The Action Stream Framework

Here's the framework, summarized within one image (click on it below to download to your device).

Mini-Assortments in Action Streams

You probably already do this (if you aren't doing this, wow) ... but you already know which items sell best in each attributed marketing channel ... right?

In other words, you have an assortment of Dresses ... pretend there are 100 styles you offer. You already know that The Positano (click here) sells best in Product Listing Ads, or that the Bellamira sells best in Email Marketing Campaigns.

You also know that a lapsed eighteen month customer previously purchased via a Product Listing Ad. If that customer is about to meet one of two criteria ...

• +  The customer just visited your website.
• -   The customer is about to lapse out of the "Lapsed Spend" segment to "Lapsed Experiment"

... then you should launch an Action Stream to encourage the customer to buy something right now. And if you're going to do that, you know that for this customer, you should show Winning Items that align with Product Listing Ads ... preferably in the product category the customer prefers.

Right?

What stops you from taking action in an example as illustrated above?

Use Mini-Assortments ... the items that sell well within key Marketing Channels ... to kick out Action Streams to encourage a lapsed buyer (or even a recent buyer) to purchase again.

Switching Up Creative

A large e-commerce brand sent me forty (40) consecutive email marketing messages featuring a percentage off message above the fold.

Interestingly enough, this brand sells stuff I've previously purchased. Why in the name of Chris Donald couldn't this brand be remotely bothered to feature what I purchase or might be inspired to purchase?

If you employ Action Streams, the logic isn't difficult.

1. Does the customer pay, on average, at or above-average prices for individual items? In other words, if an item normally sells for $50 and sometimes sells for $30, does the customer skew toward $50 or toward $30?
2. If the customer skews toward $50, why not feature a half-dozen items that align with the interests of the customer ... featuring these items above-the-fold, featuring the ridiculous discount elsewhere?

In other words, send me (Kevin) down an Action Stream that features "what" I am interested in buying.

Conversely, for Calvin (let's assume Calvin craves discounts), send him down the typical Action Stream with 40% off or 60% off or 80% off of clearance or whatever ridiculous discount gets Calvin to pay $10 for a $30 item with a $7 cost of goods.

Switch up creative based on the interests/needs of the customer.

How Accurate is ChatGPT in Describing Action Streams?

When I asked ChatGPT to give a specific example of implementation of an Action Stream, here's the answer I received.

Do you think the answer is credible?

Example of Implementing Action Streams in Marketing

Let’s say an online fashion retailer wants to improve customer engagement and increase conversions using Kevin Hillstrom’s Action Streams model.

Step 1: Identify Key Customer Actions

The retailer tracks customer behaviors in real-time, including:

• Browsing Activity – Customers who repeatedly view certain products.
• Cart Abandonment – Customers who add items to their cart but don’t complete checkout.
• Purchase Recency – Customers who bought something recently vs. those who haven't in a while.
• Email & Ad Engagement – Customers who open marketing emails or click on ads.

Step 2: Assign Customers to Action Streams

Instead of placing customers in broad segments (e.g., "high-value customers"), they are placed in action-based groups:

1. "Window Shoppers" – Customers who browse but don’t add items to their cart.
2. "Cart Abandoners" – Customers who add items to their cart but don’t purchase.
3. "Loyal Repeat Buyers" – Customers who make frequent purchases.
4. "Reactivatable Customers" – Customers who haven’t shopped in a long time but just engaged with an email.

Step 3: Take Real-Time Action Based on Behavior

• For "Window Shoppers": Send a personalized email showing items they browsed + a limited-time discount.
• For "Cart Abandoners": Trigger a retargeting ad or send an email with a reminder and an incentive (e.g., "Your cart is waiting – checkout now and get free shipping!").
• For "Loyal Repeat Buyers": Offer an early-access sale or exclusive VIP discount to encourage another purchase.
• For "Reactivatable Customers": Send a re-engagement email with "We've missed you!" messaging and a curated selection of new arrivals.

Step 4: Optimize & Iterate

By monitoring how customers respond to these actions, the retailer continuously refines the approach, ensuring the right message is sent at the right time based on real behaviors, not static segments.

Why Action Streams Work

• They react to live customer behavior, increasing relevance.
• They prioritize high-value interactions over generic mass messaging.
• They create a more personalized customer experience, boosting conversions.

Action Stream Attributes (ASA)

It's common to have a simple table in your database to help execute Action Streams. You probably already have something similar to this ... your use of the table likely varies.

Here's some of the attributes I routinely provide for clients who ask for Action Stream Attributes (ASA).

