Winning Items? Or Other Items?

Look at the bar chart below.

A first-time buyer buys a mix of items, no doubt, but the items marginally skew toward "anything", meaning that the items are not skewed to best sellers (those items have a value of 0.00 in this analysis ... low-selling items have a value of 1.00 in this analysis).

Outside of the crazy customers who buy all the time (often employees), loyal customers tend to skew to best-selling items, in this case.

Every company has a personality ... this company gives discounts to best customers, and best customers appear to buy winning / best selling items. One might surmise that this company is discounting best-sellers to loyal buyers.

A common trap clients fall into is the trap where loyal customers like long-term best-selling items, thereby paralyzing the brand when a merchant wants to move the brand forward with new items not guaranteed to sell well. When best customers like best items (which have been sold for a long time), the brand cannot move forward ... customers will hold the brand hostage.

Life Stage Data

Some of the most enjoyable work I perform surrounds the life stage the customer is in.

In the bar chart below, we evaluate customers after a 1st / 2nd / 3rd / ... / 51st+ order. Within that order within the life stage, the chart evaluates the percentage of the amount spent that was on items selling below their historical average price point.

For this brand, what does the data suggest?

It suggests that the brand discounts heavily to best customers. Less than 25% of sales from first-time buyers is discounted ... after ten purchases, more than 40% of sales are discounted.

Every company has a personality - what the company believes in is revealed through the data sent to me. This company cares about loyal buyers.

Chicken Efficiency

Via Twitter user @popwreckoning, we are introduced to the concept of "Chicken Efficiency". This is one of the best stories I've read all year. Read the text in the image below to the end.

Chicken Efficiency! 

You'll hear about the amazing Cyber Monday that "brands" had all day today. Realize that research professionals, vendors, trade journalists, and pundits are measuring Chicken Efficiency. 

You, however, are measuring profitability. Which is why you do not care about Cyber Monday.

Cyber Monday

It's the day that trade journalists and vendors praise companies for selling a $50 item that costs the brand $25 for just $30. They smugly suggest that all of the additional "traffic" will make up for the $20 of gross margin dollars you give up to make a theoretical "consumer" happy.

a) Selling a $50 item that costs you $25 gives you $25 of gross margin.

b) Selling a $30 item that costs you $25 (40% off, woo-hoo) gives you $5 of gross margin, requiring 5x the traffic converting at the same rate for you to make the same amount of profit.

I've told you this story before ... but it symbolizes Cyber Monday. I'm sitting in an Executive Meeting (as a consultant) at 8:00am on Cyber Monday. An employee walks into the room, hands the CEO a folded piece of paper. The CEO opens the folded piece of paper in a manner not unlike a 16 year old boy would in high school chemistry after having the paper passed across six desks from a girl named Becky, reads the contents, looks up, and then pauses the meeting. He addresses the room.

"Brand X just announced they are at 50% off today. Fifty percent! We're only at 40% off. If we want to remain competitive, we need to be at 50% off today as well. Charlie (the marketing executive), tell you team to redo the creative on the 9am blast and switch to 50% off. Have them create a new discount code and get that code on the website as soon as possible." Charlie, the CMO, headed out of the room with a look on his face like his scrambled eggs for breakfast weren't fully cooked and his tummy was just realizing that fact.

Humans.

Here is a chart from a company that doesn't even care all that much about Cyber Monday. We look at percentage of sales that day that come from items selling below their historical average price point. Tell me what you see.

Yeah, for one day they just gave some merch away in an effort to please, well, um, I have no idea who they were trying to please. By the way, if you average sales for the Monday before Cyber Monday and for the Monday after Cyber Monday, you learn (for this brand) that sales did not increase on Cyber Monday. Prices were lower, profit was lower, sales were identical.

Notice the heavy discounting in the week prior to Christmas. This company switches into "liquidation mode" and starts moving out winter stuff in anticipation of Spring/January.

In mid-November, the brand begins to discount more. The brand discounts even more on Cyber Monday, then reverts back to pre-Cyber-Monday behavior, before clearing out stuff in late December.

What does the graph depicted above look like for your brand?

Which Customer/Product Combination Yields The Best Gross Margin Percentage?

Back to our table format studied this week. This time, each product / customer combination measures gross margin percentage.

The green cells represent the highest gross margin percentages.

Do the best selling items have the highest gross margin percentages?

No.

Do the best customers have the highest gross margin percentages?

No.

If you think your best customers and your best products should yield the highest gross margin percentages, trust your instincts.

This business isn't being optimized properly.

Have you run this table for your brand?

No?

The contact me (kevinh@minethatdata.com) and let's get busy.

New Items Are Not Best Items

Back to the table framework we've been studying this week, take a look at share of sales from new items by product decile and customer decile.

