For the last couple of months, I have been trying to get my head around the analytical nuances of email-centric marketing (as a part of my new gig with Responsys). At a first glance, there seems to be a plethora of metrics – some around deliverability, then open rates, click rates, conversion rates, etc, etc. Which ones of these are the true measures of success? To some extent, they all are – but I can’t help but look for a top-down hierarchy which looks at the bottom line first, and then delves into the various supporting factors.
The success of your email (or any other customer contact for that matter) depends on whether the email was successful in influencing a desired action on the recipient’s part -- Did they actually open the email? Did they actually click on any of the external links in the email? Did they actually perform the action that was the goal of the email – like making a purchase or signing up for a program? Measuring each one of these events and understanding how they are influenced by external factors are crucial to optimizing the success of your email campaigns.
A common misperception about emails is that they are more-or-less “free”. Compared to the cost of a direct mail piece or a telemarketing call, the physical cost of sending an email is definitely lower by several orders of magnitude. As a result, what is also common is that organizations get somewhat sloppy about measuring the cost and impact of their email campaigns, and they tend to measure email campaign effectiveness on a more general level.
Particularly important are the hidden costs of email campaigns. Since it is easier and cheaper to send several emails to a recipient, often “list fatigue” is realized much sooner in the email than in offline channels. This means people in the email lists opt out more quickly and the response rates decrease more sharply, indicating a more negative experience for the email recipients.
This breakdown in relationship manifests itself in the following email engagement metrics:
- Decreasing click-through rates and conversion rates
- Higher opt-out and spam-out rates
However, a common trap most email marketers fall in is to address the above issues by increasing the volume of emails. The thought pattern is something like this:
I am sending out 100,000 emails today, and I only get a 0.1% conversion rate. Since my target is to drive 10,000 new web sales by email this quarter, I must send out 10,000*100 / 0.1 = 10,000,000 emails this quarter. And because it costs me only a penny per email, I am basically spending $100,000. If the gross margin per sale is higher than $10, then I even have positive ROI.
While this calculation may be mathematically correct, the major flaw in this approach is that it evaluates email purely on short-term revenue potential, and does not consider the impact of email volume on the overall quality of customer engagement. In a way, it is very similar to the bad image telemarketing has gotten over the years because telemarketers have been entirely focused on making a large volume of calls without caring much how many people get infuriated in the process. The difference is that in the case of emails, people can easily unsubscribe with a click, tag you as a spammer, or simply hit delete any time they see an email from you. They can do a lot more damage to you in a much shorter amount of time.
So, it is essential to think of email as an engagement tool first, and revenue generator second. The primary goal of your email campaign should be engagement, which in turn drives value. It does not work the other way around.
In the coming days, I will post more of my thoughts around measuring engagement and its influencers. I will also like to hear from you – what are your thoughts around measuring engagement driven by email, and/or do you see the problem space differently? What should be the role of analytics in driving success of email campaigns?
The more I think about this, it seems that analyzing success of email campaigns is not just looking at one particular success outcome, or even looking at individual metrics like open rates or click rates -- the task seems more about developing some sort of "Email Engagement Index" (EEI -- do we really need another acronym?) that incorporates all the different manifestations of engagement that were influenced by the email (open, click, purchase, forum posts, calls to the call center, web site visit, etc.) with their appropriate "weights". Then we need to track the impact of our email campaigns with respect to this index. Else we run the danger of looking at this multi-dimensional issue in a more unary manner (such as looking at only revenue or open rate, one at a time), and thus, not realize the actual impact of the email campaign.
I'll have more on this soon, but thought I should at least get the dialog started.
5 comments:
Very true Sandeep. I have seen this weakness of just increasing volume of emails to get the target by fellow email marketers. This is actually killing the advertising medium itself.
I have seen this trend not only in smaller companies but even in large organisations like Usfis, E-bay etc too. They sent me so many mails for the action I was not interested and forced me to list their ids in the spammers list.
I would say, people should be more worried about customer needs and wants and try to understand them than just pushing their product. Google is very kind on customers in this space and have found their mails fairly relevant; but along with this they are very strict on spammers. Spamming on a gmail account will almost drive all the future mails to the junk mail folder of all the gmail users (my personal experience says this). So, people need to be extra careful.
I just think, why dont people just analyse their data before they send the mails. A good customer segmentation and a response model will make them much better in the market.
I will probably write a blog entry on this topic sometime soon. You have raised a genuine and a alarming issue here. Hope marketers read and learn something from your post.
--Bhupendra
Well, I think if "a good customer segmentation and a response model" were easy to do, people will do it. But from a typical marketer's perspective today, this still gets categorized as "complex and technical", and something that requires hiring Stat PhD's and/or integrating a heavy-duty business intelligence or analytics platform such as SAS or Business Objects. And once you realize that your analytical solution will come from a different system that the one that does your campaign or marketing program execution -- all of a sudden, red flags go up everywhere.
It is this perception that solution providers need to fight. What needs to happen is that the marketing program execution platforms need to include these analytic features as a part of their solution portfolio and make it easy for the marketers to incorporate these tasks in their campaign strategy development and implementation.
Most marketers you will talk to will agree with you that they need to do a better job in segmentation and tactical optimization. Their challenge is that they haven't found a tool/process that enables them to do what they want -- it is usually a mix of multiple solution providers - which is why they see it as fairly complex and/or resource-intensive. And worse, when it comes to email, because of the relatively cheaper cost of delivery -- it also becomes easier to ignore in the short term.
Well, I agree with the application problems with these techniques (good segmentation and response model). It is hard to do and more to actually implement in business.
Probably Marketers will have to shop for an marketing analytics vendor (for good segmentation and response model). [Good for us :)]
In any case spamming has to be stopped or reduced for their own benefit. Spamming is only a very short term solution with very high side-effects for the long term. And email marketers should understand this.
Hi Sandeep.
In the end it's all about building emotion. More emotion equals more engagement.
We can measure that emotion by recording different levels of engagement: opening rates, CTR, conversions and unsubscribes. Not to forget page visits, repeat visitors, repeat buyers and what have you. These measures allow us to express, with subjective standards, different levels of emotion and engagement. From pratically none to eternal love.
And when the real engagement does become love your revenues will follow.
Indeed, making the numbers and systems talk can be complex and sometimes frustrating. The vast load of data too.
My rule of thumb: don't go for perfection. When it comes to analytics the good old 20-80 rule or a variation thereof is often all you need to determine in which direction to look. We're not web accountants. Think trends !
Hi Remco:
Thanks for your comment.
I agree that the answer is not becoming web accountants and optimize and customize each and every contact, and that we should be focusing on identifying trends. And "trends" is where analytics should be focused on.
To me, "trends" means identifying your most relevant segments and understanding what makes them happy, and get moving towards "eternal love". This is where technologies such as OLAP and data mining can help a lot. But they have to be applied on looking at this problem holistically and not focus on one metric at a time.
I'll have a followup post soon on the our approach towards a holistic measurement of engagement, so stay tuned :-)
Sandeep
Post a Comment