• Email Address
• Household_ID
• Kevin's Action Segment (Elite, Loyal, Quality, Average, Struggling, Lapsed Spend, Lapsed Experiment, Lapsed Save $).
• Kevin's Action Segment Score (Probability of Buying in the Next Year).
• Kevin's +/- Indicator (+ = Customer Just Visited Website or Engaged in a Trackable Manner) ... - = Customer Is About To Fall To A Lower Action Segment).
• Months Since Last Purchase.
• Number of Purchases, Past Five Years.
• Average Order Value, Past Five Years.
• Total Demand/Sales Spent, Past Five Years.
• Primary Marketing Channel Purchased From (Email, Paid Social, Paid Search, Affiliates etc).
• Secondary Marketing Channel Purchased From.
• Primary Merchandise Department Purchased From (Mens Tops, Mens Bottoms, Mens Outerwear, Womens Tops, Womens Bottoms, Womens Outerwear, Home, Kids, Accessories etc).
• Secondary Merchandise Department Purchased From.
• Price Point Preference (High, Medium, Low).
• Discount/Promo Buyer (% of spend below average historical price point for an item).
• New Item Preference (% of spend on items new in the past year).

There are plenty of other attributes that end up being client-specific. But what you see above is more than enough to launch a credible Action Stream program.

Happy Hour!

I mentioned in December that I was thinking of hosting a Happy Hour where you could ask me any questions you might have as well as having general discussions with the attendees. So let's try this ... I don't care if four people sign up or forty.

Happy Hour

• Thursday, February 13.
• 8:00pm EST / 5:00pm PST.
• 55 Minutes.

If you want to attend, send me an email (kevinh@minethatdata.com) and I'll forward you a link for Happy Hour.

Bring your favorite beverage - whatever that is, water is fine, something more interesting is fine as well.

I'll work on a topic to get us started.

Who's in?

+/-

In the NBA, players are occasionally evaluated on a "+/-" metric. If you come into the game with the score 42-34 and you leave the game with the score 54-44, your +/- is +2 (you were ahead by eight when you came in and you left leading by ten).

In Action Streams, +/- represents a change in customer state.

• A "+" is when the customer does something positive, like visiting your website or interacting with your community or watching one of your videos on YouTube or clicking through an email campaign.

A positive action requires a new Action Stream. If the customer has not purchased in eighteen (18) months and suddenly appears on your website, you suspend your normal email cadence with a new Action Stream. You already do this when the customer abandons a shopping cart. Why not extend the concept when the customer visits your website and looks at low-priced Home merchandise (for instance)?

• Contact 1 = Show customer what the customer looked at.
• Contact 2 = Show customer related products.
• Contact 3 = Your best discount/promotion (since your lust for discounts/promotions requires action).
• Contact 4 = Show customer content options to keep the customer engaged.

In Action Streams, a "-" represents a customer who is about to lapse to a lower / less responsive customer segment.

Remember the segmentation strategy I use to classify customers?

• Elite:  75%+ Annual Rebuy Rate + 12 Month Buyer.
• Loyal:  60% - 74% Annual Rebuy Rate + 12 Month Buyer.
• Quality:  40% - 59% Annual Rebuy Rate + 12 Month Buyer.
• Average:  20% - 39% Annual Rebuy Rate + 12 Month Buyer.
• Struggling:  0% - 19% Annual Rebuy Rate + 12 Month Buyer.
• Lapsed Spend:  12%+ Annual Rebuy Rate (worth spending money on).
• Lapsed Experiment:  5% - 11% Annual Rebuy Rate (worth testing ideas on).
• Lapsed Save:  0% - 4% Annual Rebuy Rate (don't spend money, they are unresponsive).

Any customer with a "-" is a customer in the lowest percentage point of the band. The customer is about to drop down to a lower segment.

• Example:  A Loyal customer drops to a 61% annual rebuy rate. This customer is graded with a "-". This is your signal to DO SOMETHING to prevent the customer from falling from Loyal to Quality status next month.

In your customer database, you flag any customer with a 76% / 61% / 41% / 21% / Lapsed 13% / Lapsed 6% rate ... you flag them as "-", and then you DO SOMETHING to stop the customer from sliding down to a lower customer segment. You set up an Action Stream to mitigate the demise of the customer.

Does that make sense?

Great Moments in Omnichannel History

Traditional retail is in a lonely, unsatisfying alternate timeline.

Action Streams: Compatibility Scores

It's 2001 at Nordstrom.

The e-commerce Executive wants to tailor email campaigns to the merchandise customers previously purchased. Her team tells my team they can support ten (10) different versions of an email campaign twice per week. My team is brilliant. They tell her they can develop a methodology to pair customer purchase history with the content offered in a version of an email campaign.

What they came up with is similar to what I currently call a "Compatibility Score".

Let's oversimplify the process. Say you have two versions of an email campaign, and you want to decide which version the customer should receive. Now, there are better ways to do this, and if you don't like how I oversimplify my example below, by all means, add the level of complexity necessary to suit your needs. But for the love of Nordstrom, do something, ok?!

Ok, here is the merchandise composition of email version #1.

• Mens = 70%.
• Womens = 20%.
• Kids = 10%.
• Home = 0%.
• Accessories = 0%.

Here is the merchandise composition of email version #2.

• Mens = 0%.
  
• Womens = 50%.
• Kids = 30%.
• Home = 20%.
• Accessories = 0%.

