TMAI #238: The OG of Analytics -- Segmentation!
| |One of my most quoted sayings:
Almost all data in aggregate is crap.
I can see the value of data in aggregate for some CEO reports.
For example, what was the overall revenue last quarter? Did our Profit per Sale decline last month? How many pre-orders do we have for our grass-fed bison meat?
Giving the aggregate answer in all these cases is ok.
But, you can imagine how quickly that answer becomes useless when the what will fall short of answering the why questions that’ll come up.
Why was our revenue above expectations last quarter? Which products contributed to the fall in profit? Did our paid search campaign drive our pre-orders?
Now you know why I’m such a huge fan of segmentation as a key part of an Analyst’s practice.
Smart segmentation is helpful in preemptively answering questions that’ll come up when your data leaves you, as well as in helping you find insights that matter.Smart segmentation increases the chances that your peer employees and leaders will take action.
What is Segmentation?
You had 96,017 Visits to your website, by 68,125 Visitors.
That's that data stated in aggregated.
Here’s that data segmented:
Instantly more useful than knowing that there were 96,017 Visits.
And, it causes an exciting rush of smart questions. Are the conversion rates the same for the top five? What about Average Order Size? And, on and on.
This is how you go from data puking to beginning the work that’ll drive action.
This is how you go from a Senior Leader saying meh to the data to truly learning because you piqued their curiosity.
Give me segmentation, or give me death!
Why is your segmentation strategy not awesome?
I bet the very first questions you’ll get, or seek to answer personally, by segmenting data are related to your Paid Media efforts. Search. Television. Facebook. Display. Radio. So on and so forth.
In part, this is because if you look at the Marketing budget such a big piece of it goes into these types of inbound efforts that it is a little understandable that they take on a big part of our mindshare.
My worry is that most Analysts just obsess about that, and little else. If I’ve got a dollar for every time I’ve heard the word attribution modeling, I would be a billionaire now.
With Analysts focusing so much on Paid Marketing efforts, building great user experiences falls by the wayside (or at least does not get as much obsession as it deserves!). Not spreading analytical love beyond Paid efforts results in a less than robust understanding of the macro outcomes (and micro outcomes!).
Let’s commit to not being that person in 2021.
To help you along, here are three clusters of segments that your data efforts should incorporate.
| |1: Acquisition Segments.
Acquisition refers to the activity you undertake to attract humans to your website/retail store/engagement points.
This would, obviously, include campaigns you run, like pay per click marketing (PPC), television, affiliate deals, display ads, radio, Facebook marketing campaigns, and more.
(Do you know the Bounce Rate, Profit per Sale for each of them?)
But, it also includes SEO.
(The moment you focus on SEO, your smart leader will ask you what’s the incrementality of our PPC campaigns over SEO. You’ll rush to read the TMAI on incrementality, and your company will be doing something brilliant!)
It also includes email marketing, inbound leads from your call center, organic social (I do hope you are not doing organic social!), and more. Owned. Earned. Paid.
Ask yourself the question: Where is my company/client currently spending the most amount of time and money acquiring traffic?
The answer has your main acquisition segments.
Your task is to figure out how to both lower the cost of acquisition (by identifying and eliminating the losers) AND increase revenue (by identifying and investing to the point of diminishing returns where things are going well).Important: Don’t stay at an aggregate level when you create segments (a good place to start, just don’t stay there). Ex: Don’t segment by Paid Search and stop.
Split into Brand and Non-Brand terms. As you know there is no such thing as a keyword, the game is about key phrases - so create mico-segments of 3 word phrases, 7 word phrases and 10+ word phrases, identify the difference in behavior. Split Paid Search traffic from Florida, New Mexico, Arizona and Utah. Split between Paid Search campaigns where you give discounts vs. those where you don’t. Do match-back with your internal systems and split New Customers vs. Existing Customers acquired by Paid Search. Wait six months, split Paid Search acquired customers by the number of tech support calls they made, the subsequent purchases they made, even product reviews they wrote.
Phew! And, I'm not kidding, I’m just getting started. :)
The deeper you go, the better you’ll understand the business (HUGE benefit) and the more precise and actionable the recommendations from insights you identify.
(This tip applies to all the segments below as well.)
| |2: Behavior Segments.
Behavior refers to the activity humans are undertaking on your website.
It is almost criminal that most Marketers don’t care about where they dump people on your site/store - because their senior leaders don’t create the right incentives.
Behavior segments begin to change that reality in a positive way.
When humans show up at your site, what is it that they are doing? Is there anything discernable/important in their behavior that is adding value to your digital existence? Or, on the flip side, what do we want people to do on our site, and is anyone exhibiting that behavior?
Such fascinating questions, seeking answers that will drive fatter conversions (!). Yet, it breaks my heart that behavior segments are so frequently overlooked.
Most behavior segments fall into two clusters: People who see x pages. People who do y things.
Visits with more than three page views (critical for content-heavy sites, important for eCommerce sites where this demonstrates an ability to cross-sell and upsell).
People who add to cart, vs. those who don’t.
People who’ve visited more than 5x this month.
