Split-Testing: What is it? Where to Start?

In the marketing and advertising world, the terms Split-Testing, A/B Testing, or just the word Testing describe the process of finding out what type of messaging generates the best response for your product or service from your specific buyers.

This is how marketing and advertising campaigns are made profitable and then made even more profitable.

In brief, Split Testing or A/B Testing is a process of showing two (or more) versions of the same ad or Landing Page to different users. It’s like a performance race in which the winner is determined by how many people respond. Whichever version generates the most responses (visitors, subscribers, leads, sales, etc.) is the winner. But instead of receiving a trophy, it’s designated as the “Control” ad and a new race (test) is initiated to achieve an even better response.

This type of testing is not new. In fact, a prominent book on the topic was first published a century ago in 1923: Scientific Advertising, by Claude C. Hopkins. It outlines the basic principles of testing that are still used today. Although nowadays testing done via the internet is much faster than back then, when it was conducted through direct mail and print ads.


The “Split” refers to the fact that different versions of a message are run at the same time and “A/B Testing” means one ad is called “A” and the other is called “B.” In either case, the ads are being shown simultaneously online to the same target audience, although any given viewer only sees one ad and not the other.

For example, you might have a video ad that emphasizes the “quality” of your product or service and another video ad that highlights how it “saves you money.” Whichever generates the greater response becomes the “control” ad. Then you test another “challenger” against the control ad. For example, maybe you would emphasize both “quality” and “saving money.” That would seem like it should perform better, but only Split-Testing will provide an objective answer and sometimes what may seem obvious is not true at all. It’s quite possible that emphasizing only one benefit (the highest performing benefit as determined by testing) might out-perform an identical ad that highlights more.

As another type of example, you can run a Split-Test of the exact same service with the only variable being that one ad has a lower price. It might seem intuitive that the lower price should perform better. But sometimes the higher price generates more sales if it conveys more value. (This would be less likely on a commodity type of product where its average market price is well established.)

In brief, Split-Testing is an objective way to answer questions about how to increase leads and sales. A series of such tests is the foundation of a marketing strategy to generate the most predictable sales for the lowest cost.


Whether through text ads, picture ads, video ads, or TV commercials, there are three fundamental areas that are ripe for testing. Which of these should be tested first can vary with any given product or service. More on that further below.

1) The Landing Page
2) Geographic and demographic Targeting (also mobile vs. desktop devices)
3) The ads themselves

All three are vital and are interrelated.

Effective online advertising is comprised of all three optimized to generate the most productive response for the lowest cost.

By the way, testing can be used for more than ads. For example, testing email subject lines is common. Even testing potential business concepts, such as a new product or service. Testing the product or service idea before it’s developed is an ideal way to gain some initial insights into how well the concept will be received. In some cases, you may find it will be too expensive to launch the new product or service and therefore testing can save you a fortune in lost money and time to learn that now rather than some years later. However, for this article, we are focusing on ad testing for high-performance online sales expansion.


The Landing Page is where prospects arrive once they click on your ad. It’s usually not the home page. In fact, this page may not even be visible to regular website visitors. This means the only people who would arrive on this page are those who clicked an ad.

A Landing Page can also be a product page and for retail businesses with hundreds of products, there may be hundreds of ads and Landing Pages.

An ideal Landing Page should be tailor-made to work hand-in-glove with the ads and vice versa.

From a practical perspective, it’s often easiest to begin a new ad campaign by developing a Landing Page first. It should use industry best practices to present your product or service, while also contemplating the target audience so that the language caters to that type of prospect. Later on, when developing the ads themselves, you should return to the Landing Page and revise it based upon the wording of the ads (more below). Especially after the testing starts showing which words and messaging are generating the best response.

There is much that can and should be tested for your landing page.

Just bear in mind that even if you have an ideal Landing Page, but the advertising is poorly targeted and/or the ads do not engage viewer participation, then you will have unfavorable results. Even the most optimized Landing Page needs a steady volume of the right prospects to be productive.


For many advertisers, the first thing they think of when contemplating a new campaign is who is the target audience?

The audience informs the ad messaging and the Landing Page. If the demographic is young families, then the messaging would be different than for technology engineers.

  • Is the target audience mature?
  • Would they be in a very narrow age range?
  • Would they be high-income earners?
  • Is the target audience interested in sports in general or just one specific sport?
  • Do they live locally, or in urban areas across the nation?
  • Or, is the target audience suburban moms?

For search campaigns, the ads are only displayed when someone searches online with a particular combination of keywords. For example, “red shoes,” or “tennis racket prices,” or “what is the best travel camera” or when you’re visiting a new city, “where’s the nearest pizza?”

Pay-per-click search campaigns tend to be the most expensive ads on a per-click basis because they are targeted to someone seeking something right now. When you put your ad about tennis rackets in front of someone searching for “tennis racket prices” you have a higher probability of generating that person’s interest and making a sale right now.

By the way, although it’s very helpful to know your targeting as precisely as possible, it’s not imperative to be overly exact in the beginning. In fact, even if you do know you want to advertise to a very narrow target audience, sometimes it can still be beneficial to start with a target audience that’s broader. This way the new metrics can re-confirm what you already know or provide additional insights.

  • Maybe your best buyers are becoming a little older?
  • Or younger?
  • Maybe some related industries to your primary target will generate enough responses to target them, as well?

Also, keep in mind that you could have your message targeted to the most ideal prospects for your product or service, but if the ads don’t invite viewer interest and if the Landing Page does not result in actual conversions (buyers or leads), then even ideal targeting will be for naught.


When used effectively in conjunction with Split-Testing, the Ads provide a substantial volume of performance data.

Of course, the ads are the most prominent part of any advertising campaign. Yet, as important as they are, they should be conceived holistically — meaning in combination with the target audience and Landing Page.

