Deliverability, Engagement, and the Theory of Email Marketing
Ever since the final session at EEC 2015, it’s been a seismic few weeks for the email community when it comes to understanding deliverability from the perspective of the inbox providers. In that session, a seemingly innocuous question from yours truly turned what would have been a worthy, but predictable panel on deliverability featuring 4 major inbox providers (AOL, Comcast, Gmail and Outlook.com) into one of the most controversial and talked about panels in the history of the EEC.
It would be disingenuous to say I wasn’t expecting some controversy to result from my intervention, but even I have been surprised by how visceral the reaction has been. Looking back, I should have been more prepared because my question exposes a deep and enduring fault line within email marketing with a simple word at the heart of it: engagement.
Putting revenue to one side (because everyone claims their approach will lead to more money), the two sides can be broadly defined as follows:
- Those who believe that engagement is the goal and should be measured by rates such as open, click and unsubscribe (campaign-level metrics).
- Those who believe that email is like any other marketing channel with the goal of maximizing the number of people who get the message and the number of times they see it, also known as reach and frequency (subscriber-level metrics).
For a fuller understanding of this I would urge you to read this white paper in Digital & Social Media Marketing, authored by me.
The reach and frequency point of view roughly translates into “don’t remove inactives” and “send more email” and this is what drives my detractors mad. Even more so if you do not include the infamous rider that goes with everything I say or write… “and don’t be stupid (#DBS)!”
I acknowledge what follows is a simplistic view in most organizations but typically the people who sit on my side are the business owners, those who get fired if the company doesn’t hit its targets. Those who sit on the engagement side are the people who are fired if the email doesn’t go out, render properly, has the wrong offer, etc. These are the guys who are woken at 3am if there are deliverability problems. There are some who wear both hats. But historically thought leadership in email has been driven by the latter camp (and fear and self-loathing has made people listen despite the lack of data – more on this later).
Most deliverability experts (and I don’t mean the abusive zealots who set up and monitor spam traps) are regular, hard-working, much-put- upon guys and gals who work for ESPs. They are the ones who get woken at 3am and harassed if your emails don’t get to the inbox. They are also at least one step removed from your business goals. Sure, they care if you leave money on the table, but they don’t get fired and, as you are locked into a 2 or 3 year contract, the account team have plenty of time to make it up to you.
This is why, up until this year’s EEC panel, reach and frequency have not sat comfortably within the deliverability camp. Let me explain. All things being equal, if brand A sends two emails a week and brand B sends 3 a week, brand B wins. Sending emails to those who haven’t opened in a while, say 12- 24 months, is also VERY profitable, particularly if you include sales driven by unopened emails in other channels – what we call emails “nudge effect”. It is very difficult, if not impossible, for anyone who doesn’t agree with this to provide consistent empirical evidence to the contrary.
But what both of these approaches will do is lower the campaign level-metrics used by deliverability experts as measures of engagement. So this is the data that’s used when they play the deliverability card, which in turn feeds what I call “fear and self-loathing in email marketing”. This is the fear that everyone hates email, inboxes are overloaded and your emails are being marked as spam or not making it to the inbox. Again, these fears are not based on facts and if you don’t believe me, then check out the latest State of the Inbox Report by Return Path.
As soon as I heard that influencers Dennis Dayman and Ryan Phelan had persuaded 4 of the major inbox providers to a panel about deliverability and engagement, I realized that this was an important opportunity to shed light on one of the most vexing questions to face email marketers: the nature of engagement. And although this issue is not a new one (I first wrote about it in 2010), it was a chance to clarify and get on record what inbox providers mean when they say “we measure engagement” and how that impacts inbox placement.
For those who were not there in the preamble to my question, I pointed out that because the inbox providers have never stated for the record what engagement they actually measure, there has been a knowledge vacuum. Unfortunately some people, consciously or otherwise, take advantage of this vacuum to pass all responsibility for deliverability on to the client and scare them into leaving money on the table for the sake of an easy life.
So not only did the panel definitively provide everyone with those answers, they threw in a couple of real surprises too. They said:
- We do not use clicks as a measure of engagement
- User engagement will not affect your overall reputation
- If you remove people who do not open or click you are leaving money on the table
- Microsoft, Gmail, and Comcast clearly stated that they do not convert old accounts to spam traps.
In other words they do NOT use campaign-level metrics to measure engagement.