Finally, we have a customer with the following weighted purchase history (I weight historical transactions ... often along a 100% for 0-12 month orders, 60% for 13-24 month orders, 35% for 25-36 month orders, 20% for 37-48 month orders, 12% for 49-60 month orders ... your mileage will vary).

• Mens = 15%.
• Womens = 40%.
• Kids = 10%.
• Home = 12%.
• Accessories = 23%.

Which version of the email campaign should the customer receive? I multiply weighted historical percentages by product percentages in each campaign, then sum the multiplied percentages to acquire a "Compatibility Score".

The second version of the email campaign has a 25.4% Compatibility Score, compared to 19.5% in the first version. You'd assign an Action Stream for this customer where the customer receives Version 2. 

Yes, you should test this concept to see if it works for your customers.

As you can see ... it's terribly easy to create a Compatibility Score.

Twenty-four years ago at Nordstrom ... an entire generation ago, we obtained sales increases of 20% within email marketing using Compatibility Scores.

Are you using Compatibility Scores within your Action Streams?

Action Streams and Returns

In case you were wondering, no, "AI" is not going to solve the retail "returns" problem.

It can make your returns problem better, but come on. Most "AI" you see in my industry is Snake Oil, a rebranding of an old problem and an old solution, allowing the salesperson to charge a premium.

Why do I say this?

In 1993 (yeah ... 1993) I built a model using the hyperbolic tangent function (look it up) to predict the return rate of every one of four million twelve-month buyers. Today the snake oil folks would call the hyperbolic tangent function "AI". Anyway, we learned that returns behavior was not predictable until a third purchase. At that point, any customer who returned 60%ish of items (i.e. bought 9 items, returned 6 on those first three purchases) was a customer likely to return in the future, costing us $$$$. Our decision? Cut back on mailings to those customers ... from 20-25 per year to 4-6 per year. Boom. Another million dollars of profit.

In the modern world, some of you have Action Streams for customers who return too much merchandise. There is no reason you have to email customers who return 60%+ of what they purchase. You create an Action Stream once the customer passes a threshold, and you send the customer maybe 4 email campaigns a month instead of 4 per day.

You are under no obligation to keep emailing customers who return most of your merchandise. The customer can still buy from you. But you don't have to waste resources, do you?

Set up an Action Stream to bypass customers who return too much merchandise. You'll also bypass all the snake oil salespeople pushing "AI" solutions upon you.

It's Time!!

Four months go by in the snap of a finger!

Remember just two months ago when you were told you HAD to have a PERECT Cyber Monday experience with 70% off or you'd disappoint today's savvy shopper? Well, how did all that work out for you?

In this run of the Elite Program ($1,000 for existing clients, $1,800 for new clients, no commitment moving forward) I'll look at how many new customers you generated from Cyber Monday and I'll see if they purchased since vs. other customers acquired just before / just after Cyber Monday. That and my usual Rolling Twelve Month Analysis / Comp New Buyer / Comp Segment / Annual Repurchase Metrics / Next Year's Forecast.

It's virtually free. Here's the criteria.

• Pre-pay the fee on or before February 15.
• 5 Years of Customer Data Through January 31.
• Data due by 2/15.
• Analysis completed by 2/28.

Contact me now (kevinh@minethatdata.com) and we'll get busy!

What Would An Action Stream Look Like?

The results were absolutely compelling.

Let's say the customer last purchased eight (8) months ago, and just visited your website. What should you do?

• If you are obsessed with discounts, it's time to kick those into high gear. Instead of 25% off, make it 35% off. You'll love it, your customer will like it.
• Show the customer new arrivals for the product category that the customer previously purchased from.
• Ask the customer if the customer needs to replenish what was previously purchased.
• Promote the items that the customer looked at last month.
• Promote complementary items to the items the customer just looked at.
• Promote complementary items to what the customer purchased previously.
• Ask your customer to engage with more of your content.
• In retail, tell the customer of upcoming in-store events the customer may want to participate in.

When I log into the customer forums on Headphones.com, I am sent into an Action Stream based on the content I looked at.

This stuff works.

And there are dozens of vendors who will help you do it.

So what stops you from doing this stuff?

It Happens With Best Customers As Well

Here's our table for customers with 5+ purchases, segmented by recency and a website visit in the past month.

The "Index" column is of interest here ... we were seeing gains of 2x-5x among lower-quality customers ... the index here is much lower, though any one of us would love to market to a customer 50% more likely to buy than a comparable customer, right?

I mean, look at 13-15 months of recency.

• No Website Visit = 9.5% chance of buying next month.
• Yes Website Visit = 18.7% chance of buying next month.

Come on!

This stuff is so easy ... we're shooting fish in a barrel here. Kick off an Action Stream for the website visitor, with unique marketing activities tailored to that customer BECAUSE the customer just signaled to you that the customer is interested in you ... the customer took action, now it is your turn to take action!

• P.S.:  Taking action does NOT mean sending your same email campaigns that you send to everybody else. You have to do something different, for this specific customer.

