Share what you know with millions of people
Focus is the best place to turn what you know into remarkable content
0
What are some common mistakes made when doing testing for email marketing?
What can I do to avoid those mistakes and save some time?
Events
- Dos and Don'ts of Small Business Marketing May 29 @ 11 am PT
- Lead Nurturing 202: The Next Generation May 31 @ 11 am PT
- The Tricks to Paid Media June 6 @ 11 am PT
- Display Advertising for Brand Awareness June 20 @ 11 am PT
Related Activity
7 Answers
Hi Jessica.
I'll wade in here with a few common mistakes:
1 - Testing with too small of a sample. I like to use 10% of my database but there are no rules. However using just 1% of your database as the guinea pigs may throw up some nasties.
2 - Use random data - Obvious, but don't select your newest subscribers (for example) for testing if you plan to roll out the campaign across the database as they will likely act different to your oldest subscribers. Keep it random.
3 - Avoid jumping to conclusions - for example when running a split test on subject lines, don't rely too much on early info as results can change in the coming hours/days.
4 - Have a plan. Don't just test for the sake of it. Come up with a theory (I think doubling the size of my CTA button will increase clicks) and action the test. These theories will enable you to get a stronger grasp on what works.
5 - Don't test too much at once. It is hard to gain insight if you change so many things that you have no idea what is working best. I prefer to use A/B tests and change just one variable at once to see solid results. If done over time, it is possible to change and test many elements whilst retaining a clear insight on what is causing the lift.
6 - Don't Stop Testing. Constantly evolve your campaigns with testing. don't stop after one good result!
Hope this helps you out and hopefully you will get some more thoughts from others,
Jay Chambers
So Jessica,
Jay already named some very common mistakes. But seeing that you are a small business owner, the biggest win is in the prize. So think about what you want to achieve (your goals) and how you would be able to do so by optimizing your e-mail marketing.
OK, so if you want to sell more of a certain product, maybe you should firstly test product pictures, offers, landing pages, target groups, etc instead of for instance sendtime. The less testing you do to achieve your goals the better. So try and start off with the tests that have high potential gain and you can re-use.
And if you want to save time, some ESPs offer functionality/tools to test and optimize more efficiently. If you have any questions about that, don't hesitate to contact me.
Hi Jessica,
The mistake I've seen in testing is often testing too many variables at once. If you change the subject line of the email AND the landing page background color AND whatever, you can't control it.
One of my clients is a spokesperson for a large small electronics manufacturer. They test her infomercial in six markets wearing three different colored aprons for different segments in each city, if you can believe that! And it makes a quantifiable difference. Make sure you pick the best constants you can, and then sequentially test the variable with a representative population sample.
Good luck to you!
Wendy
Jessica,
Great question. What Jay, Jordie and Wendy said....but here is one other little nugget to consider as well.
A common mistake I always see is that people do email testing, see a result that would result in a change and then never do anything about it. One of the things about email testing that makes it so cool is that you never really know what direction it will take.
What I mean by this is that sometimes the results can be surprising and sometimes shocking. You must be willing to adjust your program as a result of the tests. In other words, make sure that what you are testing is actionable fromt the standpoint that EVERYONE and I mean EVERYONE in your organization is willing to make a change as a result of that test. It might be painful, it might be weird, but ultimately the subscribers in your program are in charge.
Hope this helps and good luck with your tests.
Andrew Kordek
Co-Founder, Trendline Interactive
An email marketing consultancy.
Twitter: @andrewkordek & @trendlinei
Andrew makes a great point here.
For instance, one of my clients is a large property portal. We are seeing some great results with simple, 90% text emails.
It is hard for us to imagine this to be the case, after sending the typical image heavy real estate ads. But the plain text ads are working well - and forcing us to rethink our plans!
Jay
The biggest mistake? Not testing at all. I've seen too many just skip past this and end up wondering why they made mistakes and why they underperformed. It's too powerful an asset not to use, especially when it comes to split tests.
Everyone here has made excellent points. Yes, we should be testing. That's what makes email great: The ease of testing. And I won't add specifics because the others have nailed them down pretty well.
Here are a few thoughts to add to the discussion:
1. Know your metrics and benchmarks first. You can't know if you're improving if you don't know where your metrics have consistently been.
2. Understand your purchasing cycle, and know where your natural peaks and valleys may be before determining winners. For example, I used to work for an online university where January and August were crazy busy months. Engagement would be higher in those months regardless of any tests, so I'd have to remove those months from consideration in terms of true testing.
My most effective testing occurred in the normal months, where I could better determine trends in metrics. And don't assume, for example, that your August numbers can be accurately compared to your July numbers if your purchasing cycle has a big difference between the two months traditionally.
3. Always test against a control. Always. Take your existing, benchmarked templates and creative and put them up against revised or entirely new templates and creative. If you think you can afford to, do a 50/50 split. If not, do a 90/10 or 80/20. Enough to have statistical significance.
4. Understand that testing is about finding a positive (or negative--though hopefully not) trend in metrics. This trend may take a few months to determine. So patience is key.
Answer This Question