Testing & Analysis INSIGHTS
INSIGHT #31 - Measuring Across Time
Our last Insight was about the philosophy of triggered lifecycle programs.
And now I've promised you a "How To" session on measuring the incrementality of lifecycle marketing.
The philosophy in a nutshell: Lifecycle triggered campaigns are about influencing lifetime value.
So even though triggered lifecycle marketing campaigns can get upwards of double the immediate direct response for open, click and conversion vs. a blast marketing campaign....
...don't stop at simple campaign response
Because that will cut short the real value of a triggered lifecycle program.
Triggered lifecycle marketing programs generate LIFETIME value in a way that blast marketing campaigns do not.
So how do you measure the incremental lifetime impact of a triggered program? Well, you measure it over the lifetime of the customer. Oops, that's too long! But the point is, you must measure it over a long period of time, maybe 6 months or a year.
WHY so long? Because the value of a triggered lifecycle program shows itself more and more over time. You'll start with some set of lifecycle triggers, but as you measure the incremental lifetime value, you'll continue to develop more and more of these triggers along with corresponding targeting, creative and offers.
Here are 5 basic "how-to" steps:
- Set up an initial lifecycle segmentation
and contact strategy.
That's for another day. But for now...
Just define some lifecycle segments.
Base it on purchase recency, first purchase, birthday, category of purchase, channel of purchase... Pick some lifecycle segments.
Whatever your segments, customer lifecycle is NOT about what your brand is planning.
Customer lifecycle is about what the customer did or did not do as of a certain poitn in time.
- Split your segments into Test and Control
Start with 50/50 and you won’t go wrong.
Don’t worry, some of these will come out of the Control and into the mail stream shortly.
You can always reduce the Control Group after your analysis shows statistical significance; but you can never increase it (because you will have already tainted the comparison with that first lifecycle mailing).
- Suppress the Control Groups from any lifecycle
For a period of X months, your lifecycle Control Groups will receive whatever they would have received, but absolutely NO lifecycle campaigns. The Control Groups continue to be treated as normal for everything else, just no lifecycle campaigns.
- Implement lifecycle marketing for the Test
What's that look like?
Your lifecycle Test Groups will receive every lifecycle campaign they qualify for.
For emails, the lifecycle campaign should pre-empt whatever else they would have received that day.
For direct mail, the lifecycle campaign can be a version of what they would have received or it can be in place of what they would have received.
- Measure the growing net spending disparity
between the Test and Control Groups ACROSS TIME.
At the end of no less than 6 months (and at the end of every month until then), you’ll analyze the multi-channel net spending of the Control Groups and compare it to the multi-channel spending of the Test Groups.
This long-term analysis has nothing to do with campaign response or direct response. You're not filtering on promo codes or who got any specific campaigns - you're just comparing the Test and Control Groups with How frequently did they buy and much did they spend? Over a long period of time.
And ... over time ... you will see a growing
disparity in spending between the Test and Control Groups.
The longer you hold out the Control, the greater will be the disparity. And THAT growing disparity between Test and Control is the incremental value of running a lifecycle trigger program.
At some point, you'll find sufficient statistical significance to go on and move some portion of your Control Groups into the lifecycle mail stream, reducing the size of your Control Groups.
And eventually, you'll decide that enough is enough, and you'll end the analysis and move all the Control into the lifecycle mail stream.
You'll likely only ever do this analysis once. Use the results as a benchmark for the scale of annual incremental value from triggered lifecycle programs. You'll gain insight into the ongoing impact of this type marketing vs one-off blast campaigns. You'll muster the necessary buy-in to create more and more triggers, campaigns, and offers. And forever more, your marketing will grow more and more relevant and it will drive more and more loyalty and retention.
INSIGHT#30 - Tracking Across Time
If you’re only tracking ROI during the campaign response period, you’re missing it.
Especially for any triggered, lifecycle-driven contact strategy.
Think about the point of lifecycle marketing. “Lifecycle” implies two things:
1) Something today
2) That affects the future
Think about raising children. If it were all about how they behave today, we’d just bribe or cajole. Instead, we discipline to raise our children into law-abiding adults.
Think about dieting. If it were all about today, you’d throw caution to the wind and eat <fill in your blank>. Instead, we diet in February to fit into swimsuits this summer.
Think about exercise. Don’t know about you, but I hurt after exercise. For me, it’s all about a long and healthy future.
Now think about marketing. Is it all about making today’s plan?
Well let me have a hey ya'll and yee-haw!
But wait! That really is NOT what triggered, lifecycle campaigns are about.