Seasonality is Amplified

Back to our table from yesterday, reviewing customers with two life-to-date purchases, measuring how they perform in the next month based on their recency and based on whether the customer visited your website in the past month.

Did you do you homework assignment from yesterday?

Look at recency = 12 months. If the customer last purchased 12 months ago, the customer is more responsive than at recency = 2/3/4/5/6/7/8/9/10/11 months ago. That's some serious seasonality going on.

Now look at what happens if the customer is at recency = 12 months AND the customer visited your website last month? That customer ... has a 16.0% chance of buying next month. It's the customer most likely to purchase in the entire table.

Is your old-school RFM segmentation system able to see that happening?

The job of a marketer is to get seasonal buyers to the website JUST BEFORE the season of interest is about to happen. And if the customer actually cares enough to visit, you kick off an Action Stream of unique marketing activities tailored to that individual customer.

Do Better Customers Behave Similarly?

Yes!

Which means better customers also deserve Action Streams that are uniquely tailored to them.

Here's an example of customers with two life-to-date purchases, segmented by recency and visit / no visit in the past month, measuring rebuy rates in the next month.

If you last purchased 19-24 months ago AND visited the website in the past month, you are more likely (8.8%) to buy in the next month than a customer who bought 1 month ago and did not visit your website (7.8%).

So yes, better customers behave similarly. 

If you last purchased 25-30 months ago (for instance) and visited the website in the past month, you darn well better craft an Action Stream of unique marketing activities for that highly responsive customer.

Homework Assignment:  Look at the row with twelve months of recency. We'll talk about that row tomorrow.

What Is An "Action Stream"?

You'll hear me talking about "Action Streams" in the next few weeks, and for good reason.

What is an "Action Stream"?

It's the (unique/different) marketing activity you initiate when a customer "takes action". You want customers to take action (to "engage" as the muttonheads say), it shows you that the customer is interested in you.

Want to see an example of why you need Action Streams?

The table below is for customers with one (1) life-to-date purchase. Each row represents months since last purchase. There are two columns of interest ... one column represents repurchase rates in the next month for customers who did not visit your website last month ... the other column represents repurchase rates in the next month for customer who DID visit your website last month. Which group of customers do you think performed better in the next month?

Oh my goodness!

A customer who last purchased 31-36 months ago AND visited your website last month had a 5.6% chance of purchasing next month.

A customer who last purchased three months ago and did not visit your website last month had a 3.3% chance of purchasing next month.

Which customer is more valuable?

Which customer, according to your antiquated RFM-based system, would you prioritize?

• 3 Months of Recency Customer.

Which customer, based on facts and reality, should you prioritize?

• 31-36 Months of Recency who visited your website last month.

An "Action Stream", therefore, is a series of marketing events you initiate because the 31-36 month customer just visited your website. There should be a unique "Action Stream" for this customer, different than your tepid batch-and-blast email campaigns, different than the postcard you send to remind all customers that it is January clearance and you are 60% off this week.

If you don't understand what unique marketing activities you'd employ once a lapsed customer visits your website, tailored to that unique customer, well, it's time for you to start thinking about Action Streams.

More on the topic tomorrow.

Bob Uecker

He passed yesterday - he was the soundtrack of a Wisconsin summer for five-and-a-half decades, though you might know him from the movie Major League.

My Grandfather used to keep score of Brewers games in the mid-70s (helping fuel my budding interest in statistics) ... he'd just get up from the living room at 6:55pm, pour himself a 7oz glass of Kingsbury beer, open up his season-long scorebook, and lose himself in Uecker-isms via a transistor radio. His Father-In-Law lived upstairs, listening to games. That guy was born 19 years after the end of the Civil War. Time flies, folks.

What a colorful life.

18 Months Away And Then BOOM, The Customer Reappears: Action Streams

In old-school print, you segment the customer (16-18 months of recency, 1 life-to-date purchase), look at historical performance (profitable or not), then decide whether you send a bunch of paper to the customer. You throw 84 pages in the mail and hope for 1% of customers to buy something, the rest is pure waste.

In traditional e-commerce, the 16-18 month 1x customer gets your standard array of 5-20 batch-and-blast email campaigns per week, not personalized (from a merchandising standpoint).

Smarter e-commerce professionals personalize "what" the customer sees in email marketing based on historical purchase activity. These professionals generate 20% to 50% better click-through and conversion rates than traditional e-commerce professionals generate.

Modern e-commerce gurus POUNCE on the slightest bit of customer interest. When the 16-18 month 1x customer decides to visit your website, it's "action time".

• The customer is moved into a separate email "stream".
• Content shifts from education / entertainment (designed to create engagement) to "buying something now". Many of you already do "some" of this (cart abandonment programs) ... but your personalized "Action Stream" should match purchase history with what the customer just looked at. If you were ever going to offer a discount/promotion, this might be a good time to explore the topic.

Are "Action Streams" hard work? Sure.

Action Streams are also your future, if you're not already doing something comparable.

