Helium Blog 20180209 web

Chiseling Your Way In

Communication, Development, Strategy

Why Split Testing Ads Is So Important


50% of the traffic on Google goes to the top 2% of advertisers.  But what if you’re not one of the ‘800 pound gorillas’ in the top 2%?

How can you realistically compete on any paid advertising platform if you’re an underdog in your market?

The real solution is NOT to just throw money at it.

If you are getting killed by the competitors because your ads suck, then increasing your advertising budget is just going to accelerate the bleeding.

Before we get to the solution, we need to cover a fundamental online advertising metric.


Google, Facebook, and Bing assign a quality or relevance score to each of your ads.  There’s a few different criteria that gets used to calculate their quality score.  The biggest factor by far that determines quality score is click-through-rate (what percentage of people who see your ad are actually clicking on it).

In a sense, people are voting on what the best ads are by clicking on the ones that are the most compelling to them.


The search engines and platforms like Facebook want to show good quality ads because people will have a better experience if the ads are relevant to them.  The more relevant your ads are, the more people will click on them, the higher up on the page & more often they will be shown, and the lower your cost-per-click will be.  So it’s in everyone’s best interest that good quality ads get shown.

When this is executed well, a snowball effect happens and that’s how the top 2% of advertisers dominate their markets and get 50% of the traffic.

What’s the secret to starting this snowball effect and end up with great ads that dominate your market?

Split testing!


By testing two or more ads against each other and continuously trying to beat your control ad, you can ratchet up your click-through-rate, increase your quality score, and drive down your cost-per-click.  This is how to get your ad to show in the top positions without increasing your bid prices.

The key is to brainstorm many ads that are drastically different from each other, NOT 10 minor variations of the exact same ad. Think of completely different angles to come at the problem your potential customer is trying to solve.

Once you have enough data in each split test, you simply look for the worst performing ad and pause it.  Then you start your next round of split testing by adding a new ad that you’ve already brainstormed and is ready to battle against your current winner.  (It can be very exciting to watch the numbers and see which ad comes out on top!)

Eventually, you will come up with a new ad that beats your control ad.  Then it becomes the new control ad.

This is how you systematically keep increasing your click-through-rate.  This is how you chisel your way into the toughest cutthroat markets.


Below are 3 tips that you can implement immediately to split test your ads the right way.  This is how our agency approaches ad split testing.

Feel free to use these methods for your own online advertising.  In fact, we encourage it.  (It’s hard to believe how often we still see supposed “expert” agencies and freelancers who are still doing split testing wrong…or not even split testing at all!)

Embrace your inner scientist.  Start experimenting with new ads today!

1. Brainstorming Ads

  • Look for obvious things that might be good to put in the ad
    • Study existing ads
    • Visit the landing page that the ads point to
      • Include some of the text, images, videos, or other key elements of the landing page in the ad
    • Check the keywords tab if it’s a search ad for Google AdWords or Bing Ads
      • See which keywords & search terms have the most impressions and include those in the ad text (that’s what people type in the most often, so if you repeat their own language back to them, they often respond better)
  • Use an ‘Ideal Prospect’ statement to nail down exactly who you are creating the ads for
  • Ideas for ads
    • Use the Swiss Army Knife tool created by Perry Marshall to brainstorm emotionally driven ads that use many different knife blades in each ad
    • Borrow ideas from ad swipe files
      • Sometimes ads from a completely different industry work well to modify and use in another unrelated industry
    • Study top competitors’ ads
      • Create a Frankenstein ad with the best elements of each competitor’s ad combined into one new ad

2. Determining Losing Ads

  • It’s ideal to have at least 1000 impressions or 20 actions in each ad since that’s statistically significant data
    • Conversions should be used as the 20 actions, although a click is also an action, but it’s a less valuable action
  • Often the winning and losing ads are obvious
  • Primarily use conversions and conversion rate to determine winners and losers
    • Click-through-rate and Google Analytics data (bounce rate, pages/session, avg. session duration) should be used if there are no conversions
  • Super Split Tester spreadsheet works great to determine winners and loser if one ad has a higher click-through-rate, but the other one has a higher conversion rate

3. Structuring Ad Split Tests

  • Have at least 2 active ads in each ad group (in most paid search platforms) or in Facebook Ads, have at least 2 ad sets that have the same targeting and put 1 ad in each ad set
    • You can leave more than 2 active ads running in each ad group if you have 2 strong ads and you’re worried about losing conversion volume if the new 3rd ad ends up performing much worse than either of the 2 active ads that you would normally pause
  • Label the ads and the ad group using this naming scheme:
    e.g.  Ad Test 03 Jan 26/18