“Lifecycle” implies two things:
1) Something today >>> bellwether event coupled with a bellwether contact
2) That affects the future >>> long-term impact
I love Wikipedia’s definition of bellwether: any entity in a given arena that serves to create or influence trends or to presage future happenings.
Lifecycle triggered campaigns are about influencing trends, future happenings.
So If you’re only tracking ROI during the campaign response period, you’re missing it.
Next up: how to track impact across time.
INSIGHT #29 - Needing Time
I haven’t sent you one of these little “Insights” since Dec 3. Between the holiday marketing frenzy and a couple of big projects, I was busy.
Also, my mother passed away, and I just needed time.
Needing time. Let’s divert from the mini-series you’ve probably forgotten about anyway. And let’s talk about NEEDING TIME.
Here’s the thing: If you’re only tracking ROI for your marketing campaigns during the campaign response period or through a promotion code, you’re missing it.
Everyone needs time. Your customers need time. And they don’t always respond on YOUR timeline.
If near-term ROI is all you’re measuring, you’re selling your marketing impact way short.
Your customers remember. Not every marketing campaign. But the direct mail piece you sent last summer may be the very reason they think to buy from you this January. Your continued, relevant messaging over time might be what keeps them engaged.
It wasn’t just what my mother said to me last month, but the impression she made my whole life, that impacts my thoughts about her right now. And it’s not that different with customers and brands.
Are you measuring the long-term impact of your marketing?
Next up, measuring long-term ROI. Meanwhile, check out Insight Outcomes #12 below; it's about measuring the long-term impact of a frequency test - over time.
INSIGHT #12 - Frequency Test
Last week, we looked at what REALLY matters in a frequency test – CUSTOMER metrics.
But this week I promised to look at CAMPAIGN metrics within the context of an email frequency test.
What’s the easiest way to set up the segments, the frequency cap codes and the reporting template?
ONE. Set up the test and control groups for various customer segments.
- You'll want to optimize frequency for each segment so do some sort of segmentation, even if it’s just Active vs. Inactive
- Leave out anyone who already has a frequency cap; include EVERYONE who doesn't
TWO. During the frequency test, add NEW SUBSCRIBERS to similar but separate frequency cap codes.
- The biggest gap between the test and control will be amongst the new subscribers – so don’t leave them out of the test
- Hint: How you onboard your subscribers, including frequency, will make a big difference in their future performance
- But eventually, you’re likely to prove that it pays to start low, but ramp up to a higher frequency
THREE. Assign frequency codes that are distinct from any of your other frequency codes
- Every existing subscriber will have a frequency code assigned
- For example, let’s say you typically send 7 emails per week, if you’re testing 2, 4 and 6 times per week, depending on the size of your segments, your codes might look something like this:
Size of group
2 times a week
4 times a week
6 times a week
7 times a week
4% of total Active
2 times a week
4 times a week
6 times a week
7 times a week
4% of total Inactive
2 times a week
4 times a week
6 times a week
7 times a week
All new subs during week 1
All new subs during week 2
All new subs during week 3
All new subs during week 4ff
1 time a week
These are not in the test;
they were already capped
FOUR. Set up your email reporting tool to track these distinct frequency cap codes separately.
- Since you assigned EVERYONE a frequency code, everyone will show up in the results
- Campaign results are additive over time. Your email tool never attributes any response to more than one campaign, so you can use simple addition to sum up the volume metrics across time
- If you’re multi-channel, you MUST take offline orders into account – if not, your frequency test will lead you astray, but if systems make that difficult you might plan to do the multi-channel analysis at the end of multi-month test, while constantly keeping your eye on the immediately available online results
FIVE. Set up a Frequency Test reporting template.
The Reporting Template needs to do the following:
- Let you easily paste the results of each frequency cap group for each campaign
- Sum up the volume metrics (total sends, opens, clicks, orders,
across time for each frequency cap group
- Calculate efficiency metrics (open rate, click rate, conversion,
using the summed volume metrics across time
- Display apples-to-apples Total Volume Metrics as if each of the test & control groups were of equal size
- Display apples-to-apples Net Efficiency Metrics based on the ORIGINAL quantity in each group (to account for the impact of frequency on unsubs)
In the end you want to show which frequency drives the most profits from each customer segment over time, so your test must run its course over several months time; the longer the better.
Next week, let’s move on to something less technical.
Something fun from the “Profitable Programs” bag!
INSIGHT #11 - Campaigns vs Customers
Typically, an increase in send frequency will result in an increase in clicks and revenues. At the same time, it will drive a decline in campaign response rates.