• Use social / video to keep the customer interested.
• Use batch-and-blast email campaigns to link social / video to commerce.
• Use Action Streams to POUNCE on customers who just visited your website and showed the slightest bit of interest in your brand.

Some of Those Customers Suddenly Reappear

In the stone ages of e-commerce (1999), I worked at Eddie Bauer. We had a Home business.

Those of you who have worked in Apparel and Home fully realize that these are two completely different business models.

• Apparel = Purchases every few months.
• Home = A cluster of purchases, followed by dormancy, followed by a potential future cluster of purchases.

You can spend marketing dollars marketing to existing Apparel customers, because their purchase cycle is short. Future purchases are always coming.

Home? Be careful. Our Home business at Eddie Bauer never turned a profit. The closest we got was in 1999 when we lost a quarter-million dollars. What did we do differently? We treated customers who just bought differently than we treated prospects.

• We stopped mailing most existing Home customers after three months.
• We rented every possible customer we could from Pottery Barn and Williams Sonoma, mailing those customers instead of our own Home customers.

We live in a different world today. Regardless, you shouldn't waste marketing dollars on customers who are not likely to repurchase. When a customer isn't likely to buy, you use email marketing and social channels and video to maintain a low-cost relationship - to remind the customer that you're still there.

• The minute the Home customer visits your website and looks at Home products, you should immediately "Action Stream" a series of in-house marketing designed to close a sale right now. It's a different marketing cadence than you can get away with in other product lines.

Some of those Home customers suddenly reappear. Modern software allows you to know the minute a lapsed customer visits your website, enabling you to trigger specialized campaigns with a personalized merchandise assortment.  Why we don't take advantage of website visits among lapsed buyers with triggered campaign paths is beyond me.

Momente të shkëlqyera në historinë e të gjithë kanaleve

 

Subscriptions

More than a decade ago I worked with an e-commerce startup. You could buy their products online, or you could sign up for a subscription, allowing you to continue to receive products on a monthly basis.

My job was to evaluate the subsequent behavior of a subscription customer vs. a classic e-commerce customer.

• The e-commerce customer bought +/- 3 times per year, had a reasonable repurchase rate, and continued to buy +/- 3 times per year moving forward.
• The subscription customer? The customer bought for three consecutive months ... unsubscribed ... then just disappeared. No/Few subsequent purchases.

You should have seen the mortified look on the face of the Owner - I met him in Seattle for lunch, to go over the results of the analysis. He looked like somebody slapped him in the stomach with a pickleball paddle. He realized, quickly, that he simply accelerated all purchases that would be spread out over time into a short window, got a sugar high for his efforts, and then immediately needed new customers or he was finished.

A year later, his business was finished.

It isn't easy to do, but a good subscription program accelerates purchases without dissuading subsequent activity. The best subscription programs capitalize on perishable goods that must be replaced. You need 5G to make your iPhone work, so you keep paying for it. You don't necessarily need to keep eating pretzels with specialty mustards.

It's $690 For a Brown Scarf

Yeah, take a look.

Do you have any idea how hard it is to get a customer to pay $690 for a scarf?

In a world where you can pay $9.99 for a perfectly credible scarf delivered in just hours, it is infinitely hard to accomplish this task.

And yet ... somewhere, there are marketers who convince their target audience that this is a perfectly good use of customer funds. And don't tell me "well, those are rich people so the rules are different" because yes the rules are different, but you still have to convince the customer to spend the money that way vs. an infinite number of ways that rich people can spend money.

You'll find a thousand vendors who will help you figure out how to use technology to offer the exact right discount at the exact right time to the exact right customer.

You'll find virtually nobody who is able to convince a customer to pay $690 for a scarf.

I just wonder how much we've all been deluded by third parties to chase the lowest-common denominator?

Judy, Jennifer, Jasmine

Back in 2012 I put together a popular series about three personas ... Judy, Jennifer, and Jasmine (click here). So popular, in fact, that I was invited to speak at a conference in New England about the personas (here's the agenda for those of you who enjoy nostalgia). These were heady times ... you didn't need to bring $20,000 to the agency running the conference to be able to address the entire audience like many conferences require of their speakers today. Keep that in mind as you think about the NRF's "Big Show" that is coming up (yes, somebody sent me the pitch deck they received for how much people have to pay to play there .... that's what modern conferences are all about ... you are most assuredly not getting brilliant insights from the best and brightest, you are getting content from people paying for access to you).

Thirteen years later, the personas have aged.

• Judy = 72 years old.
• Jennifer = 56 years old.
• Jasmine = 40 years old.

Judy is the catalog-loving shopper ... notice that she's been retired for seven years. If you manage a catalog brand and Judy is still your customer (she would have been 42 years old in the era of J. Peterman on Seinfeld), project her spending levels into the next decade as she enters her eighties ... then project your paper/printing/postage costs, and you're in for a scintillating exploration of the Darwinistic principles of Capitalism.

Stunningly ... Jennifer is nearly Judy's age from that presentation in 2012. And she behaves nothing like Judy.