Two things are worth pointing out in this table. First, new items are seldom best-sellers. New items require patience ... they need to be developed (just like customers need to be developed). Second, there are interesting trends in the table. The best customers are less likely to buy new items than are other customers. We teach our best customers what they should purchase, and guess what? They purchase what we tell them to purchase!

The second point is an important one. If our merchandising/product teams want to sell something different, our best customers frequently say "no" ... the best customers hold back brands looking to change. This is why, so often, it is best to launch a whole new brand than it is to try to transform a brand from within. Best customers simply don't want to change.

This Happens A Lot!

I see this one a lot ... look at the New/Reactivated row, and compare the price these customers pay for merchandise compared to the best customers.

Among the best selling items (the 5% column), best customers paid an average of $62.43 per item, whereas new/reactivated customers spent $71.96. This difference isn't as extreme through the remainder of columns, but the difference exists.

This happens A LOT in my project work. Either best customers are being given deals/discounts (whereas new/reactivated buyers are frequently given free shipping), or the merchandise preference is slightly different. Given that I'm controlling for the sales potential of the items in this table, it's a discounting issue. 

Sometimes my clients wonder why it is so hard to acquire customers? Sometimes it's hard because you're asking new/reactivated buyers to pay 10% more per item than your loyal customer base is asked to pay. 

Project Opportunity

I am in the process of finalizing a project package around discounts/promotions/pricing.

As is usually the case, I offer a significantly discounted project opportunity for loyal blog readers as I finalize the project code and test usability of the concept.

For you, that means I am offering two (2) readers the opportunity to purchase the discount/promo product (you've been reading about elements of what this project will become) for just $12,000 (it will be a $24,900 project when it is officially released).

Contact me immediately (kevinh@minethatdata.com), because both spots will be gone shortly.

Where Are The Discounts Being Applied?

Sometimes discounts/promotions are applied to items via liquidation efforts. This is one of those cases.

In this table, each row represents a group of customers. Best customers are at the top of the table (5% = best decile of customers, 15% = next best decile of customers). New/Reactivated buyers are at the bottom of the table. Meanwhile, item sales are rank-ordered across the top of the table. The best items are in the 5% column, the worst-selling items are in the 95% column.

Which cells (red) represent customers buying discounted products? The columns at the far right side of the table tell the story ... the worst-selling items are selling at/above their historical average price point about 70% to 75% of the time. The best-selling items sell at/above their historical average about 85% of the time.

So we have two dimensions at play ... marketers are offering discounts/promotions, no doubt about it. But at the same time, the merchants are liquidating lousy products. Both parties contribute to the challenge ... as is generally the case.

Optimal Price / Discounting Ratio

Ok, this isn't optimal because I'm not looking at gross margin dollars (that's reserved for clients), but you'll get the idea here about what I'm looking to accomplish.

I created regression models that predict the impact of prices on rebuy rates, and another model that predicts the impact of discounting on rebuy rates (again, gross margin work is done for clients).

The average price point for this category was $50.00, and the average percentage of items sold at/above the historical average price point was 75%. This yielded a 51.6% rebuy rate.

What happens if prices increase to $55.00 and discounting remains constant?  Rebuy rates decrease to 49.7%.

What happens if prices increase to $55.00 and discounting increases so that just 56.3% of the items are selling at/above their historical average price point?  Rebuy rates maintain at 51.5%.

This dynamic is coming to an omnichannel brand near you (your mileage will vary). Prices increased a year ago and continue to increase, customers balk at higher prices and purchase less often, leading "brands" to discount more to maintain response (which means that gross margin dollars decrease and the brand is less profitable).

It's hard to fight the customer. The customer is telling us how much s/he is willing to spend, and the customer does not care one bit, one bit, about how our cost-of-goods increased. Many of us are going to give up profitability, either by increasing prices / decreasing response, or by increasing prices / discounting (which decreases gross margin dollars).

Sometimes Discounting Leads To Positive Results

The secret to discounting is to generate more gross margin dollars and build a stronger customer file (for some, it is to liquidate merchandise, but that is a topic for another day). The goal should never be to steal market share, because honestly you'd have to be a ten billion dollar brand (or larger) for that to even make the slightest difference.

So, if you choose to sell a $50 item with a $20 cost of goods for $35 via your promotional/discounting program, you must sell twice as many units to equalize gross margin erosion.

I worked with a smart brand who knew what this relationship looked like, and they discounted only enough to make the math work (or, in this case, come very close to working). In their case, I created regression models that predicted how much customers would spend next year on items selling at/above their historical average price point, how much customers would spend next year on items selling below their historical average price point, and how much gross margin dollars customers would generate next year.

