7: Predicting Customer Demand
Retailers have always looked for techniques to provoke customers to spend more or purchase higher-margin items - but much of what has been done thus far are merely pseudo-psychological "tricks" based on specious theories and scant evidence. It is only within the past few decades that computers are powerful enough to handle the calculations, and "reward card" programs enable retailers to recognize the buying activity of individual customers, that any real science can be applied.
It's also observed that most of our buying behavior does not involve decision-making. It's habitual. When we purchase a product for the first time, there is intense thought and scrutiny - but if we are satisfied with the outcome, we tend to purchase the same brand automatically the next time, and the next. People remain "loyal" to the same brand for decades not because they repeatedly re-evaluate their choices and find that a given brand is superior ever time, but out of laziness to reconsider and remake the buying decision.
Another observation is that, in spite of efforts to change spending (e.g., to "cut back" on groceries), people tend to purchase the same amounts of the same items over time. They will find excuses to purchase the items they can do without, or to stay with their preferred brand even when a different one is on sale. It's also difficult to get people to try something new or different: a person whose weekly pattern never goes down the potato-chip aisle will not see the product or promotion.
Moreover, the wide variety of products provides an enormous number of possible combinations and the heterogeneous nature of American culture makes the shopping patterns largely unique to each person. It requires an enormous amount of data and calculation to identify even a small group of customers with similar buying habits.
Mass-retailer Target attempted to apply statistics to consumer behavior: with thousands of retail stores and a reward-card program, they had ample data and the ability to track purchases (what and when) to the individual shopper across stores. The company could also intensely scrutinize website behavior, knowing not only what was purchased, but what was browsed, and every click made on its site. Data could also be obtained from other sources, giving them a very detailed dossier on each of their reward-card customers.
From all of this data, Target can assemble a "guest portrait" that is very detailed. For example, they have noted that the buying patterns for orange juice are quite different, depending on whether the customer is purchasing for himself or others. A 24-year old running enthusiast has a different juice-buying habit than does a 38-year-old mother who purchases it for her school-aged children.
Originally, the firm used this data in a vague way, to better target advertising messages to customers: it makes very little sense to promote a sale on milk or beer to customers who never buy those items. But then, it began to predict future buying habits: that a person who bought a bikini in April would likely buy sunscreen in June and weight-loss books in November - and custom printing capabilities enabled the firm to send a monthly "coupon book" promoting the exact items that an individual shopper would be likely to purchase.
While we often follow habits, we do not do so perfectly: people do change brands. Once study is cited that indicates 10.5% of people change their toothpaste brand in the past six months. (EN: this seems to de-emphasize that 89.5% did not change and remained faithful to their brand - so brand-switching is highly unusual.)
Examination of brand-switching led to the concept (Andreason) of "life events" that disrupt existing patterns. Married couples seldom have separate toothpaste - most often, one spouse changed to the other's brand. A person who relocates to a different area and starts shopping different stores may be unable to find their preferred brands, and switch as a result of the life event.
The key life events that precipitate brand changes are:
- Getting married or divorced
- Change in household - someone enters or leaves the domicile
- Changing residences
- Change in employment
Of all life event, there is no greater upheaval than having a baby. From the moment it is expected to the time it leaves for college, the dependent causes major shifts in purchasing behavior. And what's more, people accept rather than resist the changes that take place as a result of having children.
Back to Target, they noticed that when people become parents, they begin to consolidate their purchases. They no longer have the time and energy to shop different retailers - so when they begin buying diapers at Target, they begin buying groceries and other merchandise from them. To new parents, convenience is a critical factor.
Because the new-baby life event is so critical and potentially profitable, Target set its statisticians to work predicting pregnancy. It was easy to focus on shoppers who registered for baby-shower gifts or started purchasing maternity goods - but this only identified a fraction of parents-to-be. By analyzing the purchases of new mothers months before childbirth, they arrived at a formula that could be up to 96% accurate in identifying when a woman was pregnant.
This led to the famous incident in Minnesota, where a man came into a store, very upset that Target was marketing maternity products to his daughter, who was still in high school - and then later had to apologize when he discovered that his daughter was, indeed, pregnant.
There's another long narrative passage about identifying which songs will be popular. Aside of the popular artists, whose fans will endorse anything they put out, what was found is that people crave familiarity. If a song seems different to what is popular, people will stop it from playing. But if it follows in the conventions of music that is already popular, they will listen through it. This is why the most popular songs are often the most insipid and uninspired.
To popularize music that is unusual, radio DJs use playlists - then "sandwich" a new piece of music between two established hits that have some similar elements. One example is cited in which 26.6% of listeners changed stations when a new song was played - but this was reduced to 5.7% by sandwiching it.
There is neurological evidence behind this, and it is theorized that we are evolved to ignore most of the noise in our environments. So when it comes to sound, something entirely new creates sensory overload and the listener hears only an incomprehensible cacophony of unfamiliar sounds - but when a set of sounds is mostly familiar, the mind can focus on the few that are unusual, which makes the music seem interesting - different enough to avoid boredom, similar enough not to overpower the processing capacity.
There's a diversion about dining habits, which are some of the strongest and most primitive. Generally, the way to get people to adopt a new food is through similarity and diversion. In wartime, people gradually mixed chicory into their coffee and included organ meats in their diet as part of a meat loaf, slowly increasing the percentage until people were drinking chicory tea and eating organ meat. Vegetarians attempt to lure omnivores to their diet by means of an array of products that attempt to be meat-like: tofu hotdogs and bean burgers. Abrupt changes have never been successful.
But there is also evidence to the contrary - that if there is a cacophony in which there seems to be no logic at all, new products can be slipped into the mix. One example is the way that cereals seem to be shelved at random - not in alphabetical order or grouped in similar clusters. This facilitates the introduction of new brands because there is no "order" to be violated by the addition of a new product.
At the end, there's a bit about consumer opinion versus consumer behavior. When asked what they valued in a gym, most members mentioned new and up-to-date equipment, a sufficient number of machines to accommodate demand, and the cleanliness of the facility. But when these things were addressed, there was no appreciable change in visit frequency. What was found by observational studies is that going to the gym is a social event - and having employees greet members by name and setting up exercise groups/classes had a much higher impact on visit frequency than any of the factors that were ranked highest in the survey.