When I gave the presentation in 2012, the audience audibly groaned when I introduced Jennifer. They ... did ... not ... like ... her ... at ... all. Which is too bad, because today these brands realize they NEED Jennifer. They need to know who she is, they need to know what she desires. The time to start knowledge exploration was back in 2005. The second best time to do this was back in 2012. 

The third best time to start is today.

P.S.:  There are some in this audience who strongly believe in certain marketing tactics. Let's ignore tactics for a moment. Answer the following two questions for each individual below ... first, how much disposable income does each individual have, and second, what needs does the individual have that you are able to meet with your current merchandise/product assortment?

• Judy, 72 Years Old.
• Jennifer, 56 Years Old.
• Jasmine, 40 Years Old.
• Jadyn, 24 Years Old.

The answers to your questions should be telling, and should inform you of what you need to do in 2025 and beyond to be relevant.

Weighted Email Attributes

I'll use my headphone hobby as an example.

Weighting (you'll develop your own weighting scheme based on a regression model).

• 0-12 Month Purchases = 100% Weight.
• 13-24 Month Purchases = 60% Weight.
• 25-36 Month Purchases = 35% Weight.
• 37-48 Month Purchases = 20% Weight.
• 49-60 Month Purchases = 12% Weight.
• 61+ Month Purchases = 7% Weight.

Let's say a customer purchased two times.

• October 10, 2024 = $200 on in-ear monitors, $140 on a dac/amp.
• May 1, 2021 = $800 on open-back headphones.

Weighted spend:

• In-Ear Monitors = $200 * 1.00 = $200.
• Dacs/Amps = $140 * 1.00 = $140.
• Open-Back Headphones = $800 * 0.20 = $160.

Weighted percentages:

• In-Ear Monitors = 200/500 = 40%.
• Dacs/Amps = 140/500 = 28%.
• Open-Back Headphones = 160/500 = 32%.

On the surface, the customer spent 800/1140 = 70% of historical spend on open-back headphones.

Because the open-backed headphone purchase happened a long time ago, the purchase isn't weighted as heavily ... so it's "relevance" is 32% of past purchase activity.

You'd market all three categories to the customer, given that weighted history is mostly equal. And yes, you'd correlate historical weighted percentages with future activity to "know" that the weights are relevant.

Regardless, use weighted email attributes to personalize your assortment to each individual email subscriber.

Two Clicks

There's a nice hierarchy in email marketing.

1. Segment buyers by the Elite / Loyal / Quality / Average / Struggling / Lapsed framework.
2. Segment email subscribers by Inactive / 1 Click-Through Past Year / 2+ Click-Throughs Past Year / 1+ Email Purchase Last Year.

Overlay (1) by (2) and you have something!

Regardless how many email campaigns you send per week (or per day) clicking through two campaigns in the past year is a meaningful metric ... those are the email subscribers you focus extra effort on.

You've Probably Done This Already ...

... but I sense some in the audience would appreciate a refresher course.

Test:

• Group "A" receives all emails in the next month (i.e. 25 contacts).
• Group "B" receives some emails in the next month (i.e. 10 contacts).
• Group "C" receives no emails in the next month (0), though I realize some of you find this unpalatable, so you could make it a small number (i.e. 4, one per week).

Results:

• Measure total spend after a month, at a customer level.
• Group "A" = $25.00.
• Group "B" = $23.00.
• Group "C" = $20.00.

Adjusted Results:

• We adjust for the fact that the customer would have spent $20.00 no matter what ... that's the organic amount.
• Compared to Group "A", your organic percentage is 20/25 = 80%.
• Group "A" = $25.00 - $20.00 = $5.00.
• Group "B" = $23.00 - $20.00 = $3.00.
• Group "C" = $20.00 - $20.00 = $0.00.

From here, we can fit the relationship with a power function.

And you now know the amount of incremental demand/sales you generate at each number of contacts in the month studied. Also, yes, I understand that your analytics guru thinks this is a poor way to design a test and then measure results. No worries. It's important to be in "do something mode" in 2025.

Lost

There are many ways to tell if a marketing team is lost.

One of the most important metrics is new customers acquired vs. new customers needed to maintain flat sales (or whatever your sales goal is). If you aren't measuring this, you are already lost.

Meanwhile, we have email marketers. 

Here's how one brand treated me from 1/1 - 1/5.

• January 1 = 3 Campaigns.
• January 2 = 3 Campaigns.
• January 3 = 5 Campaigns, Including 3 Within 9 Minutes of Each Other.
• January 4 = 3 Campaigns, Including 2 Sent Within 1 Minute of Each Other.
• January 5 = 3 Campaigns.

That's 17 Campaigns within 114 Hours. One every 6.7 hours. Throw out 8 hours of sleep per day and they sent me one every 4.8 hours.

Of the 17 Campaigns sent within 114 Hours ... a total of zero (0) featured any merchandise above-the-fold. Zero (0). Every single message featured a percentage off, with anything goes between 30% and 60%.

This is where the email marketer typically emails me (see what I did there), suggesting they are not lost, suggesting instead that the Management Team is lost because the Management Team demands they send this cadence with this style of messaging. That's fine. But somebody is lost.

