如何运行A/B测试[FAQS]

When was the last time your boss said, "We have enough leads for the day. Why don't you head on home?"

是的...我要刺，说那从未发生过。为什么？根据经济文献的报告，只有28％的营销人员对他们的转化率感到满意。It's more than likely that your team falls into that unsatisfied bucket.

One of the best ways to get out of that conversion rut is to embrace your mathematical side and run anA/B test。这是一种实验，显示了两个不同版本的一件内容（例如着陆页，电子邮件或通话行动），向两个类似大小的受众群体展示了哪个表现更好。

(奖金：您可以在营销中以多种不同方式使用A/B测试，并且它们相对简单。）

If you're one of those marketers who isn't satisfied with your conversion rates, keep reading. I'll walk you through each step of designing, implementing, and measuring an A/B test. You can use these steps for anything in your marketing -- we're going to explain the process using呼吁行动(CTAs) for our example.

Want to follow along to help get your own A/B test up and running by the end of this post?Download our free call-to-action PowerPoint template to cut your design time in half.

准备运行A/B测试了吗？这是您的工作方式。

1) Decide what you're going to test.

The great thing about A/B tests is that you have two pieces of content to test -- but you can test both large and small elements in your marketing. You can test something as small as the color of a CTA to something as big as an entirely redesigned page.

当您在两个内容之间添加更多差异时，您需要记住的唯一一件事是，您只能将结果归因于您整体测试的每个内容 -notindividual differences. This means that if you're testing two versions of one landing page against each other, and you've changed the CTA copy, the form length, the image you've added, and the headline copy on one of the landing pages, you不能attribute that landing page's success to the form itself. You'd have to attribute success toallfour of those elements.

需要您自己的A/B测试的想法吗？查看本文以获取一些小巧的胜利and这篇文章从只有for advice on running larger A/B tests. If you're trying to fix your visitor-to-lead conversion rate, I'd recommend trying some landing page, email, or call-to-action A/B test.

For the purpose of this article, we've decided to run an A/B on our CTAs to what changing the color of the button does.

2) Figure out the goal of your test and decide how to measure it.

为了进行成功的A/B测试，我们不能仅仅在阅读上一步的最后一句话后才达到直接运行。您必须更深入地思考要通过A/B测试发现的内容。您是否要衡量CTA的颜色如何影响有多少人点击它？那是您可以做的最直接的测试。但是您还可以测试以查看颜色是否影响有多少人点击more than onceon the CTA.

在我们的示例中A / B测试,我们想要发送大量的people to the landing page, so we're going to use the number of clicks on the CTA as our indicator of success.

3)设置您的控制和治疗。

Ignore the jargon -- the control and treatment of your A/B test are quite simple. The control is simply the "Version A" of your test -- it's what you normally use as your landing page, email, call-to-action, headline, etc. The treatment is the "Version B" of your test -- it's the version that has the changes you're trying to test.

在我们的示例中，控件（A版本）将是深灰色，这是我们大多数博客CTA的颜色。这是现状，规范。治疗（B版本）会有所不同 - 假设明亮的蓝色。

4) Create your A/B test and release it into the wild.

Once you've design how your experiment is going to work, get on with making it! First, actually design and创建内容对于您的控制和治疗：在我们的情况下，下面的灰色CTA和蓝色CTA。请注意两者之间唯一不同的是颜色 - 两者的副本和图像都是相同的。这样，我们实际上可以测试颜色是否影响点击数。

Variation A:

Variation B:

然后，您必须在营销软件中设置A/B测试。bob电竞官方下载每个工具都是不同的，通常，对于要测试的每种类型的内容，A/B测试步骤都不同。如果您是HubSpot Enterprise客户，他正在跟随我们的示例，哟u can use these instructions。If you're a HubSpot Enterprise customer running a different A/B test,这是其他可以帮助您的资源。bob体育苹果系统下载安装

5) Promote your test -- but maybe not to everyone.

如果您想进行测试根本没有任何意义 - 换句话说，是statistically significant, which we'll go over in the next bullet point -- you'll have to promote the heck out of your content. Send your email to a large enough list, promote your landing page across social, or even throw some PPC behind the blog post link to get enough people to see your test.

Keep in mind that if you're running an A/B test for a specific audience, you need to keep your promotions tailored toonly那个观众。例如，假设您很好奇Twitter关注者是否会在着陆页上喜欢某些内容，那么您不想在Twitter以外的任何地方推广A/B测试内容。不是在Facebook上，而不是通过电子邮件 - 仅通过Twitter。

在我们的示例中，我们只是在研究CTA转换，因此我们会重新宣传博客文章，以使对页面感兴趣。

6) Gather data until it's significant.

Now comes the waiting game. Keep promoting your test until it's statistically significant -- one way of saying when the results of your tests are determined to be most likelynotdue to chance alone. If you're really into math and statistics,您可以按照此处的步骤来计算意义。否则，您可以使用这个方便的花花公子工具。Once you've hit significance, you can see if the treatment is more effective than the control.

But what happens if you never hit significance? Wait a few more days. Sometimes, it can take up to 30 days to get enough traffic to your content to get significant results. That being said, if it's been a month and you've sent a lot of traffic to your test, but you haven’t seen a statistically significant result, then your test probably wouldn’t make a big impact on conversions. Don’t be afraid to move on to another experiment.

7）调查您的整个营销渠道。

好的，现在您知道您的实验是否适用于最初设置的指标。惊人的！但是你不能停在那里。

Even though I told you to only focus on one metric before, this is the one time in the test where you can look outside the test's original purpose to see if it's had effects on any other part of your marketing funnel.

虽然认为更改CTA上的颜色可能会影响点击数量以外的任何内容可能会很愚蠢，但它可能会发生。如果你有closed-loop analytics, you can track to see if people who clicked on that CTA actually become customers. Maybe the people who click on blue CTAs become customers faster.

That's probably a ridiculous claim, but you see what I mean. By looking at other parts of yourmarketing process, you may discover that an A/B test has consequences you didn't anticipate. And if those consequences are good, you can lean harder into them. Otherwise, you might want to rethink whether you should make that change.

只需记住，您的A/B测试比仅仅是测试指标本身具有更大的含义。

8）迭代您的发现。

现在，您已经收集了数据（重要与否），并检查了您的A/B测试是否有其他意想不到的后果。你完成了！

Just kidding. You've finished your first A/B test -- and that's cause for celebration -- but there's so much more you can test. In the CTA example, you can try placing the CTA elsewhere on the page or see if switching up copy can affect how many people click through.

Or maybe you just don't trust the results of the A/B test you just ran. Maybe you ran it during a holiday and got a ton of traffic to your site -- but that's not indicative of how your audience normally behaves. Run the A/B test again -- except make sure you're not doing it during a holiday.

如果您一直在测试，则可以在转换率方面取得长足的进步 - 嘿，甚至可以使您的老板感到满意，以使其对他或她的团队的转换率感到“满意”。祝你好运！

Have you run A/B tests before? Share any secrets you have about the process with us in the comments.