People who visit only certain parts of the site (sports, editorial, international edition, crosswords, and games, etc.).
(Obviously…) Everyone who bounced.
Oh, and people who abandon during the checkout process.
How can you forget people who do and don’t use the internal site search (and keywords they use - intent, baby!).
All that is off the top of my mind. I have 40 more of these, I’ve done so much behavior analysis (because I am OBSESSED with improving user experiences and removing roadblocks).
People whose days to transaction is more than 5 and less than 5.
The more you understand what people are doing on your site, the more likely it is that you’ll stop the silliness on your site (kill content, redo navigation, make cross-sells better, eliminate 80% of the ads - and still make more money! - learn to live with 19 days to conversion, don’t sell too hard, and so much more).
My guilty pleasure from effective behavioral segmentation is proving how wrong Senior Leaders (HiPPOs!) are in what they think is happening on the site vs. what actually is. Or, even sweeter, when it is so obvious that their favorite UX features - here’s what I want you to do, do thing x, that is what works for me! - actively harm outcomes.
While that is temporary joy, what feels really great, every day, is delighting customers by improving their experience.
I don’t need to tell you… Making customers happy is highly causal to making loads of money. :)
Behavioral segmentation is hard. That is why there is so much profit involved when you do it right.
| |3: Outcome Segments.
Outcomes are site activities that add value to your business/non-profit.
In my humble experience, the problem with outcome segmentation is not that Analysis Ninjas don’t segment, rather it is that they are incredibly unimaginative.
So yes, segment people who convert vs. those who don’t. Figure out the patterns of behavior that lead to the first outcome and patterns of acquisition and behavior that lead to the second (sad) outcome.
But, please don’t stop there.
Segment people whose order size is 50% bigger than the average - where did these whales come from, what did they do after they got to your site? Can you do more of either?
Segment, omg I love this so much, macro AND micro outcomes. Just about the only way to solve for medium-term and long-term value for your company is to understand exactly what micro-outcomes your company is solving for.
(Note: That last link is from March 2008. 2008!)
People who add a product to a wish list. Those that sign up to show up for a protest by your non-profit. Those that apply for a trial or download a pdf. Those who give you their email address. Those who send a sales inquiry. Those who open an account. Those who submit a job application. And, on and on and on… Outcomes that deliver economic value to you.
If you've read my (did you know bestselling?) books, you know my #1 most favorite metric of all time: Task Completion Rate. If you’ve integrated that into Google Analytics, segment by that and wait for your mind to be blown by the insights you’ll accumulate.
Net, net. . . it is absolutely critical that you segment your data by the key outcomes important to your business. Not just because your site exists to add economic value, but also because I cannot think of another way you can earn the love of your boss or get promoted.
What I’ve primarily covered in the digital analysis view in my above recommendations, you can see how segmentation in these three clusters applies to every type of business out there and every department in your company.
| |Super Special Bonus Tip.
My favorite type of segmentation? Segmenting the data by company VPs.
Step 1: Identify the five key metrics to report to our CMO indicating the essence of quarterly business accomplishments. (Literally the answer to her question: “Avinash, what did I get for all the money I spent this quarter?”)
Step 2: Take everyone who reports to our CMO, segment the performance for those five key metrics by each VP.
Step 3: Send the CMO dashboard out (and brace for impact!). :)It is not easy to do this.
Technically segmenting key metrics by company leaders is a tricky challenge, but one that can be solved with some effort.
Psychologically, the challenge is no VP wants to see their name against poor performance (and like at any company, there is always poor performance, in case of some folks consistently).
When at Step 3, I guarantee that every single VP will lean the heck into the data. No one wants to look bad. Everyone wants to be praised when they are exceptional. They will look at the data. They will ask questions. They will push their departments to change and do better. They will deliver better performance.
All because you segmented the data by them.
No human ever wants to see their actual real name against a row of key metrics.
(If you use the name of the department they lead instead of their name, it has 1/10th the impact. Such is the power of their name in the limelight.)
So. Segment by VP names. Get magical results.
| |Bottom line.
Simple: Segment, or be irrelevant.
The next time you start to do true analysis of your data, I hope you have your minimum of six segments in hand (two for each category). If you do, you'll find that analysis suddenly becomes a lot more interesting for you, profitable for your company, and the resulting improvements delightful for your current and future customers.
PS: For Premium subscribers most weeks I share a link of the week and an image of the week.
Link of the Week:
Frederic Filloux has written a super-insightful analysis of Google’s decision to give about €150m to the French press over the next three years. As he says, the details show a mixture of a genuine and impactful arrangement and the usual convoluted dealings to make a new set of subsidies looking like a sound business deal.
It is fascinating: Inside Google’s Deal with the French Media.
Image of the Week:
It is so corny, but it really made me smile… And, made me think…
I don’t know where the picture is from, but from the architecture I could almost swear it is downtown San Francisco.
Thank you for being a TMAI Premium subscriber - and helping raise money for charity.
Your Premium subscription covers one person. It’s fine to forward occasionally. Please do not mass forward or pipe it into Slack for the whole company. We have group plans, just email me.