Inflection Point Strategy

FYI - I'm about to complete a new booklet called "Hillstrom's Business Review". This booklet links my Marketing Budget Experiments work with Inflection Points and Merchandise Forensics.

One of the fun aspects of my work over the past two months is the development of an Inflection Point strategy for companies. I build models that predict how likely a customer is to purchase in the next year, the model is converted to a segmentation strategy that is generally used in email marketing to better target customers.

That's an oversimplification illustrated above - this stuff is performed/modeled at a customer level, so you know the appropriate strategy for every single customer within every single email campaign.

I know, most of you are just blasting 1-2 messages per day and wondering why response is so tepid. I get it. With a little bit more work, you can perform an awful lot better than you perform today.

NOTE:  If you want to get ahead of the curve and get in on a "Hillstrom's Business Review" project before I announce the beta, you'll get the absolute best price ... better than the beta/pilot price that will be announced in 1-2 weeks when the booklet is available. Contact me for details (kevinh@minethatdata.com).

Comp Segment Performance

Next week I begin analyzing a ton of Elite Program projects.

And unfortunately, sometimes I see something that looks like this.

That's a comp segment analysis ... I've been talking about comp segment analytics for a decade, as you are painfully aware. Here's a case where merchandise productivity crumbles in March, slumps even more in September, and is just an all-around disaster by the end of the year.

This tells us that the merchandising team messed up.

How did the mess up? Take a look at new merchandise.

My goodness. What a mess. The merchandising team just plain failed to properly manage new items (or failed to introduce enough new items or did both).

How about existing items?

They don't have a problem with existing items.

It's a new merchandise problem.

You're probably already running this table, right? If you aren't (for some reason), let me know and for just $1,800 as a new client we'll get busy with this run of the Elite Program.

There Are Months That Yield Bad Customers

In a recent project, customers buying from the following months were worth 10% less in the future:

• October.
• December.
• January.
  
• February.

Customers buying from these months were worth 10% more in the future.

• April.
• September.

Christmas and Discounting Season (Jan/Feb) yielded customers who weren't buying from the brand, they were buying for Christmas or because merch was cheap. That's an inflection point. You need to act upon the inflection point.

Similarly, this brand has new merchandise in April and again in September ... those are two months were new products are introduced. Turns out customers like that. Hmmm? That's an inflection point. Act upon it!

When Search Goes Bad

Want a search-related inflection point?

When analyzing inflection points, I measure the impact of orders attributed to search on future response. Here's a fun tidbit:

• Search can be a strong positive on a first order.
• By the time the customer orders for a third time, the fact that the order is attributed to search becomes an inflection point ... the customer is frequently less responsive going forward.

In other words, at the very time the customer should be becoming loyal to your brand, an order attributed to search suggests the opposite ... that the customer doesn't trust you.

Measure the dynamic for yourself. If you can't do it, contact me (kevinh@minethatdata.com) and I'll get busy doing just that for you.

Driving Business ... To Amazon

It's 2006, back in the stone ages. Gnarls Barkley sang "Crazy" and the song ended up being the top ranked song of the year. You could get a loan for a home by breathing on a mirror to demonstrate condensation. Amazon was becoming a force.

My team at Nordstrom had massive catalog mail/holdout tests. If our circulation was 2,000,000 catalogs, we held out 200,000 customers. Yeah, 200,000! This allowed us to measure lift at a segment level.

We also had a proprietary credit division at Nordstrom. This meant that if a customer purchased at Amazon (or Neiman Marcus or Target or your local gas station) using our credit card, we could attribute purchases to our mail/holdout test.

Yeah, that's good stuff!

So good, in fact, that we learned the following fact.

• For every incremental $1.00 we generated at Nordstrom from mailing a catalog, we drove $0.20 to Amazon.

Yeah, that's not so good.

I recall one test where we for every $1.00 we generated at Nordstrom we generated $1.00 for Amazon. Millions upon millions of dollars of free business we sent to Amazon via our own marketing efforts.

Is it possible that this happens at your brand? You send out an email "blast" and you drive business to Amazon?

At Nordstrom, we knew that Neiman Marcus store openings drove business to our stores, and we knew that our marketing efforts drove business to Amazon. Imagine what your marketing efforts are truly doing?

Panel Discussion

A funny one for you.

Remember when catalogs had source codes / key codes on the back of the catalog? You'd call the contact center and Besty would say "please read the key code off of the back of the catalog" and the customer would dutifully say something like "FN2AJPFB" which actually means "Fall November Drop 2 0-3 Month 3x+ $100 AOV".

I was on a panel discussion at NEMOA in 2007 on the "future of cataloging" or something like that. A question came up ... "how will we measure catalog performance in the future?"

I remember sharing with the audience that at Nordstrom we eliminated key codes years ago and simply measured incremental lift via A/B tests (mail vs. no mail). I mentioned that there would be no key codes in the future - no need for them, and they were not calibrated to measure the complicated performance of a customer visiting a website several times a week after receiving a catalog.

The audience laughed ... in the "oh that will never happen, idiot" way that audiences tend to respond.

The late Alan Rimm-Kaufman agreed with me. The woman from L.L. Bean was mysteriously quiet, suggesting she was headed down the same path and didn't want anybody in the audience to follow her. Seriously - if she agreed with me the audience would have stopped laughing instantly and would have done whatever she told them to do because she was from L.L. Bean and L.L. Bean could do no wrong in New England in 2007. A fourth member of the panel spoke glowingly of the importance of key codes - the audience nodded with loving approval ... the "She gets us, catalogs will be here forever" look of approval.

It's late in 2023 ... we all know what happened.

Vet your panelists ... some speak about the future ("it's an AI world and we are just living in it") and know nothing about what they are saying. Others are already practicing the future and the audience isn't smart enough to realize it. Then you have the vendors selling incessantly to the uniformed ("We think all brands will leverage Pilgrim Models in the future" ... and guess what ... they sell "Pilgrim Models" ... whatever those are).

Panel Discussions ... they're an interesting pot pourri of thought leadership and actual knowledge.

An Odd Inflection Point

In a recent project, customers (across the board) who purchased new merchandise in their order were 10% more responsive going forward.

Which is a big deal. When I run a Marketing Budget Experiment project, 10% translates to a lot more than 10% over time (due to the compounding interest generated by constantly increasing the size of the customer file).

If you know this fact, how would you (for instance) change your email marketing tactics? You're sending out a daily "blast" ... surely you'd find a way to change two of those five contacts each week to speak about new products that you love, right?