Now, it is possible that somebody A/B tested these ideas and these ideas worked best.

But if the cadence was not A/B tested? Then the marketer is lost.

P.S.: Yes, my teams executed email marketing campaigns back in the day, and they executed campaigns where, at any time, a store manager could request that their campaign trumped the e-commerce message ... and yes, at any time, four store managers could request that their campaigns trumped the e-commerce message and we had to prioritize which of the four store managers got to communicate to the individual customer. And if no store managers requested access to the customer, the customer received one of ten (10) possible campaigns, all ten with the same core marketing message but personalized with a different merchandise assortment based on prior customer purchases.

My team did this. In 2004 they did this. They were spectacular marketers/analysts. 

So do not tell me how "difficult" your email marketing job is, resulting in having to send three campaigns to me within minutes of each other communicating different messages. Your email service provider would LOVE the opportunity to execute something sophisticated on your behalf.

P.P.S.:  My best e-commerce clients generate about 30% of annual sales from email marketing, and their programs are not communicating sale messages with varying percentages off every 6-7 hours. You'll know when your email marketing program works well when you generate 20% to 40% of sales from email marketing. Generate less and your email marketing program isn't working well as it should. Generate more and the rest of your marketing program is not working as well as it should.

Not An Outpouring, But ...

I performed a Saturday test (click here).

Turns out that if I talk about Affiliate Marketers scamming you, you respond. En masse. Not an outpouring, but one of the larger post-COVID outbursts I can remember.

Your responses suggest that some vendors are out to mess with you, without regard for the health of your business.

The past year has been so darn fascinating. In catalog marketing, your paper/print partners messed with you. You didn't like that. Remember that feeling.

Yesterday, the e-commerce audience responded, loudly, with stories of being ripped off by digital vendors.

When you (unscrupulous vendors) scam my clients, how does that make you feel? When you buy a new Lexus because you siphoned off a portion of client profit for your own benefit ... by lying ... how do you sleep at night?

I worked in the vendor world for one (1) year, back in 2000, at a retargeting startup. My goodness. What a train wreck. While the company generally tried to do good for clients, the employees were all about stock options, and would do what the felt was "right" to boost the stock price so they could retire as a 28 year old. I have a patent for a product that I created in response to employees choosing to not serve clients, causing me to invent a competing product within my own company so that clients could benefit. I know a little something about vendor employees and self-interest, even if that knowledge is dated.

P.S.:  What's awful is that nasty vendors get all the attention. I had a video conference a few months ago with a catalog agency that keeps things quiet ... minimal publicity, working hard to do "what is right". Really good vendors are out there, folks. They just keep their good works quiet.

Greatest Twitter Marketing Thread Ever

Just read the entire thread ... every word (click here). Every word, please.

Don't have Twitter? I did you a solid ... here is the text. Everything that is wrong about digital marketing, corrupt vendors, and muttonheaded marketers at brands who can't be bothered to ask simple questions is on display here.

My team built the leader competitor to

@honey

This "scam" story is absolutely wild to me.

In his video today,

@MKBHD

makes two claims about Honey's foul play:

1. Honey effectively steals commissions from their affiliate partners by being the last to drop their cookie when a user checks out.

2. Honey's value prop to retailers is dubious since they are giving customers discounts on products they were already about to buy. So, why is this wild?

Because this is exactly how it has been since the beginning.

In 2016, I was at a startup that aspired to take on Amazon by consolidating every product on the internet via a Chrome Extension.

Turns out that's pretty hard to do, and after working at it for years, we were coming up short.

Our business team noticed another e-commerce Chrome Extension (guess who?) seemed to be making waves and growing fast. They dug in a bit, and while I wasn't in the room, I gather their reaction was something like this: They're getting paid HOW MUCH to do WHAT?! From the start, it sounded ridiculous. A user on the checkout page is at the bottom of the funnel. They are the highest-possible-intent customers. Right on the cusp of purchasing of their own free will. At full price. Of course consumers were loving it -- it was the closest thing to "free money" they could get.

But the retailers? Why the hell were they paying for this?

It all makes sense -- sort of -- when you understand how affiliate marketing works.

(Feel free to skip this part if you already know it)

The original purpose of affiliate marketing was to enable hardworking content publishers (like Marques) to get paid for referring customers to retailers (like Best Buy). (This is just an example, I don't know who he has affiliate agreements with, if anyone). It's a brilliant idea, and it makes quite a lot of the internet go round. It's largely done via "affiliate networks" -- companies like Rakuten, Impact Radius, CJ Affiliate -- who make it easy for retailers and publishers to pay and get paid.

It's quite obvious why Best Buy would want to pay someone like Marques (again, just an example) for referrals -- he *sends* customers who otherwise wouldn't be shopping right to Best Buy's door. That's value!

