Get in touch with our team
Feature image for 14.03.2024


4 min read

How to overcome diminishing returns with ad & keyword testing in Paid Search

This article was updated on: 20.03.2024

It’s a familiar situation for many marketers – we’ve set up our campaigns, they’re running efficiently, we’re happy and our stakeholders are happy but soon enough we’re all thinking what else can we do? How can we improve performance even further? 

The first thought for most is: to increase spend to increase revenue. However, this is where we may experience diminishing returns. The law of diminishing returns states that after a certain point each additional unit of input results in a smaller increase in output. This is often the case in advertising – as spend increases, we expect to see revenue increase at the same rate, however, once we hit a certain spend level we will see revenue growth slow. Therefore, increasing spend may result in paying more per click without those clicks translating into conversions or increased revenue.

How does this relate to Google Ads? 

Google Ads’ job is to hit the targets we as advertisers set. Whether it’s a target ROAS, target CPA or to drive as much traffic as possible through a maximise clicks bid strategy. Users and search terms that are most likely to hit those targets are prioritised which is why we often see high ROAS or low CPA when budgets are low. Once we’ve picked the low-hanging fruit, the audience pool expands to lower-intent users and more expensive audiences which is why, incrementally, we see each pound return less revenue. This is where we need to get more creative with reaching relevant audiences and hitting targets. Unfortunately, simply doubling the budget rarely means doubling the revenue, particularly if no changes are made to the campaign.

How do we find the sweet spot?

So, how do we find the ‘Goldilocks zone’ regarding diminishing returns? The answer is through testing and experiments. Google allows us to set up experiments, analyse performance and implement what works. By running tests and experiments we can identify new customers through optimise messaging through ad copy tests and improve budget efficiency by reaching the most relevant search terms through keyword experiments. 

Ad copy tests

Most brands have a particular tone of voice that is specific to them and their audience, however, is your ad copy alienating or discouraging some people from clicking? Is your promotional discount enticing enough? Does your CTA lack direction? One way we can find out, and grow the pool of potential converters is by testing different styles of ad copy. Below are some ideas of how a monthly subscription service could test copy to uncover which is most likely to drive clicks & conversions.

For example:

Save 25%One Week Free
Discover MoreShop Now
Valentine’s Delivery AvailableOrder By 10th February

Keyword experiments

One of the best ways to grow an account is through broad match (BM) keyword testing. Historically, BM keywords were considered a little risky with the potential to spend lots of budget on largely irrelevant search terms, however, Google has improved its ability to utilise audience signals for more efficient targeting and Broad Match keywords have become more important for campaign growth. 

  1. Choose a campaign that performs well and has sufficient data – the more conversions the better. Steer clear of branded campaigns to start with and go for product type instead,
  2. Carry out an in-depth SQR – identify any themes or specific keywords to add to a negative keyword list,
  3. Identify what you want the campaign to achieve i.e. low CPCs & increased conversions,
  4. Set up your BM experiment through Google Ads experiments, splitting the budget 50:50 between the control campaign and the test,
  5. Regularly review the experiment paying particular attention to budget pacing, search terms and CvR,
  6. Once Google has deemed the experiment data as statistically significant, review the campaign performance.

If the campaign has performed well and you are successfully reaching in-market users through new search terms, you can choose to ‘apply’ the experiment, in which case the BM campaign will be a fixture in the account. You could also choose to go a little more granular and pick out the top-performing BM keywords and add these to an existing campaign. If the experiment didn’t have the desired outcome then that is also valuable learning that can be used to set up another experiment, perhaps testing BM keywords on another campaign type.

Once experiments have been run and results analysed it’s not long before ad fatigue sets in and refreshes and updates are needed. Adopting a test-and-learn strategy is important but being able to continue the cycle is essential. Consistency with a testing roadmap will ensure the growth and efficiency of campaigns, limiting the impacts of diminishing returns and avoiding stagnation of ads.