You’ve just sent your 10,000 customers an email with a terrific offer on your 2013 Chardonnay. (You do need to make room for the incoming 2014 and you still have lots of the 2013 left.)
And you wait for the orders to start rolling in. And you wait. And wait. And those pallets of 2013 Chard are still sitting.
What went wrong? The wine is good. The price is excellent. The offer sounded appealing to you (and to your staff). But it didn’t move.
Now what? Ah, perhaps the email never went out. That must be it. So you check your list for your seeded email. (What does this mean? It means you deliberately added emails for yourself, your family, your friends on this list so you can know or ask if the email was received. Maybe your email service reported it as being sent, but it got locked up in the computer.) Uh oh, they all got the email. So probably everyone got the email.
Could it have been the wording? Yes. Could it have been the placement? Yes. Was the “buy” button inconveniently placed? Yes. Was the typeface too small or too big? Yes. Could the offer have been better? Yes. Could shipping costs be too high? Yes.
Yes to all. But how do you know what the response will be before you send out the promotional email? You don’t. Hence, A/B testing.
What is A/B testing? That’s creating two versions of the email, but not sending both of them to everyone. You take a small subset of your list and send some of them Test A and some of them Test B. (You can create multiple versions of the tests, but we won’t call this A/B/C/D testing.)
This sounds simple enough. It’s not though. It takes time and thought. But it’s necessary. You can do a limited test or you can do a more complex test. But you do need to test, to see which version produced the most sales, and then send that version to the rest of your list.
Let’s start with simple, especially if this is the first time you’ll be doing A/B testing.
The point of an A/B test is to see which version of your email will produce better results. Now there are so many areas to look at, so this is really a never-ending task. We’ll mention some of the things you should look at, but for now, let’s start with some wine pricing choices.
You are going to put some wine on sale. Do you:
- Offer the wine at 20% off?
- Offer free shipping* with a case purchase? (*or shipping included if you’re in California)
Let’s say your wine sells for $300 per case. 20% off will save the customer $60. Free shipping on a case will save the customer anywhere between say, $30 locally to $60 going across the country.
Should you send out two emails? Can you determine where someone lives with info from your database?
Supposing you sent 100 people the percentage off the case, and another 100 people the free shipping option. Based on the sales in the next few days, you can then decide which offer to send to the rest of your mailing list.
If you only chose to send out one offer to everyone, and few people choose to buy, would you know why?
There are many other choices too. You could say 20% off or you could say Save $60! Which is better? Some retailers know when to offer a percentage as a savings and when to offer a dollar amount. If you don’t, perhaps you can simply use your email to find out. What you need to do is make your offer enticing. A/B testing is a good way to know the route you should follow.
There are many things you can do via A/B testing. But if you make many changes between Test A and Test B, you won’t necessarily know which version triggered the change. So it’s best to start with one for each email you are planning on sending and simply compare the responses. You can take the choice made that produced the best results, keep that in your next email, and then change something else for the next one.
Here are some thoughts of what are the best items to test (and best in your case may be different from someone else’s best):
- What is the best way to phrase a subject line: Test A: 2013 Chardonnay on Sale! Or Test B: Save 20% on our 2013 Chardonnay.
- Personalize the greeting or not: Test A: Dear Bob. Or Test B: As a customer of Winery B…. (The danger of personalization is that if your database contains errors, you might wind up with “Dear (firstname): or “Dear Jones Alice” when the name is Alice Jones.)
- Call to action button: Test A: Buy Now! Or Test B: Save Big! Click Here!
- Text sizes, placement, number of columns, large pictures, small pictures, no pictures, mention reviews or medals or comments from happy customers, or not mentioning them, the list goes on.
There are really so many choices to test, but you can’t do them all. But as you do each one, check your analytics for the results. Do more people buy from Test A or Test B? Are the results similar, but more people clicked on Test A than Test B, but you didn’t convert them?
Run your tests at the same time. Don’t send Test A on Monday, and Test B on Tuesday, though it might make sense to A/B test your emails on Monday the first time, and Tuesday the second time. (Or whatever days and time of day you choose.)
There’s a never-ending series of tests you can do. But you should test. After all, you do want to sell your wine. And increased sales pays for the time you took to do these tests.
By the way, does everyone (does anyone) ensure that their emails can be easily read on a mobile device? And better yet, link to a mobile version of your site? What about if you texted some and emailed others? But those are separate blog posts. More and more people are using mobile instead of PCs. But we’ll deal with them (and you) in upcoming posts.