Acquisition secrets can decrease your media costs for acquiring customers, there by increasing your gross margin, stat! With the mighty competition for the consumer dollar, we know that every penny counts. You may not even push these pennies to your bottom line. You may decide that improving your customer service will increase retention, so you opt to invest in your service and increase revenue through an increased lifetime value. Regardless of where you invest your savings in customer acquisition, we all know savings proves beneficial.
Many self-serve ad platforms have asked how we understand their platform and develop sophisticated strategies with little knowledge of their platform, including the famed Facebook Ads team. The answer is always simple, we pride ourselves on using fundamental direct marketing strategies; hypothesize, test, optimize, rinse, lather, repeat. Sounds simple…just like anything in marketing, it’s not. Attention to detail, documentation and consistency are the toughest challenges here. Funny enough, most think the challenge will be math and it’s not. You do need to understand basic clean testing practices and we will explain these.
Your first mission is to hypothesize. This is the fun part! Get folks from different areas of the organization to tell you what they think would work. You will be surprised by the creative accountant or the call center representative that really knows their customers. Hold quarterly meetings and invite different employees, managers and leaders to support this initiative. Each quarter you should have several dozen ideas. Many times, employee collaboration will develop innovative solutions and ideas through discussions that started somewhere else entirely. Celebrate this process, it’s a great team building and marketing exercise. Once you have your many testing hypothesis, you will retreat to the marketing team to determine which tests you will start with. Pending organization size and resources, you may start with only 1 test or you may have 10-20 tests going when you start. Direct marketing companies have 500 customer segments and dozens of tests going simultaneously within those segments.
Your next objective is to design your test. Without fancy software, you can do simple A/B tests. This means you test 1 variable only and identify the impact this has on your acquisition. If you try to test more than one variable on each test, you will not be able to determine with certainty or significance which variable impacted the results. Depending on the type of ad you are testing, you can have many variables to test:
- Copy writing in the physical ad
- Copy writing introduction on social media ad
- Call to action copy
- Inducement/Offer: Percent goes to charity, cash back, premium available, sale price
- Product imagery, positioning, lighting
- Audience selection: What demographics are you targeting for your prospect
And this is why we choose one variable and continue to optimize upon finding the winner of the initial test. A sample test would be to choose changing the call to action. Everything in the ad is identical leaving no chance to what changed response. Now you will select 2, 3, 4 or as many as you want to test the call to action, “Learn More”, “Buy Now”, “Get more info”, “Request information”, “Get Coupon Now”, “Limited Inventory”, “Limited Time Offer” with all ad components the same with the exception of the call to action words. The button must also be the same size, position and color. These simple tests can yield response rates so wildly different, that you will be surprised. I have seen up to 1,000% lift in response.
Below you can see a test design based on the hypothesis that color research has shown that consumers favor the color green and that it elicits happy emotion from people. So in one acquisition ad, we are using our brand colors and in another we are testing the hypothesis that people will respond better to the color green. Notice that EVERYTHING in the ad is identical with the exception of color.
You also must have enough people in each acquisition ad group to ensure your results are significant. All tests need to be statistically reliable and significant. In order to determine reliability formally, you can do internal or external reliability tests. For external, you can test 2 – 3 sample populations to determine results stay similar. You can determine internal reliability by using a fancy Chronbach’s Alpha, but for today…let’s just rely on statistician guidelines of how many people in a study will contribute to reliable outcomes; much debate over this number, but it has been commonly noted that a sample size over 30 will yield reliable results not due to error or mitigating variables (ours will be mush bigger with affordable Facebook ads). To determine significance, you will set 95-98% reliability expectation and use one of many significance tests to determine: T-test, ANOVA, Z-Test or Chi Square. Though for those of us without fancy stat software and an analyst to run it, we will be doing very simple and minor elements as to not leave mystery or debate. If your response is less than 5% difference between the 2 acquisition ads, you may consider testing a new variable for greater variance and lift in response. Also note that not only do you need a lift in response, but you want to ensure your cost per acquisition is lower as a result.
Now you unleash these acquisition ads to the exact same audience selection criteria. Be sure no matter which ad platform that you are using, to save your audiences so you can use the same one when testing. Also note this same ad can be turned into 5-10 tests by selecting different audiences. Now you have increased your testing matrix to not only test color on same audience but test audience on same ad.
Once you get your results and determine that this is the winner ad with variance, you will hypothesize again, set up new tests and go! The next item suggested to test would be the intro copy. Edit the copy and keep the images the same and you are on to your next test with yet again same audience selection.
This covers simple A/B testing you can do without fancy tools, statistical training or an analyst on board. Once you get a handle on this, you will be ready for finding your analyst and sourcing tools that can automate testing and include multi-variate testing, meaning you can test more than one variable at once.
Share with us what your most surprising test was, and why.