How To Launch A Kick Ass SEM Campaign (3 of 3) – Adwords Testing And Optimisation

Ming Yeung

This is part 3 of a 3 part series on clickTRUE's "How to" mini course where we share knowledge on Search Engine Marketing.

In this series of articles, I will do a basic walkthrough of the entire SEM campaign on the Google AdWords platform. Part 1 is about Research, part 2 will go through the process of Setting up an AdWords ad, and part 3 will discuss Testing and Optimization.

This article will focus on split testing of ads and landing page so that optimisation can be achieved.

Warning: This is a long read! (meaning it's chock full of content too!)

The goal of optimisation is to increase the conversion of traffic into customers/leads and maximise the ROI of your advertising budget.

This step is what separates the big boys from the beginners, as businesses which test will gain knowledge of which combination of online elements will produce the optimum results, whereas beginners live or die with their untested assumptions.

So do you want to make a few sales/get a few leads per month or do you want to turn your website into a sales/leads generating machine?

Even Google itself does a ton of split testing on its AdWords platform – colour, font size, font type etc – and this turned it into the world’s largest online advertising space.

If you want to grow your business, then do this step well.

1. Split Testing

1.1 What is Split Testing?

Also known as A/B testing, split testing allows you to run two or more different versions of the ad/landing page which you want to test.

They will be rotated at a rate you can set. For instance, you could set ad A to appear 50% of the time and ad B to appear the rest of the time.

This is the best part about online marketing, as it gives you the ability to track and test the effectiveness of your ad and other marketing materials in real time.

As compared to traditional advertising, online advertising offers much more flexibility and allows you the ability to test and improve your ads as many times as you can, whenever you want.

With this ability, you can easily increase your clickthrough rates from 0.8% to 2.7%, like how Howie Jacobsen did.

You can find a full collection of his test ads here – he tested a total of 15 ads before hitting gold with his last ad!

You can hop over to as well for more case studies of split testing.

There are some very interesting examples there where a simple change of a few words or a colour change has led to a significant increase in clickthrough rate.

Again, it proves that assumptions we make are normally wrong, and testing is essential if we want to make every advertising dollar count.

1.2 Profits per Impression

But before we get ahead of ourselves, let’s not forget that what we want ultimately is to grow our business.

And to grow our business, we need more customers, not just more traffic.

As such, we must not stop at improving clickthrough rates of ads alone.

Rather, testing and optimising of ads should subsequently lead to an increase in profits. This is otherwise known as profits per impression.

So even if a newer ad has a higher clickthrough rate, scrap it if it leads to a decrease in profits.

At the end of the day, realise that split testing is not the mere testing of different variations of a page.

What you need to do, is to measure results, and the differences between the results generated by each variation.

1.3 Statistical Significance

Split tests are only useful if they are statistically significant.

In statistics, a result is statistically significant if it is unlikely to have happened by chance. Click this link to learn more about the mathematics of statistical significance.

Put another way, a test that is statistical significant provides enough data for you to confidently say that if the test continued indefinitely, the results will be similar.

For a split test to be statistically significant, you want it to be at least 90%, and ideally 95% or more to be sure.

For your convenience, you could use this A/B split test significance calculator for reference. Do not skip this step as it is not always obvious that landing page A is better than landing page B.

For instance, you would think a page which received 961 visits with 7 conversions works better than a page which received 974 visits with 5 conversions right?


There is a 54.73% chance that the difference was based on chance, and hence this test is not statistically significant.

2. Split Testing of Ads on Google AdWords

On Google AdWords, click on the settings tab and click advanced settings.

The options are as follows:

Optimize for clicks: Ads expected to provide more clicks are delivered more often

Optimize for conversions: Ads expected to provide more conversions are delivered more often. If there isn’t sufficient conversion data, the system will optimize for more clicks

Rotate: Each of your ads will enter the ad auction an approximately equal number of times. This allows ads with lower clickthrough rates and conversion rates to show more often, so this option could result in a lower average position or fewer clicks and conversions.