For customers who spent $100 on items at/above their historical average price point and spent $0 on items below their historical average price point ...

• $64.41 spent next year on items at/above the historical average price point.
• $8.44 spent next year on items selling below their historical average price point.
• $73.05 spent next year, total.
• $45.37 gross margin dollars generated next year.
• 62.1% future gross margin.

For customers who spent $0 on items at/above their historical average price point and spent $100 on items below their historical average price point (discount-centric customers):

• $56.51 spent next year on items at/above the historical average price point.
• $28.74 spent next year on items selling below their historical average price point.
• $85.25 spent next year, total.
• $44.27 gross margin dollars generated next year.
• 51.9% future gross margin.

This company comes really, really close to getting the math right. Full price customers generate $73.05 next year while discount-centric customers generate $85.25 next year. Discounting led to customers who spent 16.7% more in the next twelve months. However, full-price customers generate a 62% gross margin next year while discount-centric customers generate a 52% gross margin next year. Full-price customers generate one extra dollar of gross margin in the next year.

This is one way to make discounting work. As long as gross margin dollars are equal/greater and as long as the customer file is stronger, the math "can" work. I have a responsibility to share this fact with you.

But you have to do the math, perform the analysis, and understand the tradeoffs. At minimum you have to take the math down to future gross margin dollars. In a smarter world, you'd include pick/pack/ship information as well, and take into account any additional p&l metrics that matter.

By Year of Introduction

This example is from the pre-COVID era ... and the example didn't turn out the way I expected it to turn out.

When I looked at the percentage of sales at/above the historical average price point for an item by year of item introduction, I expected the brand to discount items more and more as the item aged. I didn't expect this:

Some of the trends make sense to me. Newly introduced items (look at 2016 / 2017 / 2018) are more likely to sell at full price than are older items. This happens all the time ... older items are discontinued and are consequently discounted.

2019 looks different - you can see that this brand made a solid effort of re-establishing price integrity after several years of discounting.

As many of you are experiencing in 2022 ... when prices go up (either due to the end of discounting or because of cost of goods sold increases), response goes down. That's how this stuff works. Eventually, you land on a local maxima where pricing is theoretically optimized ... until it isn't once again.

Pricing Evolution

This is from the pre-COVID era ... the before times ... I'm analyzing a brand that, over time, was becoming healthy. Look at the percentage of sales sold at/above historical average price points by year by pricing band (0 = lowest prices, 9 = highest prices).

When the company discounts, where does the company discount? With the lowest price points. The brand maintains price integrity on the most expensive items.

Compare 2019 to 2015. Tell me what you see?

Yeah, in 2015 this was a sick brand, selling half of merchandise at a discounted price. In 2019? Much healthier.

If you are a consultant like me, it's pretty easy to see when a brand is healthy or unhealthy, based on the prices that items are selling at over time. I realize that in 2022 many of you were forced to increase prices AND your business became unhealthy in the process ... that's a different issue than the one we're analyzing above.

Who Gets The Deals?

This company doesn't do a lot of discounting, but they marginally prefer to offer discounts to the very best customers or the very worst customers. This signature comes up often in my work.

Again, the differences aren't huge (77% of items sold at/above their historical average price point for best/worst customers, 81% for middling customers).

But there are instances where you'll see 65% sold at/above historical average price point for an item among average customers and 25% sold at/above historical average price point for best/worst customers. Those situations don't end well ... you train the customer to wait for deals, and the "average" customer simply never purchases. Seen it a lot. No bueno.

Which Cell Generates The Most Gross Margin Dollars?

We've learned several things in our case study this week, haven't we?

• Customers trained to enjoy discounts continue to enjoy discounts.
• Customers trained to pay full price continue to buy via full price.
• The customers who spend the most in the future pay mostly full price with "some" discounting.

So, which cell generates the most gross margin dollars on an annual basis?

The results are similar to the future demand/sales table.

• 1x Buyers?  51% to 75% on first order.
• 2x Buyers?  76% to 99% historically spent on items at/above their average price point.
• 3x Buyers?  76% to 99% historically spent on items at/above their average price point.
• 4x Buyers?  76% to 99% historically spent on items at/above their average price point.
• 5x Buyers?  76% to 99% historically spent on items at/above their average price point.
• 6x+ Buyers?  51% to 75% historically spent on items at/above their average price point.

In this case, you don't want customers who only pay full price. You want a mix ... but the mix needs to be heavily skewed toward items selling at/above their historical average.

Look at 1x buyers ... first-time buyers in the past year.  If they only bought discounted items, they generated $14.96 in gross margin in the next year.  If they spent 51% - 76% of their wallet on items selling at/above their historical price point, they generated $20.65 of gross margin dollars in the next year. 

Which customer would you prefer to acquire?