They do however track the following:
- Open - although they know that open has become a less relevant metric (images downloaded by default in certain email clients), they still track it
- Reply - a reply to a message is considered a super-strong signal of engagement. If you ever needed evidence that using a “no-reply@…” email address is a bad idea… here we go!
- Move - to junk strong, negative signal. Two of these on AOL are enough to automatically place that message in the spam folder from then on, for that recipient.
- Not junk - strong, positive signal that the message should not be considered spam. One of these on AOL is enough to “reset” the previous behavior.
- Delete without open - a quick glance at the sender/subject, and they didn’t like it: a somewhat negative signal.
- Move to folder - if you are moving certain messages around, it means you care about them. Add to address book it shows that the sender matters to the recipient
All of these metrics are measured at the inbox (subscriber) level, in other words your open rate may be high, but your message may still not get delivered to a particular inbox. The only way to report on that level of engagement is to measure every interaction to every email over time. And as far as I am concerned that is great news for my clients.
"We often refuse to accept an idea merely because the tone of voice in which it has been expressed is unsympathetic to us."- Friedrich Nietzche
With the above quote in mind, I’d like to start with a bit of context about me and my thinking in the hope that it allows me to get my points across…
I have ALWAYS been pro engagement. If you are a responsible email marketer with household-name brands for clients, you have to be. But what may surprise all but those with long memories is that I too used to advocate brands remove inactives from their lists, most memorably at a Silverpop client conference in 2006. There I played The Dead parrot sketch by Monty Python to the audience and followed it up with a presentation referencing The 6th Sense by running around the stage in true Dela style yelling “I see inactives! They’re everywhere!”.
Back then we had been monitoring email activity at the individual subscriber (inbox) level for about a year and, unsurprisingly given what we know today, the number that jumped out at me was this: between 45% and 55% of people on our clients’ databases had not opened or clicked a single email in the previous 12 months. Incidentally, all of the clients we were monitoring had above average open rates and none of them had deliverability issues.
I shared this data with the audience. And the solution I advocated? Stop mailing them. Not for reasons of deliverability, mind you, but because I assumed the lack of engagement had to indicate that these people did not want to be on the list. They were the so-called emotionally unsubscribed. The irony that I may well have been the catalyst that created a generation of people who think that this is true is not lost on me!
So what made me change my mind? In a word: data. Our clients wouldn’t listen to me when I told them to stop mailing inactives, so we tried to prove them wrong using our subscriber behavior data.
The first client I tried to persuade to purge their list of inactives was a content provider that sent a weekly newsletter to a list of 2 million subscribers. 700,000 had not opened or clicked a single email in 12 months or more, so we started to monitor their engagement at a subscriber level across three distinct groups:
- Group A (active): subscribers that interacted in the last 12 months
- Group B (become inactive): contained no subscribers
- Group C (currently inactive): subscribers that were inactive for 12 months or more
For the next 3-4 years, every time we sent an email we would move anyone in Group C who opened or clicked into Group A. Anyone from Group A who crossed the 12 month inactive threshold moved into Group B, where they were treated as inactives but kept separate from Group C. Guess what we found? The inactive Groups B & C stayed about the same size, and so did the active Group A. Why? Because the number of people reactivating was about the same as the number of people crossing into the inactive pot – roughly 3% of the database per mailing.
I’ve never come across anyone else that has conducted a piece of research like this over such a long period of time.
But my light bulb moment came when looking at what happened to Group C – the so-called inactives. The group started out shrinking by about 3% per mailing. However, slowly but surely the percentage of people re-engaging with the newsletter from this inactive group began to drop down to 2%, then 1%, then 0.5%. Eventually there were about 180,000 people who had not opened or clicked ONCE in nearly 4 years. A group of 700,000 subscribers that I’d been looking to purge had shrunk to 180,000. Put another way, 75% of those inactives had re-engaged when being mailed after 12 months.
Even after 4 years of no activity, there was never an email sent to this group that didn’t have any interaction with it. What we did observe at this point was almost every open resulted in a click on the unsubscribe link. Or in other words, a list is dead when the open to unsubscribe rate tends towards 100%. At that point, after 4 years and genuine insight into subscriber behaviour, they did stop mailing them and removed the remaining addresses from their list.
The data had spoken. And I’ve never said or recommended anything since if it hasn’t been backed by the data. Which is why I never recommend removing addresses from a list that has been inactive for less than two years UNLESS the return from sending to them is that revenue or traffic is less than the cost of the send.
So you can imagine I find it extremely galling when people armed with nothing but open rates or a survey intimate that I do not know what I am talking about.
