a 9th grader!
Interestingly, a lot of best-selling authors are doing the same thing. Shane Snow at Contently found that most best-selling authors are writing at or below a 9th grade reading level:
Note that Hemingway is even more readable than Goodnight, Moon…
Mashing all this data together, we can gather that the ultimate best-performing Google Ads ad of all time would look something like this….
This ad uses:
Some of this data surprised the heck out of me. What about you?
Please read our Comment Policy before commenting.