Don't be distracted by campaign response rates.
What really matters is customer response rates.
Do you see the difference?
Revenues per Send (that's a campaign response rate)
Revenues per Customer (that's a customer response rate)
As sends per customer go up, so do revenues per customer.
Yet you'll see a decline in revenues per send. So what!
What matters is:
overall revenues are increasing, and revenues per customer is increasing.
Focus less on how your CAMPAIGNS perform and more on how your CUSTOMERS perform.
If you want to improve a campaign, do an A/B test, and that's where campaign response rates matter.
If you want to grow your business, focus on period-over-period trend in customer responses per week.
If your customer list is growing (or stable)...
... and your customer responses per week is growing,
... then don't fret soft or declining campaign response rates.
Insight #10 – Which Is It?
Retail is not like sports where a single score determines the winner.
Depending on your role, you're accountable to multiple "scorecards." And some scores may be in direct conflict with others!
So WHICH IS IT for you? Volume or Efficiency (ROI)? To optimize against both, establish a minimum acceptable ROI and go for as much volume as that permits.
Even then, the list of ROI metrics and Volume metrics goes on and on. So you must prioritize the various scorecards that are inherently present in any test.
For example, here are some volume metrics:
- Top line revenues
- Bottom line profits
- Customer List Growth
- Customer Satisfaction
- Customer Retention
- Customer Acquisition
- Customer Value
- And on and on...
And no fewer Efficiency scores:
- Revenues or Profits per Campaign
- Revenues or Profits per Send
- Revenues or Profits per Customer
- Revenues or Profits per Order
- Visits per Campaign
- Visits per Send
- Visits per Open
- Visits per Customer
- Acquisitions per Campaign
- Acquisitions per Send
- Clicks per Open
- INCREMENTAL (all of the above)
- Cost per Customer
- Cost per Acquisition
- Cost per Order
- Cost per Revenues or Profits
It comes down to this:
What are you aiming to increase? And how much decrease in efficiency is acceptable in exchange for that increase in volume?
By the way, these scores exist, whether you track them or not!
Next week, we'll look at a specific example of testing, analyzing and optimizing various volume and efficiency scorecards.
Blog #7 – Scraping the Bottom
In Blog #4, we talked about your List Growth and List Engagement reports fueling your segmentation and analysis strategy.
Today, let’s consider your LOWEST-PERFORMING SEGMENT. Chances are, that's half your list! Non-buyers (or so they seem) and/or non-engaged. As in, they’ve been around for more than a year and you haven’t detected a pulse in all that time.
Don’t just cut them off. Scrape them off the floor. If you’re spending anything on prospecting, THIS is the top priority group you should be prospecting! They’re going to give you better marketing ROI than going after prospects. Guaranteed!
And even just a tiny lift will pay off big for such a big segment.
Blog #4 - Right-Brain Pre-Analysis
With your List Growth and List Engagement KPIs in place (Blog #3), now it's time for some right-brain pre-analysis.
Recall, we started with any readily accessible segmentation scheme for our List Growth and List Engagement KPIs – gender, source of collection, region, channel, etc. If we only have it for the TOTAL list – or if we don’t have weekly or monthly trend over time – then we’re all dressed up with nowhere to go.
First step > identify the biggest segment of your list. Is the segment growing or shrinking? Is engagement increasing or decreasing? How do metrics for this segment compare to metrics for the other segments?
Any little thing you do to improve the biggest segment will pay off big.
Here’s a short list of possible factors that are likely in your control if you’re reading this:
- You’ve hidden your sign-up form
- You’re sacrificing sign-up volume in favor of a form that is too complicated
- Your sign-up form or data feed has a glitch
Here’s a short list of factors that are likely beyond your control but any Marketing workaround can bide time:
- New competitors
- Merchandise issues (inventory, fit, pricing)
- Fulfillment issues (high shipping fees, slow delivery, poor store experience)
- Changing society or demographics
Next step > Identify a smaller segment of your list that is already doing well. It’s growing and engagement metrics are strong and improving. Right brain: how can you make this segment a bigger piece of your list or customer base? If you can turn this segment into critical volume, that’s a win all the way around.
Some ways to grow a small segment:
- Targeted banner ads, keywords, promotions, sweepstakes, and co-reg partnerships on other websites
- Target some Refer-a-Friend campaigns and incentives to this segment
- Drive them to specially designed sign-up forms; incent the new sign-ups with a short-life promo code; get them marketable, and thank them with a drive to purchase conversion
- Test some in-store changes targeted at this segment
These explorations will start to fuel your segmentation and analysis strategy.