You could have a well-optimized advertisement that generates a lot of clicks, but if it’s not targeted to the right audience or if the Landing Page is ineffective, or worse, seems disassociated from the ad, then the ad and resulting visitors will not produce good results.

A fundamental part of ad testing is obtaining objective data on what moves your prospects to action. This very same data should not only be emphasized in new ads but on the Landing Pages, as well.

For example, let’s say your product or service has a dozen benefits. That’s too many to make any kind of lasting impact on a viewer. Sometimes a very simple (and effective) ad may only feature one benefit. Other times, three benefits can be featured. But featuring more may result in no message being conveyed to a viewer who is considering the ad with an indifferent eye. In other words, the whole message is tuned out.

  • So, which three benefits should be highlighted?
  • Or is one benefit clearly the most compelling?
  • Find out by testing.

And then be sure to translate the same information to the Landing Page. Although the Landing Page has the space to outline all of the benefits, its performance will be increased by mirroring the data from the ads.

In other words, the highest-performing benefits that are featured in the ads should stand out on the Landing Page. This might mean:

  • Using a bold font
  • Featuring them as subheadings
  • Surrounding them with white space so they are more prominent
  • Or perhaps each of the featured benefits has its own graphic

Alternatively, a Landing Page test can be conducted comparing one that offers all the benefits versus another that only features the few benefits that have performed the best in earlier tests.

In my experience, some of the biggest improvements in an ad campaign’s performance are made through Landing Page optimization. Yet, for many who do at least a minimal amount of testing, they tend to focus solely on the ads and targeting. All three areas are critical for high-performance advertising.


In our work at Skyworks Marketing, we primarily test:

  • Paid ads
  • Targeting and
  • Landing pages

However, testing has broader applications.

Regardless of whether you invest in paid ads, search engine optimization (SEO), content marketing, or any other lead generation or sales channels, the more resources you intend to invest to grow your business in the future, the more important testing will be now.

Even if you were planning to primarily use SEO to attract new prospects to your website, you can use the concept of Split-Testing to inform which keywords will be the most effective at attracting the right buyers. In this SEO circumstance, it’s still better to use pay-per-click advertising to run a series of tests, since it’s much faster to obtain meaningful comparative data with paid advertising than SEO. Then, with the right performance data, you can focus on building your SEO with greater confidence that your work will be remunerative for the longer term.

Having said that, if you use testing to develop a profitable paid ad campaign, you may want to continue with paid advertising since, for many businesses, it offers a more predictable flow of the right buyers than SEO.


What should be Split-Tested first to optimize an online ad campaign?

Of course, we do need to start somewhere, but it’s best to think of testing in stages and holistically. This means that wherever we begin, the knowledge gained from one area (landing page, targeting, ads) will inform the other two areas and vice versa.

In truth, there are numerous stages of business and marketing development. But the following outline establishes a basic way to categorize two opposite ends of a spectrum of testing to help inform where you should start for your circumstances. Many businesses will fall somewhere between the two.


If the product or service is new and is going to be sold in a competitive market, then testing will be crucial to gaining an edge.

Maybe your new product or service has a benefit that makes it better than the competition.

  • But how much is that worth to your buyers?
  • Perhaps you are pricing the new product or service 20% higher than the competition.
  • But what if 25% or 30% would sell more?
  • Or, what if your buyers will only pay 10% more?
  • Or, what if they won’t pay anything more than what already exists in the market, but at least they will try your new product if it’s the same price?

Testing will point the way.

Here are some fundamental tests for a new product or service for landing pages and ads:

  • Price testing
  • Landing Page headline testing
  • Primary feature and benefit testing
  • Broad target-audience testing
  • General music testing (video only)
  • General voice-over tests (video only)
  • General offer testing (offer = shipping, guarantee, payment terms, etc).
  • Basic image testing (which graphics perform better).
  • People or no people testing


If you’re selling an established product or service, you already have some insights on what works and what doesn’t, in terms of marketing and advertising. Nevertheless, those same insights could be self-limiting if you’ve been selling something for many years and the buyers and competitive offerings have shifted over time. Are you still operating from principles you formulated years ago, but may not be as true today? Testing will provide answers.

Note: Even if you have an established product or service it can be beneficial to test some or all the points in Stage 1 to ensure your messaging is built on a foundation of high certainty. Having said that, the following are further testing factors to increase campaign performance:

  • More detailed Landing Page features and benefits testing
  • Image/graphics testing including size and page placement
  • More segmented targeting testing
  • Different types of people
  • Testing more detailed features and benefits
  • More nuanced music variations (video)
  • More nuanced voice-over (video)
  • Different channel testing (Google, LinkedIn, Facebook, Amazon, TV, etc.)
  • Call-to-Action button placement and colors

These two ends of a testing spectrum are guidelines only. In practice, the stages your own testing will evolve through are based on your unique products or services and their relationship to their overall market.

The basic principle is to start by testing broad concepts and then in the subsequent stages test narrower comparisons.


It may occur that you become quite inspired when you start gaining real-world, objective insights into what generates more leads and sales from your marketing and advertising. That’s great!

  • Testing is a direct communication channel from your customers to you.
  • Testing is a profound listening opportunity to hear what your customers need and want by way of their actions and wallets.

However, it bears emphasis that you don’t want to focus all your testing on just the Ads, or just the Targeting, or just Landing Pages. All three areas need to be tested and the knowledge gained from any one area should be used to benefit the other areas, whenever possible.

The most profitable gains will result from a synergy of testing and optimization in all three areas — Landing Pages, Targeting and Ads.


Sometimes we offer special promotions for Split-Testing and/or video advertising trials. Visit our Skyworks Marketing home page to see what we’re offering this month.