A/B Testing Tutorial
Learn a/b testing: for PPC advertisers & professionals
what is a/b testing?
A/B testing is a Search Engine Marketing (SEM) industry term referring to the formal process to running two variants against one another to measure performance and select a winner. A/B testing refers to Alpha & Beta, Alpha being the better performer over the Beta variant or version of the test.
what can be a/b tested?
An A/B test can be performed on a website or landing page to improve conversion rates or on an ad including testing ad copy or banner ads for higher engagement via the Click-Through Rate (CTR) and Quality Score (QS), key components in determining how an ad will perform on the PPC advertising platform's ad rank listing via the auction algorithm.
a/b testing a landing page
Of equal importance is the user friendliness to submit contact information via a form submission, to initiate a chat, or to make a phone call to the business. As paid advertising's goal is to entice and capture leads, the landing page should be optimized by a Conversion Rate Optimizer (CRO) to have a bold and prominent phone number listed in the above the fold section of the landing page, a simple form with fields to input contact information such as name, phone number and email, or other tools to contact such as a listed email address or chat box.
Manipulating the look and feel of these conversion submission tools is one way to A/B test a landing page. Showing a larger or smaller phone number, using fewer form fields, or a different Call To Action (CTA) submission button at the bottom of a form submission are examples to A/B test two variants. By A/B testing one variable at a time, collecting data, and analyzing performance for Conversion Rate and other metrics such as bounce rate (which can indicate problems with a user's ability to complete a desired action), it can be statistically determined what variables users behave more favorably towards. After a landing page A/B test is completed, and the results determine a winner, the "loser" landing page is paused, a new variant is created, and the A/B test is run again. The idea is to continually optimize the landing page further to increase user experience, increase conversion rates for the PPC advertiser, and ultimately create the ideal landing page for the target audience.
a/b testing text ad copy
A/B testing text ad copy for search ads is also a continual process. In the Google AdWords interface account managers can apply labels at the ad level to streamline and control the process of gathering data for similar ads across ad groups or even campaign levels, and is ideal for PPC advertisers with very large accounts. The labels can use terms like "challenger" for new, experimental ad copy and "champion" for proven winners in prior A/B adtext tests.
Potential text ad copy variables for testing including changing the headline, flipping the description lines (soon to be an irrelevant test with the rollout of Extended Text Ads or ETAs), testing different description line copy, and testing different display urls. Professional account managers should consider testing one variable at a time to ensure that the results are statistically significant for one piece of the copy, rather than for aggregate changes.
Good tests for A/B testing adtext includes changes the language from business differentiators to language that addresses a customer's concerns and the solution and benefits of how the PPC advertiser's offering will address their issue. Using customer-centric language entices the user to click on the ad, improving Click-Through Rate (CTR) from a feeling that the business understands what they are experiencing and has the expertise to solve the issue. Other tests include changing the Call To Action of the description or headline, mentioning different benefits of the offering, or adding a limited-time offer such as a discounted rate available only for the next 3 days using a special coupon code.
a/B testing display image ads
A/B testing image ads for display campaigns that target display networks is a good practice to continually improve the imagery, branding and any other items used in the banner ad, to entice higher Click-Through Rates (CTR), View-Through Conversions, a conversion type that creates top of the mind brand awareness and leads to conversions later on via organic or direct user behavior, leading to a decreased maximum Cost per Click (CPC) or Cost per Thousand Impressions (CPM) to show ads on partner websites. Using different image sizes, changing the colors to align better with the brand and create a different feeling in the user, and the hero image (the prominent central object of focus) are great ways to test for better performing banner ads.
A/B testing banner ads is more difficult than text ads as multiple image ads of various sizes can be housed within a single ad group. Effort should be made, such as through the use of Google AdWords labels, to track sets of banner ads, or if the work will be commensurate with the return, testing banner ads of the same size against one another. Keep in mind the latter account management strategy will lead to A/B test results with banner ads of different set types winning different sizes and positions on the display network placements.
a/b testing results, analysis & interpretation
A/b adtext test metric: click-through rate
The Quality Score that determines how ads will show in the auction ad rank listings, is in part determined by the component of expected Click-Through Rate (exp. CTR). The expected Click-Through Rate is calculated using historical account performance data and metrics from competitors advertising on similar keywords with similar ad copy language. Click-Through Rate is improved through the use of more compelling ad copy that engages and speaks to the targeted audience. The more granular the ad copy A/B adtext tested, the better, as it is more relevant to the individual user's search, and may contain the query in the ad copy itself (which is bolded leading to higher prominence). Altering or changing the headline, description line or display url, one at a time of course, helps to statistically test adtext variants through the A/B process. Ad variants that receive higher Click-Through Rates win the current A/B test, the loser ad is paused, and another ad is created for the next A/B test.
A/b Display ads test metrics: click-Through Rate, View-Through Conversions
Banner ads also are tested through the A/B process for Click-Through Rate, which signifies a more relevant, better performing ad for users, but also can measure for View-Through Conversions (VTCs) , a conversion sub-type that is a result of top of the mind brand awareness. A/B testing image ads for View-Through Conversions shows which ads tend to stick in the mind of the user perusing the web, seeing the banner ad, choosing not to immediately click or engage with the paid ad, and then come back via a future organic search or direct URL submission in the search engine and then converting. These View-Through Conversions are not a direct result of the paid advertising, but did influence the user by presenting brand awareness that was memorable when the user needed a product or service later on. It is estimated across industries that about 20-30% of all View-Through Conversions are directly accounted as a result of a banner, image or video ad.
a/b landing page test metrics: conversion rate, bounce rate
The intention of a landing page is to present relevant content that is related to a user's search and the associated ad, and to encourage the user to take action, or convert. Users should be able to find easily accessible and user-friendly ways to contact the business such as submitting a contact form or calling the business phone number. Optimizing a landing page for conversion behavior by testing the number and types of form fields in a contact form or altering the prominence of a contact phone number. Other A/B tests available to Conversion Rate Optimizers (CROs) on the landing page include the hero image, the prominent image in the above the fold area of the landing page or website, amount of text copy on the page, and overall brand presence and feel of the landing page such as with color schemes or placement of objects on the page. A/B tests conducted with landing page software help automatically rotate landing pages to randomly test the sample population, leading to conversion counts and conversion rate results. The higher performing landing page should be selected, the "loser" landing page should be shut off, and the next A/B landing page test conducted.
When A/B testing landing pages, bounce rate should be considered as well. Bounce rate on a landing page or website, especially when a conversion funnel exists such as steps to completing a purchase, should be considered for UX/UI user friendliness. If users tend to drop off at a specific point in the process, there may be technical errors present. A/B testing landing pages or websites helps present these issues, or uncovers unique ways to encourage users to finish the process such as with a different Call-to-Action (CTA) or more fewer form fields to complete.
a/b TESTING cONCLUSION
We have discussed what A/B testing is; that A/B testing can be applied to tests for improving landing pages, adtext testing, and image ads; how to A/B test; and how to in review and analyze the results and interpret the key metrics that determine winning and loser A/B test variants.
is a ppc audit right for you?
Think your campaign could be making you more money? Want some badass PPC experts to show you how? In this quick and affordable assessment we'll show you a bunch of opportunities you'd be crazy not to capitalize on.