2.1 What Can Be Tested?

  • Headline: try different benefits or pain points
  • Body: how you communicate, what words do you use?
  • Call to action: test how explicit your call to action should be

But amidst all the testing, remember that your ultimate goal is to attract buyers, not just to attract clicks.

One suggestion could be to include the price, so as to deter people who are not likely to buy from clicking and so save on your advertising costs.

3. Split Testing of Landing Pages

Looking deeper, an ad is only as effective as its landing page – if its landing page does not convert traffic into customers, a high performing ad is useless and a waste of advertising dollars.

As such businesses should aggressively perform split tests of their landing pages.

You could use the Google Website Optimizer, a free tool which offers all the basic functions.

You will have to fill in the different URLs of the page you want to test, the URL of your conversion page, and then install and validate Javascript tags.

Click this link for a guide on the Google Website Optimizer.

If you have the money to spare, you won't go wrong with paid tools such as Optimizely or Visual Website Optimizer.

3.1 What Can Be Tested?

  • Headline: Does it pique and sustain your user’s attention?
  • Page format: Is the design professional (colours, font, etc)? Is it long or short? What elements (video, etc) are used to communicate the message?
  • Content: Test different pain points and benefits; words; writing style
  • Others: Positioning of certain items such as your menu bar or opt in form; existence of certain items such as a FaceBook widget for social proof or trust seals, etc
  • Price
  • Images

4. Some Other Tips

1. Showing a best performing ad all the time will not produce optimum profits

Counterintuitive but true. Why?

Because most businesses offer a variety of products/services which represent different types of benefits to customers.

If an ad is shown all the time, you will only be able to target a very specific market segment, and will not be able to reach out to the other segments you are targeting.

2. You can conduct split tests past the ad and landing page level

What I mean by this is that you can conduct split tests from a settings perspective. Some examples include:

  • Time of the day: would users be likelier to buy your products at night as compared to in the day?
  • Geographical location: is it more profitable to sell in Singapore than in Malaysia?
  • Ad position: try submitting lower bids and end up lower on the paid listings, you may be pleasantly surprised
  • Latent topics: Could exposing your ad to search queries that are indirectly related raise your sales? E.g. bidding for the keyword “online flower delivery” when you sell “chocolate gifts”

3. Test one thing at a time

Obvious but not so obvious, as many beginners tend to test too many variations at a time and screw up their findings.

Actually, it is possible and it is called multivariate testing. However, it requires a much more complex set up and requires huge amounts of traffic to work well.

4. Track your experiments

In the rush to test, many forget to track and as a result lose the valuable lessons from previous experiments.

5. How clickTRUE Perform SEM Optimisation

Offering a sneak peek into our enhanced SEM system, we set "rules" (or algorithms) to manage all our PPC accounts on a consolidated platform. E.g. this system allows us to consolidate the data from all the search engines such as Google and Yahoo.

We operate by basing our strategies on the predefined rules as follows:

The system will automatically adjust settings such as bidding to ensure that our target goals/"rules" are met.

The image above shows some of the advanced settings we can set for our rules, such as impressions and clickthrough rate, to ensure that our goals are met.

With these settings, we get much greater flexibility in managing and monitoring our accounts in making the appropriate adjustments.

This allows us to optimise the various ad groups effectively to achieve our goals, such as lower cost per click or higher clickthrough rate.

To sum up the entire process of search engine marketing, you could check out this infographic – a clear step by step guide to how to go about the entire process. 

If you want to grow your business and crush your competiton today, why not contact our helpful consultants? We can assist you in achieving a higher ROI for your advertising budget that is double or even triple that of your competitiors.

No comments

You must be logged in to post a comment.
Receive FREE updates twice a month on how to improve your online marketing efforts

 Especially for SMEs and Marketers

 15,000+ people have already subscribed. Join them today.

Unsubscribe anytime. We will never sell, rent or share your email.

⇧ Subscribe To Our Newsletter