I'm not an expert on the history of Affiliate Marketing, but I gather it went roughly something like this:

1. Some time in the early/mid '90s: Affiliate Marketing invented.

2. Affiliate Marketing catches on like wildfire since it drives tons of new customers to online retailers and is much easier to measure ROI than, say, television ads. (Somewhere in here, Affiliate Networks are invented) 3. Online retailers begin hiring Affiliate Marketing Managers -- and even entire departments -- thanks to its overwhelming success. 4. Affiliate Marketing Managers are granted increasingly large budgets to spend on Affiliate Marketing initiatives. 5. ... < next section >

6. Honey is accused of massive scam

1. User lands on retailer (Macys, Foot Locker, etc) checkout page.

2. Chrome extension pops up, saying "We found N coupons. Want us to try them?"

3. User clicks "Yes."

4. -This one is critical- Chrome extension marks its territory by:

a) redirecting the browser with a retailer-specific affiliate link, or b) dropping a cookie, or c) running some custom javascript in the page.

Importantly, only one publisher can get attribution for a sale. And it's always the last one.

5. Chrome extension tries coupons, and applies the best one if any. 6. Customer finishes checking out (with their coupon discount, if applicable).

7. Retailer forwards sales data back to Affiliate Networks, who pay out commissions to publishers like Honey (and influencers like Marques).

So, why were retailers like Best Buy, Macys, Home Depot, etc paying Honey (and us) to give discounts to customers who were already about to buy a product at full price?

For us, the pitch was straightforward: "You're already paying Honey for it; why not pay us too?" (Thanks, Honey!)

But for Honey? The trailblazers?

In short: bad incentives, and incompetence.

Note: This is a bit of speculation on my part. I'm not referring to any one person or organization here -- this is broad strokes.

Here's how (I think) it went down:

Retailers set Affiliate Marketing budgets for the quarter or year. Their Affiliate Marketing teams are encouraged to spend -- if not rewarded for spending -- their entire budget.

Bonus points for discovering trendy new channels that customers love! Honey had a great brand, great user growth, and (I can only assume) a killer sales team. Some affiliate manager probably got promoted for bringing Honey on early! Honey's pitch? Something like: "Customers are more likely to convert (buy the product in their cart) when they've tried coupons first. It gives them confidence that they're getting the best deal!"

Note: This is a completely testable hypothesis...

I am 99% certain that Honey was *never* challenged on that claim by any of their retailers.

Why am I so certain?

In the throes of building our Honey competitor, I once asked our (absolutely killer) BD guy:

"How many retailers have asked you for hard data showing that trying coupons on the checkout page actually increases conversion rates?"

Those were data that we could easily collect. It's the obvious litmus test of whether we're adding real value, or if we're just parasites exploiting a (what we believed to be short-lived) loophole. His response? "One."

Thousands of e-commerce retailers paying us to try coupons on their checkout pages, and only one ever asked us to back up our central (dubious) claim.

It was site #537 or so in terms of priority. Boutique farm equipment reseller or something. The guy who asked was the CEO, if I remember correctly -- the kind of person who would actually care, or even think, to ask about ROI. Our BD guy's response to this CEO's more-than-reasonable request for evidence?

"Go pound sand!" (in so many words).

The crazy part about this Honey story coming out now is that we all saw it coming back in 2016.

Quite frankly, we thought this moment was so right-around-the-corner that we nearly didn't bother building the thing ourselves.

8 frickin' years ago. But we did, and we got acquired for it.

So did Honey, to the tune of $4 billion.

Innovation Responses

I asked readers to offer examples of e-commerce brands executing tactics that are "innovative". Based on my metrics, several thousand of you read the requests on LinkedIn and on this blog. Five responded. Here are the responses.

From LinkedIn: I don’t know how “innovative” this is — but the best ecom brands seem to proactively cultivate talent from within, rather than recruit from the outside as a first option. I’m surprised more DTC brands don’t pay more attention to player development (like sports teams).

From LinkedIn: See traq.ai for sales conversions and flow.space for growing CPG brands.

From LinkedIn (this one doesn't seem authentic): As a fellow e-commerce professional, I would like to acknowledge the innovative tactic of using AI-powered chatbots to improve customer service and increase sales. Companies like XYZ are leading the way in this area.

From Our Blog: Here we go: https://www.lavialla.com/. Product: overpriced, exchangeable Italian organic food sold by everyone else. Half jokingly. I have seen them telling their story and growing year over year while the products they sold were available for a fraction of the costs in any urban area. Or on Amazon, which they don't use. 

From Our Blog: I was shown the 'Your Style Assistant' (bottom left of home page) icon on michaelkors.com a couple of months ago. Its conversational Ai, allowing customers to input what they're shopping for, with results instantly served. It's provided by Dynamic Yield.  The use case I was shown: 

• I'm going to a wedding need a dress - statement dresses shown
• It's in Bali, do you have light weight dresses - relevant product shown
• Do you have any in brighter colours - relevant product shown
• Do you have any for under $200 - relevant product shown
• Do you have shoes to go with that - relevant product shown

A much more tailored, relevant, enjoyable shopping experience. I understand voice activation will be phase 2.