Email Segmentation’s Enemy Is Math
"Here’s the problem I have with segmentation…most of it – for email marketers – has yet to show that it actually pays in either increased sales or increased customer engagement. People seem to feel that segmentation is good, yet have little proof that it makes a difference. As an email marketer…I like proof."
As a life-long practitioner of segmentation (I actually know what the Z in PRIZM stands for…and know what ACORN is…), it’s with a certain degree of bemusement that I see all the proclamations that “segmentation is the KEY to email success” or “you can never have too many segments” or “a three hundred sixty degree view of the customer (ewww)” or anything along the lines of “if you’re not segmenting, you’re not doing it right.”
Here’s the problem I have with segmentation…most of it – for email marketers – has yet to show that it actually pays in either increased sales or increased customer engagement. People seem to feel that segmentation is good, yet have little proof that it makes a difference. As an email marketer…I like proof.
In thinking about segmentation, it’s important to understand the soil from which almost all segmentation grows – expense. Almost all traditional media elements are expensive – especially things like direct mail and/or its segmentation-centric cousin, catalogs. The cost for a badly targeted piece of mail was high – ranging from a low of .$40 per piece to a high of…well, whatever expensive catalog you can think of.
You could make really a fancy decile analysis based upon the cost of sending all that bad mail. Many times, you could cut out the bottom 40-60% of your mailings and still not suffer a profit drop.
In applying segmentation-based thinking to email, a significant problem emerges – email is CHEAP to deploy yet relatively EXPENSIVE (in comparison) to create. Even if you’re paying a ridiculous CPM of $10, you can still be wrong roughly 500 times in email for every one wrong direct mail/catalog error. The low cost of email also blows a rather significant hole in your decile analysis – you can certainly be more efficient but at the cost of total revenue.
For Example, take a look at this direct mail chart –
In this case – even using a super low cost per piece of 50 cents, savings from segmentation are significant – enough to pay for the model and any changes needed to drive more purchases.
Now let’s take a look at that same chart, except the names are delivered via email.
Your $119,000 saving drops to – are you ready - $80. That’s eighty, as in “less than Loren usually withdraw from an ATM.”
Now, it may not be fair to use the same conversion and revenue metrics. So let’s change the game by reducing the email conversion by 50% and the revenue per order by 20 %. I bet we’ll now REALLY need segmentation!
The “savings” from segmentation? A whopping $500.
Think about that for a second – by applying segmentation to your email, you save $500 per campaign. Which means a lot if you’re sending, say, 10,000 campaigns per year but at that rate of sending you sure better not be paying $10 CPM…if so, get a new ESP!
Here’s the kicker – what’s the cost of developing a good segmentation model? Let’s be cheap and say $10,000. So you need to run 20 campaigns to pay for the cost of the model.
How much to develop new and exciting creative versions to fit into the segmentation array? Let’s say $10,000. So now we’re at a total of 40 campaigns before we see any profit from our efforts. If you run 300 campaigns per year, you’re looking at a segmentation savings of $130,000…an amount that can be better saved by renegotiating your ESP deal.
The knee jerk response many times is “if I’m not segmenting then I must be a SPAMMER!” First of all, get over yourself. The only time an email marketer is not a spammer is when (a) there’s NO email program at all or (b) when the CFO is looking at the revenue you generate. Remember – for most legitimate marketers, these names did not appear on your list by Harry Potter waving his wand. These people signed up for your email which means only one thing – they’ve already segmented themselves! The most important segmentation variable you can find is interest in your product/company. If you’re doing a reasonably transparent add to your file, they are there for one reason – because they CHOSE to be.
The other response often is “but Bob…they shopped for red sox…so let’s give them more and more red sox.” While that seems logical you’ve effectively eliminated discovery from your email quiver…but that’s a subject for another day.
You can go down the path of segmentation all you want…but try the above exercise to make sure the numbers work for you before doing so. Take advantage of the low cost of distribution of your email platform by mailing deeper –and more often - into your file. Unless you like Valhella!
Bob, you are of course correct. As one's costs go down, the less segmentation really matters...until attrition becomes the metric one is managing against. But even here, a less than enthusiastic full throated support can be mounted.
The logic works if a marketer looks at a single campaign/deployment. What is lost in this analysis is the cost of an opt out which go up dramatically as relevance and interest decline. There is also a soft cost of disengagement which impacts future response and conversion metrics.
We are talking about “dividing your contacts into groups to be able to market better”. Well sometimes people are doing the one thing (the dividing) but not enough of the other (marketing better). In that case you would be right, it wouldn't pay off.
But here is a thought: Not all forms of segmentation are created equal, your costs and incremental income heavily depend on your tactics and execution. If you still think that costs too much? Then go for better audience match tactics: