Have a quick read about our Magic Green Button, a new tool that helps you avoid spamming without losing money.
Email marketing and spam
Whenever I tell people I make a product for Email Marketing, the typical response is something like “uhhh… so you make a tool for sending spam”.
Unpleasant as this may be, their reaction is understandable. I am convinced that email can be a useful channel, but the current state of email marketing leaves much to be desired.
So why is email marketing in this sad state? I think it’s because marketers have been put in a difficult position:
- On the one hand, there’s a learning curve. Direct marketing emerged relatively recently, there was little existing experience to draw on and even fewer experts who could pass on their skills. So ascending the learning curve has been process of trail and error.
- On the other hand, there is immense pressure from above. Management demands swift and outstanding results and is quick to punish failures and shortcomings.
The only solution to this conundrum is to take all our knowledge and experience gained from outdoor advertising, TV, newspaper and contextual advertising, then carefully adapt and apply it to direct marketing.
For the most part this means scaling up, making gradual adjustments and measuring their effects in Google Analytics. And, unwittingly, many of us fell into the trap where we have no choice but to keep sending more and more and more emails to keep our metrics up in Google Analytics.
The Magic Green Button
Here at Mindbox we’re determined to save the day and help marketing folks get off the email treadmill.
To this end we set out to develop a feature that would:
- be fully automated with no need for human supervision
- send fewer emails without reducing revenue
The above should also lead to happier subscribers who make fewer unsubscribe requests, and this should in turn lead to increased revenue over the long term.
How it works
A simple algorithm figures out a comfortable rate for each individual subscriber to receive emails at.
If a subscriber expresses interest in mailings, frequency remains the same or steadily increases.
If a subscriber’s interest decreases, mailing frequency steadily decreases to match.
We evaluate email opens, clicks, site visits (from any sources, not just from email) as well as purchases to calculate interest level. Each parameter has its own weight and time periods.
The email frequency is never reduced to zero. We continue to make contact even with a person who does not show any interest as long as they do not unsubscribe. The minimum sending frequency is configurable.
How we know the algorithm works
We used the same method as in A/B tests:
- We randomly split the subscriber base into two equal-sized groups. We enabled email frequency control for the first group and left the second group as it was before.
- We compare revenue, orders, and open rate between these two groups.
It’s important to note that we look at all revenue orders to calculate the pure net added effect, without making any attributions that would skew the results. Because if we tried to factor in attribution, for example using the default Google Analytics method, our results would be all messed up.
Frequency control reduces the cannibalization effect, and it’s very possible that we would see a drop in revenue even in places where there was actually no drop at all.
The short-term results were exactly what we hoped for: we reduced our unsubscribe rate without losing money.
The charts below represent the results of 1-2 months of testing.
The number of emails sent is reduced significantly while reach remains unchanged. Each recipient gets at least one email per month.
Revenue, conversion and the overall number of orders don’t change
(probability with 95% confidence interval)
Unsubscription rate has significantly decreased while Open Rate, Click Rate and RPE are greatly increased
(probability with 95% confidence interval)
We ran this test in businesses from different industries with subscriber bases ranging from tens of thousands to several million subscribers. Results were similar everywhere, but nonetheless with some differences. We compiled all the results into a single table for a complete picture.
|Results range*||Statistical significance|
|Revenue||from -1% to +8%||insignificant|
|Conversion||from -1.5% to +5%||barely significant|
|Number of orders||from -1.5% to +5%||barely significant|
|RPE (income per email)||from +90% to +200%||significant increase|
|Open rate||from +40% to +115%||significant increase|
|Click rate||from +40% to +120%||significant increase|
|Unsubscribe reate||from -35% to -50%||significant reduction|
* (“With frequency control” – “No frequency control”) / “No frequency control” x 100
Leonid, Marketing Director
We decided to try the email frequency control algorithm because while we didn’t want to spam our customers, we also didn’t want a lot of overhead in selecting recipients.
We expected the same overall effect from our email channel as before, but with a reduction in unsubscribe rate and fewer ignored emails.
Those expectations were met in full. I would like to see the service developing further and becoming even smarter.
Irina Bokarcheva, CRM and Email Marketing Manager
We understand that there is such a thing as too many emails, even for our most loyal customers. So we gladly accepted the offer to participate in Mindbox’s pilot project to test the new frequency control algorithm.
After the first week of testing, open rate and click rate increased doubled, and total revenue from emailing didn’t change despite the income per email being significantly higher. It’s very interesting to note that for the first few weeks the unsubscribe rate for the test and control groups remained identical. But after a while the unsubscribe rate in frequency control algorithm group halved in comparison with the other group.
It’s unfortunate that the algorithm doesn’t yet take email content into account. Most of our manually-created mailings are in the format of a regular newsletter, so we’d like to account for not only the optimal frequency for each customer but also their interests in terms of content.
It’s an interesting project, the algorithm works and we’ve seen a positive dynamic for the whole three months. We hope that it will soon be possible to account for content preferences as well as frequency.
Karpenko Julia, email-marketer
Email marketing is a powerful sales tool for our company, so we are constantly working to improve its performance.
In the last quarter we started thinking about how best to hold on to our subscriber base, and we ran an experiment with Mindbox to reduce the number of unsubscriptions.
The idea was to adjust the frequency with which customers received emails, depending on their response to previous mailings.
As a result we reduced our unsubscribe rate by 46% and retained the loyalty of our subscribers without any reduction in revenue from email.
First and foremost we need to wait for long-term results to show that revenue grows over time.
If that hypothesis is proved, we’ll move on with development:
- Account for content preferences
- Extend to other channels + perhaps also identifying the best channel per customer?
- Cater to needs of different industries
- Deploy Mindbox everywhere and create a new standard for email marketing =)
If you’re a Mindbox customer and would like to try email frequency control, just write write to your personal manager.