Most Valuable Customers by Price Designation?

Are the customers who buy items at a discount more likely to spend a lot in the future, or are customers who pay full price most likely to spend a lot in the future? Let's take a look at one example.

The answer? Neither!

In this example, as the customer becomes more loyal, the customer most likely to spend a lot in the future transitions from a customer who spent 76% - 99% on items at/above their historical average price point to a customer who, at 6+ life-to-date purchases, spent 51% - 75% on items at/above their historical average price point.

In other words, you want a customer who spends a mix of money (in this example) on items that are discounted and on items that sell at full price ... skewing toward full price early in the life cycle.

Tomorrow, we'll study which cells yield the most future gross margin dollars.

P.S.:  Look at the future sales numbers of customers acquired via 100% discounts (0% at/above the historical average price point). Just look at those numbers. Now go back to your customer acquisition strategy and ask yourself how much discounting you do, ok?

Training Your Customers How To Behave

Ok, here's a query for you. We take all customers and segment them based on life-to-date purchases. Then I select customers who place "average" average order values. Finally, I segment customers based on percentage of historical demand spent on items at/above the historical average price point of the item. Within the cells of the resultant query table, I measured the percentage of future (next 12 months) sales from items selling at/above their historical average.

Tell me what you observe.

The table illustrates key findings:

• Once you train a customer to bargain hunt, the customer bargain hunts.
• When you train a customer to pay full price, the customer is more likely to pay full price in the future.

Look at the customer who has bought 6+ times and has 26% to 50% of historical demand/sales coming from items selling at/above their historical average:

• 70.4% of future sales are from items at/above their historical average.

Look at the customer who has bought 6+ times and has 76% to 99% of historical demand/sales coming from items selling at/above their historical average:

• 82.4% of future sales are from items at/above their historical average.

Now, you may not think that a 12% difference in the metric matters.

It matters.

More tomorrow.

P.S.:  You create this table and have it as part of your dashboard, right?

Yeah, We Discount Online

The Executive says, and I paraphrase here:

• "It's almost like our customers don't want to shop from our catalog. We track orders we think were caused by the catalog, and even when we mail a catalog our customers seem to prefer shopping online without catalog attribution. I don't get it."

So, I dig into the information. Here's an interesting trend.

• Call Center Orders = $54.87 average price point, 29% of items sell below their historical average price point.
• Online Orders, Catalog Attributed = $47.03 average price point, 41% of items sell below their historical average price point.
• Online Orders, No Catalog Attribution = $45.25 average price point, 50% of items sell below their historical average price point.

I asked the Executive why this happens? Why do call center customers buy expensive items, while online non-catalog-attributed customers purchase items more than nine dollars cheaper, with half of the items selling below their average historical price point?

The Executive says, "yeah, we do a lot of discounting online".

And when you do that, you change customer behavior.

Those of you in the catalog world, folks who love to opine about the myriad benefits of an integrated omnichannel approach, well, you should think how integrated the approach is when you offer financial benefits to respond in one channel but not another.

Market Share and Profit

In my project work, there is a fundamental truth that is evidenced via the percentage of sales that are at/above historical averages, and the percentage of sales that are below historical averages.

• Market Share and Profit are at opposite ends of the strategic spectrum.

There are people in your company who misguidedly think that increasing sales from $88,000,000 to $97,000,000 causes you to increase market share and therefore, any profit you give up is worth it. Strong hint - increasing your market share from 0.04% to 0.06% is irrelevant.

Increasing profit is relevant. You can pay your ownership team, you can pay your executive team, you can pay all employees, you can reinvest the money in marketing, you can reinvest the money in capital expenditures. You have choices.

Do not be seduced by the co-worker who wants to increase market share. Market share is a game that Best Buy plays. It is a game not played by a brand selling widgets.

The Wrong Metrics

I frequently defer to sports, simply because the data required to understand sports is readily available. Hence, this is an interesting read ... a read about ranking the best college football teams from 2014.

I'm thoroughly convinced that, in e-commerce, we're analyzing the wrong stuff. Our fetish with conversion rates and ROAS is misguided.

So what if you executed a campaign with a 2.4% conversion rate and a 3x ROAS?

Our software tells us to optimize for campaign performance.

What if we optimized for business performance?

There is a major difference between campaign performance and business performance. One method seeks to optimize to help a brand win the week ending November 2. The other method seeks to optimize profit over five years.

Can you see the difference in strategy required to optimize conversions in a week vs. optimizing profit over a five year period of time?

Analytics Article

Read this (click here).

Now, think about your Google Analytics reporting or the dashboards you maintain/use or the vendor reporting that greatly overstates the performance of your efforts. Compare the reporting to what is discussed in the article. Do you see any parallels? You should see parallels.