8 Critical Split Testing Rules for Google Ads

Split Testing Rules

If your website gets plenty of traffic but you’re not split-testing you have a great opportunity to increase your sales and enquiries.

However it’s easy to make mistakes or do it ineffectively—so here are some rules that make split-testing more effective.

Warning: This is a long article, but I want to give you some critical tools. If I only told you half the story and you split-test incorrectly it could waste precious marketing budget. So here’s all the critical advice in one go.

Also please be aware that most of these rules apply not only to split testing Google Ads, but to direct mail, web pages, print media adverts… in fact ANYWHERE you have the ability to provide more than one version of sales copy and have it displayed in rotation.

The Perfect Split

The ideal split-test is where you get a true AB sequence. For example a magazine would print version A of your advert, then version B, then version A… and so on… as the individual magazines roll over the press.

When those magazines end up in the newsagents, each person who picks up the magazine will get A, B, A, B and so on. That’s a ‘perfect split’.

However I’ve found these days not many printers will go that far, but some newspapers do still have separate print runs for different parts of the country. And some magazines allow you to place inserts for different counties or regions of the country. And so on.

(In this examples always remember to take the regional levels of readership into account to calculate response rate accurately.

For example a North/South split — if they have 5,000 readers in the North and 10,000 readers in the South… if you get 75 respondents from the North and 101 from the South, then the Northern version has greatly out-pulled the Southern version.

(A) 75 / 5000 X 100 = 1.5% conversion

(B) 101 / 10000 X 100 = 1.01%

So the (A) version is 48.5% more effective than the (B) version. (The formula to calculate the percentage lift in response is “WINNER% minus LOSER% divided by LOSER% multiplied by 100”.)

That means through simple split-testing you found an advert that for every 200 people who see the A version of your advert, you get an extra response. (E.g. a subscription, sales-lead or actual sale depending on the call to action in your advert.)

Hopefully the above introductory text has given you the mechanics and ideas for where to split-test.

Before you do, check out the follow 8 golden rules of split-testing. (Some are Google Ads specific – I’ll tell you which those are when we get to them.)

1) Test Semantic Separation

To quickly find the greatest leaps in sales-optimised advert copy you want to test adverts that have not only different words, but also entirely different MEANING… or “semantic separation” as it’s called in the testing world.

For example, I could write and split-test two adverts for one of my home-study programmes about blogging…

(a) Discover the Greatest form of Customer Communication Ever Invented.

(b) Revealed! The Most Powerfully Effective Way to Maximise Lifetime Value.

Although worded differently, those headlines both mean the same thing. (Using your corporate blog to communicate with existing customers.)

Split-testing those would probably only reveal minor differences in conversion because the meaning is the same.

Instead of the (b) advert copy, I would be better testing a second advert with a totally different meaning. Perhaps something like this:

(b) How to Attract Streams of New Customers in Less Than 24 Hours… Guaranteed!

That has a totally different meaning–using a blog to attract new customers. A split-test would then reveal which of the two facets of blogging is of greater import for my target audience — find new customers or sell more to existing customers.

(And of course it would be that headline that sells my product more effectively and which I would roll out on a wider scale.)

2) Always Get Outside Feedback

You probably think about your products and services differently from end-users. It’s the main drawback to writing your own sales copy — you may use more technical words that don’t make sense to your general audience.

So, if you have access to an expert copywriter you may want to explain your split test ideas and get their advice. I’ve given links to recommended copywriters in the US and UK on this blog.

Also, ask your customers and clients. Interview them or at the very least have an informal chat about the main benefits they’ve received.

Make a note of those and also especially the way in which they describe those benefits. Then find a way to build those into your advert.

3) Investigate Your Competitors.

Check out competitors adverts to see what copy they’re using. Over a period of time you can start to see which adverts are winning — they’re the adverts competitors keep running over a longer period of time.

Google Ads make this tactic particularly easy, but be careful. Some advertisers don’t check their adverts and just let them run and run even if they’re NOT profitable. Obviously you don’t want to get too much inspiration from those adverts.

The easy way to identify and ignore those adverts is to check out the pages you’re taken to when you click the advert. If the page you land on isn’t targeted to the wording in the advert, or not very persuasive and doesn’t make much effort to sell the product or capture your details as a sales-lead, then you can be confident they fall in the ‘More Money Than Sense’ (MMTS) category of Google advertisers.

I was helping set up a private clients Ads the other day. We typed in his main keyword and did a quick check on his competitors. Sure enough, his main competitors were running Google Ads, and they appeared to be very well written and targeted to the keywords we typed in.

Unfortunately when we clicked them, most took us to pages that bore no resemblance to the keywords we had searched on or even the text in their advert. What a waste of money.

In fact the worst one was the advert we clicked which took us to a “Page Not Found” message on the advertisers web site. Definitely a case of MMTS!

4) Eliminate Noise.

If sales and response to your Google Ad advertising suddenly drops off… maybe it’s simply due to the volume of searches being conducted?

Be aware of seasonal fluctuations.

Check your statistics BEFORE tweaking your ad copy.

A rule of thumb is that you should run split-tests over at least a 7 day period (regardless of how many clicks you get).

Because if in your industry customers are more responsive in the latter half of the week –o r maybe over the weekend — then a 7 day test period will include that fluctuation. Whereas if you only tested at say the beginning of the week, results would be distorted.

However, my personal take on this — it depends entirely on the industry! Some of my tests are confined to a single day, and that has proven to be enough to find a winner.

Here are some tips specifically for Google Ads…

5) Always divide keywords into separate AdGroups.

That will allow you to write and split-test highly specific adverts for each adgroup. In most cases you want to use the keyword itself in the ad… you can only do that when keywords are divided into separate groups.

Here’s a slightly daft example, but hopefully one that clearly explains what I mean:

If you have a fruit shop, don’t just have a single Fruit adgroup containing “Apples” and “Oranges” keywords. Have separate AdGroups for each kind of fruit and use the name of the fruit in the advert copy itself.

Realise that the more your advert copy matches the keywords they type into Google to run a search, the more likely they’ll think “that’s what I’m looking for” and click your advert.

6) Split-test the display URL as well as the advert copy itself.

Yes, even the domain name shown in your advert can have a significant impact on response. Definitely try adding the keyword to the end of your domain name after a forward slash. www.TheFruitShop.com/Apples for example.

7) Capitalise Each Word In Your Advert.

Capitalising words in your advert makes it look more interesting (eye-catching) over adverts that are all mainly lower case. And punctuations like the ampersand ‘&’ can also increase response.

8) Don’t just look at the Click Through Rate.

The number of people who click your advert is very important. It affects how much money you pay Google. But it’s not the most important statistic. You want to keep an even closer eye on (and continually optimise) the back-end conversion rate.

In other words, how many people see your sales letter versus how many people respond to it and subscribe, sign-up or buy the offer.

Of all the different things you can do with Google Ads, split-testing is one of the most critical optimisations you can make.

If you use Ads, and know you only have a single advert, stop reading my blog NOW and go add some more!

Happy split-testing!

Subscribe to my blog and join 11,000+ subscribers who already receive free access to my tools & resources…

Leave a comment


Scale Your Brain

5-day Free mini-course


Ed Rivis