"Daring ideas are like chessmen moved forward: they may be beaten, but they may start a winning game." - Goethe
At Alchemy Worx we do not seek business from companies that have problems with deliverability. If we are approached by such an organization, we refer them immediately. All our clients run opt-in programs and, if they haven’t done so already, we help them implement and/or maintain best practice in list acquisition & management, feedback loop monitoring, removal of unsubscribes, etc. As a result, we rarely have to discuss deliverability as an issue. That does not mean they NEVER have deliverability problems or have someone at their end do something stupid. But by and large neither of us spend all our time at DELCON 1 (geddit?) state of readiness.
To be frank, if there are any members of the Only Influencers list, EEC attendees or experienced marketers running serious and long running programs who do not take deliverability best practice seriously, I would be very surprised indeed.
If maximizing reach and frequency are your goals (and they should be), then you have to stop calling non-openers inactive or unengaged and start calling them your list of opted-in prospects. Every email you send to them is then an opportunity to convert them into openers, clickers, traffic or, best of all, revenue. The last two goals can be achieved whether the email gets opened or not… because you don’t have to open an email from GAP with the subject line “Great Deals in Store this Weekend” to understand the message and respond via another channel.
The only reason ever given for not following this sensible advice is the dreaded “spam trap” or “honey pot”, where an ESP turns an inactive or dormant email address into a trap that will get you blacklisted if you continue mailing it. So to hear Outlook.com, Gmail, and Comcast clearly state that they do not convert old accounts into spam traps was a wonderful moment indeed.
It is not that spam traps do not exist. They are just predominantly set up and operated by 3rd parties with their own agenda. They are also fairly easy to avoid if you run an opt-in program and you use an address verification service at the point of collection. Think about it - a spam trap address cannot subscribe to your list without human intervention, either by accident (a misspelling) or with malicious intent.
More recently deliverability experts have started to recommend ramping down send frequency to email addresses that have not opened for 3-6 months. This is a relatively easy sell to marketers that are under commercial pressure or question the wisdom of removing inactives due to the lack of empirical evidence. I know Sri Somanchi recommended exactly this, but I struggle to see how he could when open rate is only one of the measures he uses. We can’t see the rest.
What if, for example, the person moves every email into a ‘save for later’ folder without opening the email? I challenge anyone to provide empirical evidence that a respected brand sending email to an opt-in list can increase engagement as defined by Gmail (and therefore a proxy for inbox delivery) by ramping down their send frequency to 3-6 month non-openers. Now I am not talking rates here, I am talking totals over time – both for engagement AND inbox placement
When it comes to send frequency, I give a wry smile that your ESP bombards more inboxes with more email per day than you or I ever could. How many different email messages do you think Salesforce or MailChimp manage to get into AOL, Gmail or Outlook.com inboxes on behalf of their clients a day? 30,000? 100,000? If they can get that many into the inbox, why would you have a problem getting 4 or 5 messages a week through the door? It seems to me that the ESPs are asking us, their customers, to do their work for them and make their life as easy as possible. Come to think of it, that’s not a bad business model at all!
"If at first, the idea is not absurd, then there is no hope for it." - Albert Einstein
To truly understand engagement you have to do something counter-intuitive - walk away from campaign-level metrics such as open and click rates, which are the staple diet of email marketers the world over. Instead you must do what the inbox providers are effectively doing themselves - embrace subscriber-level metrics. They have dropped rudimentary rate-based metrics such as spam complaints and started looking at the inbox or subscriber level.
Campaign-level reporting, such as you get from your ESP, is what most email marketers use and is pretty simple to understand and calculate. Every time you send an email, you measure how many people got the email (delivered), opened it, clicked it, hit spam, or unsubscribed and then divide those numbers by the number of people sent that message. However, it is a HUGELY misleading measure of subscriber engagement because it looks at the message and not the subscriber who got it.
Subscriber-level reporting is very simple in theory. You measure every interaction every subscriber on your list has with every email you send over a given period of time. When you send an email, every subscriber on your database can either be sent that email or not, receive it or not (delivered), open it or not, click on it or not and mark it as spam or not. Simple.
But to truly measure engagement, not only do you have to repeat that process every time you send an email, you also have to connect each of these individual subscriber interactions with all of their previous interactions, whether you sent them an email or not.
You want to talk big data? Imagine doing those calculations for a company with a database of 20 million that sends 5 emails a week and has over 1,000 trigger programs set up. That’s pretty big data and we’ve been crunching those numbers and interpreting them at Alchemy Worx since 2006. (As I’ve already mentioned, that’s EXACTLY how the mailbox providers measure engagement, albeit with different metrics - read rate, move from and to the spam folder, delete without open etc.).
