Test with "Experiments" Ads by Google

 A crucial element of pay-per-click marketing is testing. No matter the tactic or feature, thorough testing is essential to figuring out what works and, more importantly, what doesn't.

Automation and machine learning, though, are a problem with PPC testing today. Take the example of a Google Ads campaign that is configured to maximize conversion bids. Google will always favor one ad over the other, even if two advertisements are competing against one another and the only difference is the call-to-action. Although the advertiser intended for each ad to receive 50% of the impressions, the bid strategy will show the ad that Google expects obtaining the most conversions even if one ad obtains, say, only 20% of the impressions.

Fortunately, Google has created Experiments in three different categories:

• Custom
• Ad Variety
• Video

Individual Experiments

Custom experiments, which are accessible for Search and Display campaigns, are the most popular (not Shopping). The experiment functions by using an existing campaign as a model and making changes to it. An advertiser could, for instance, compare a base campaign with an upgraded cost-per-click bid strategy to one with a bid strategy that maximizes conversions.

Make your choice of the traffic split while setting up the trial. The best comparison between the base and trial is ensured by 50%. then decide whether to do a search- or cookie-based trial. I suggest cookie-based since it limits the change to just that variable while randomly assigning people to the original or the experiment. The user can view the basic and trial results for any given search while using a search-based system. The results could get diluted if users run several searches and see both options.

Up to two goals are allowed for custom experiments. I'm hoping for higher conversion rates and lower cost per conversion in the example below.

Up to two goals are allowed for custom experiments. This example demonstrates how conversion goals can rise while conversion costs can fall.

The dashboard displays clicks ("Clicks"), impressions ("Impr."), and other metrics with the two test metrics of conversions ("Conversions") and cost per conversion ("Cost/conv. ").

The dashboard for the experiment's findings displays the two test metrics—"Conversions" and "Cost/conv"—as well as additional metrics like "Clicks" and "Impr."

Advertisers have the option to apply or discontinue the experiment at any moment, and to conduct it with or without a set end date.

Additional custom experiments comprise:

• Conversion procedures (for example, optimize for purchases versus email signups).
• showcasing creatives
• types of keyword matches

Ad Variations

Prior to Google's Smart Bidding and Responsive Search Ads, ad testing was simpler. Google would comply with requests from advertisers to "Do not optimize: Rotate advertisements endlessly" in the ad rotation settings. Although it is still an option, Smart Bidding mostly disregards it. Up to 15 headlines and four description lines can be tested at once with responsive search advertisements, which frequently eliminates the need for additional ads.

Without utilizing Ad Variations, it is more difficult to test landing pages or advertisements with a particular message. For 50-50 tests across ads in one or several campaigns that fit your criteria, use ad variations.

Consider the scenario where "Purchase Today" is a call to action in an RSA and you want to test "Your Order Ships Today." Go to "Create variation" and choose "Find and replace" from the first drop-down menu to test across all headlines.

Then:

• "Purchase Today" should be typed into the "Find text" field.
• For the relevant drop-down, select "headlines,"
• Your Order Ships Today" should be typed in the "Replace with" area after clicking "Match case."

Calls-to-action can be tested across all headlines using Ad Variations trials.

You'll observe a performance difference between the base and the trial, just like with Custom experiments. The overall effectiveness and the outcomes for each individual ad are both available.

Unfortunately, unless the basic ad already includes a pin, Ad Variations does not permit testing pinned headlines or description lines. A message will always appear where the advertiser specifies thanks to pinned assets. The alternative message of "Your Order Ships Today" would also always appear if "Purchase Today" was set to pin in headline 2. Unpinned assets will cause the text to shift, but we won't be able to see the title or description line. Ad Variations are still valuable in my experience.

Test additional ad variations like:

• Landing pages,
• A copyright symbol on the brand name,
• URL paths.

Experimental Videos

The rise of video ads necessitates testing. The goal of video experiments is to ascertain what is successful on YouTube. Two base campaigns with identical settings but different films are needed for video experimentation. A store promoting the newest running shoe, for instance, would try using two films. While the second video shows someone wearing the shoe, the first one can feature close-ups of the shoe.

Up to four campaigns (videos) can be run concurrently in video experiments, each with its own brand lift and conversion measure. Based on survey results, brand lift calculates the effectiveness (i.e., brand recall) of the advertisements. It's especially pertinent to video campaigns because they rarely result in conversions. Like the other tests, video experiments deliver outcomes instantly.