When you are ready to dig into paid search, start with split testing. Split testing, also known as A/B split testing, is a simple, effective way to learn what works and what does not. And, while the front line of engagement can be a search ad; you will also split test the next link in the chain of action – your the landing page.
With standard split testing you compare the results (perhaps CTR) for one ad against another. Make one ad your “original” or control ad. The second ad is a test against your original. When you find out which ad performs better you can then add a third to the mix, testing the new ad against the ad which performed better previously. This process helps you refine your messaging to get the best results. Most digital ads have four components.
- Headline, 2. Description Line, 3. Image and 4. URL/link
Remember, you are a scientist and you want to find out what works and what does not. A standard A/B split test simply compares one ad against another. In a multivariate test, you test to learn which individual variable of an ad performs the best. So, if you are testing four different variables (headline, description line, image and display URL), you would need to write 16 different ads, representing all of the possible combinations, to learn which combination of variables works best. Rarely is this a feasible study because to discover your best ad combination you need statistically significant data. Unless you are a Fortune 100 company you probably don’t have either the budget or the volume of impressions you need to conduct a multivariate test. Stick with A/B split testing where you simply test the entire ad, even with multiple variables.
A Google search for “weed control” brings up several well known brands such as Preen and Roundup. Monsanto (manufacturer of Roundup) tests two different headlines and description lines in ads that drop clickers to the same landing page.
Move fast! It does not matter how large your budget or company is… you want to learn quickly what works and do more of that. The other way to look at it is “fail fast.” Find out what doesn’t work and stop doing that… Fast.
It doesn’t matter if you are running a paid search ad on Google or a paid ad on Facebook, Twitter, Instagram or any other platform. You can learn in one or two days which ad provide the best CTR. One of the variables in learning fast is going to be the size of the audience that gets ad impressions. This is the “statistically significant” aspect of testing. You can increase your audience size by widening the geographic area for viewership (i.e. statewide instead of city or nationwide instead of state).
Split Testing Optimizes Landing Pages
Once you know which ad brings your audiences to your landing page; you need to learn what combination of copy, images and calls to action will help with a conversion. This is your landing page split test. The results will help you achieve landing page optimization. Just like the ads, you create two different landing pages to test which one works.
Unlike ad split testing you want to be more specific about what you are testing. Focus on your conversion goal. If you have a specific action (a button or a link to click; or a form to fill out) focus the test on the copy/image related to that action. If you want to capture an e-mail address, then A/B the copy and the placement of form fields or pop-ups where you capture e-mail addresses.
The results you achieve from each variation of your ads and landing pages are carefully recorded and compared to determine which version shows the greatest improvement and results. This can be a daunting process if you are trying for continuous improvement and using an Excel spreadsheet to keep track. Consider using an online platform that does all of the calculations and shows you the results. A second piece of advice: use the KISS technique, or keep it simple, stupid. First test your ads to get the best CTR possible for each target audience. Second, A/B test two different landing pages to discover which landing page works best for your specific, measurable conversions. And remember, this is a continuous process. One version may work well this month; but next month your audience will respond to something different.
Know and Test the Platforms Your Audiences Use
It sounds obvious, but it doesn’t always happen. Every landing page needs to be tested on multiple platforms and devices. Android, Apple and Windows. See what your audience sees before you waste any money bringing people to a landing page that doesn’t work well.
Further, consider WHAT you are selling as your product or service may be platform specific or focused. For example, apps, ebooks, games…. it’s obvious that they are going to do better with ads that focus on mobile platforms. The same is probably true for fast food restaurants and services that people who are on the move and younger, millennial audiences are more likely to need.
