How to gather insights to write the perfect ad copy

Enjoy Digital
By Enjoy Digital
5 minutes to read

It can be easy to fall into the same routines when it comes to writing ad copy.

It’s also time-consuming to rewrite and refresh copy so it can often be overlooked. So to make that job easier, I decided to delve into some insights to see if there were any common threads from 15 year’ worth of ad campaign results that we could use to write better ad copy.

Sourcing the ads

I started by going to the ads section of every account in our My Client Centre (MCC). This gave us ads from a wide variety of industries with a range of budgets and different goals. In total it churned out 16,167 ads – plenty to play with!

Because we wanted to look specifically at ad copy, I stripped out unnecessary information so we only had headlines and descriptions accompanied by data on impressions, clicks, and conversions.

By this point my laptop was struggling with the volume of data, so I filtered out ads with fewer than 1,000 clicks and any ad that contained dynamic copy. That left us with a mixture of text ads, expanded text ads, and responsive ads.

Ultimately, I wanted to see the data at the most granular level – by the individual words within the ad. All three ad types required the same process of deduction to do this but responsive ads took longer as they have 15 headlines rather than three.

An issue I couldn’t avoid however was that while a responsive ad does have 15 headlines, Google will only show three. But you can’t see which combination of three headlines generated clicks and conversions. So this analysis assumes that all 15 headlines share the same stats for the ad which obviously isn’t ideal. But even Google can’t differentiate the data.

But because only a fraction of the 16,167 ads are responsive, it’s not really a significant issue.

Handling responsive ads

Responsive ads meant I was faced with 15 headline columns and four description columns for one ad. I needed to separate the headlines into one tab and the descriptions in the other so I could generate results for both areas.

To do this I moved all 15 headlines into one column. This meant the data for that one ad was duplicated 15 times. For example, an ad that generated 10 clicks with 15 headlines was now reading as 10 clicks for headline one, 10 clicks for headline two and so on. This meant that I could analyse results per individual headline rather than all 15 headlines at once.

I followed the same process for descriptions.


Breaking it down word-by-word

I then used text to columns on spaces so that each word of the headlines and descriptions was displayed in its own column next to the relevant data. Again, I moved each word into one column so we had the data per word. After putting this into a pivot table I then removed every industry and brand-specific word I could find. From there I could create a click-through rate and conversion rate column to order the words by what was most likely to get a click or a conversion.


What we learnt about words

The final result is a spreadsheet that provides a detailed breakdown of highly engaging, high converting words to use in future ad copy.

We learnt some useful things about the words themselves.

1. Luxurious language doesn't work for everyday products 

The likes of ‘luxury’ or ‘premium’ don’t entice engagement. In fact, the work we’ve done suggests these terms are 30% less likely to generate a click. This is perhaps an indication that these words have a connotation of expense. Unless it’s actually a specific keyword term, ‘luxury’ and ‘premium’ should be left well alone.

2. Personalisation is still key

When it comes to descriptions, ‘You’ll’ is 500% more likely to generate a conversion! This reinforces that the user journey should be as personalised as possible along every step of the way. It doesn’t have to be complex personalisation either, writing copy in the second person seems to go a long way.

3. Convince the reader

In headlines, ‘need’ is 450% more likely to generate a conversion. This could be because you’re making it clear to the reader the product is for them. While ‘need’ doesn’t add any urgency, it doesn’t leave room for the reader to consider it. ‘Need’ simply helps to reinforce that the product can’t be passed up on.

4. Some words are hard to predict

In the description, ‘Each’ is 870% more likely to generate a click. Now, I’ll readily admit we’re a bit baffled by this. The only vague conclusion we can come to is that ‘each’ is perhaps, very loosely, an indicator of care and attention or a bargain. It acts as an emphasis to the point and highlights the significance.

For example, ‘Each one of our mattresses…’ or ‘Only £10 each’

Either that, or it’s just a really common word that we use without realising it.

5. Add a time frame

In headlines, ‘years’ is 570% more likely to generate a click. While this isn’t likely to be because of urgency, it is more likely to be a form of social proof or guarantee. E.g. ‘Lasts for 10 years…’ or ‘Years and years in the making…’

6. Be sensible with discounts

When it comes to discounts, 10% seems to be the sweet spot with a CVR of 29.7%. Discounts as high as 50% perform worse with a CVR of 1.3%, perhaps because users don’t view 50% off as a discount but rather a permanent sale. Discounts between 10 and 25% feel like genuine offers that have a time frame.


What influence do character counts have?

The next factor I looked at was character length. According to best practice, you should maximise the size of your ad and include as many details, benefits and information as possible to increase your chances of generating traffic and conversions. I wanted to see how our data stood up to best practice.

Before I broke down the headlines and descriptions by word, I used the LEN formula to get the character count for each headline and description. To simplify the findings I grouped the character counts into a small, medium & large categories.


  • Small headlines are 6 to 14 characters
  • Medium headlines are 15 to 22 characters
  • Long headlines are 23 to 30 characters


  • Small descriptions are 11 to 37 characters
  • Medium descriptions are 38 to 64 characters
  • Long descriptions are 65 to 90 characters

The results were really interesting.

1. Keep headlines short

Strangely, shorter headlines (so 6 to 14 characters) are 170% more likely to lead to clicks and 130% more likely to get conversions. This goes against all guidelines but could suggest that as more people aim to write longer, descriptive headlines it’s the short and snappy ones that stand out.

2. Detail leads to conversions

Longer descriptions are 150% more likely to lead to a conversion. This could be because the more information and detailed the description, the less hard the product page has to work.

3. Descriptions are tricky

While long descriptions are more likely to convert, medium-length descriptions are 140% more likely to get a click. So that indicates that for a successful ad campaign, use medium length descriptions but optimise your landing page for conversions.

Analysis is always beneficial

This was a time-consuming process (if you know a quicker way, let me know!) and not an exact science, but it has given us some useful guidance on how to write our copy. Taking these insights and coupling them with copywriting best practices should produce good levels of engagement for our ads.

So why not take a look at the results from your own campaign to see what you can understand about ad copy? Let us know what you find

comments powered by Disqus