So without subscriber-level measures of engagement, you will never be able to optimize deliverability at the inbox level. Nor will you be able to accept that sending more email DRIVES greater engagement. To prove it, I am going to share some data from Touchstone, a subject line testing and optimization tool developed here at Alchemy Worx. The data is based on analysis of subject lines sent to a combined send volume of over 21 billion emails across a range of different industries.
As you can see, the median open rate DOES fall if you send more email! In fact, it is so consistent that you should never benchmark your open rate unless you take send frequency into account. So if you use open rate to measure engagement, you are driven to the erroneous conclusion that high frequency is a bad thing despite the fact that your revenue, total opens and total clicks go up. This is why so many marketers are conflicted.
Now let’s look at opens per person per year:
If you send at a frequency of less than 1 email per week, then your median number of opens per year per subscriber is approximately 5. This number rises to 50 opens per year for daily senders and 30 opens per year for brands who send 3-4 emails per week.
So the question is: which of these two marketers has the more engaged subscribers?
- Marketer A - open rate of 18% by sending 1 a week and generates 5 opens a year per subscriber
- Marketer B - open rate of 13% by sending 6 emails per week and generates 35 opens a year per subscriber
And most importantly of all: who do you think the inbox providers will think has the most engaged subscribers?
So the data tells us that frequency not only drives greater engagement from your actives, it also drives greater re-engagement from your inactives. Why? The answer is absurdly simple. Your subscribers cannot engage with an email you don’t send them.
Email Marketing - The Theory of Everything
"There is one thing stronger than all the armies in the world, and that is an idea whose time has come." - Victor Hugo
What I find most surprising about this whole debate is that people are concluding that I’ve finally got my comeuppance because inbox providers are now measuring things in such a way as to comprehensively debunk the idea that reach and frequency have any place in email marketing. If anything they have both validated my arguments AND removed the number one argument against my logic and the data – deliverability. In particular they have shattered the pervasive myth that inactives are dangerous because addresses are turned into spam traps and that high send frequencies are dangerous because they depress click rates.
So deliverability hasn’t suddenly become personal nor is the Third Age of deliverability (as some have called it) anything new to us here at Alchemy Worx. For nearly 10 years we have been using the exact same techniques as inbox providers! If anything the time has come to join them in moving beyond tired old campaign-level metrics and embrace subscriber based metrics like Open reach and Click Reach. It is the future.
Since the inception of email as a marketing medium, it’s been thought of as a bottom-of-the-funnel marketing tool – driving conversions and purchases using data gathered from previous emails and other channels. As a result, email marketers have been laser-focused on sending the right message to the right person at the right time. However, this has created a dominant way of thinking that views email marketing as a staccato series of singular events. It’s not wrong, but that approach doesn’t allow email marketers to fully realize email’s potential through the rest of the customer journey.
Used in the right way, email can raise brand awareness and shape consumer behavior at the top of the funnel. In 2015, brands will need to consider the power of email at both ends of the marketing funnel, particularly if they want to exploit the dominance of mobile devices in the lives of consumers. In this 24/7 world, the smartest marketers will look beyond the narrow confines of open rates and click-through rates. To do this, marketers will need to start delivering highly personalized content, which in turn will allow them to increase the number of brand impressions generated.
True engagement is driven by reach and frequency and the only way to achieve that successfully is through personalization. Which is in turn only achievable by truly automating your email campaigns using the latest generation tools provided by the better ESPs. Your mantra for 2015 and beyond should be the one we value here at Alchemy Worx and advise to all our clients: send highly personalized messages to everyone, every time.
First off thanks for the article Mr. Quist, definitely a ton of invaluable information. I do have one question for you if you have a moment, you mention there's never consistent empirical evidence to remove customers from an email list.
You also mention you started Group C with 700,000 inactive customers and after a few years of that had dropped to 180,000 and during that time period the percentage of activity that resulted in unsubscribes started nearing 100% so you decided to remove those 180k customers.
I'm not trying to knit-pick here more so clarifying so I can better understand the solution, but there is still reason to delete emails from your campaign list but it should be based on the % of users after a particular period of time who are only opening emails to unsubscribe nears 100%?
Frequency it's a challanging topic. To have the complete picture it would be helpful to have charts also regarding FBL rate and unsubscribe rate. Thanks