Price Optimization Software Big in Retail Business 121
prostoalex writes "Even if you spent only a single day in an economics class, you're probably familiar with a concept of supply and demand. The Associated Press is running an article on retailers employing mathematical models for price optimization, where some products are priced higher to generate higher margins, and some are discounted to generate larger volumes even at the expense of per-product margins. DemandTec, Oracle and SAP are some of the companies producing those mathematical models for retailers around the country, with AP listing some of the pricing optimizations employed currently."
Quick - someone patent it ... (Score:5, Insightful)
Seriously - this is NOT new. Not even in the software field.
Re:Quick - someone patent it ... (Score:5, Interesting)
Re:Quick - someone patent it ... (Score:5, Insightful)
I would also look for the best deal, but only if in doing so I'd save more per hour than I would make if I were working instead. Personally, I don't enjoy wasting my time running from store to store. Even if I'd save 50 dollars, it probably won't be worth it if that meant shopping 8 hours to find the cheapest store.
Just my two cents.
Re:Quick - someone patent it ... (Score:4, Interesting)
If I happen to be shopping at one of the other places, it is then easy to see if their current sale price beats Wal-Mart's everyday price. If so, and I need the item, or can reasonably store it, I will buy at the other store. If I notice an advertised sale on something at one of the other stores, I might go there for it.
Re: (Score:2)
The extra $15 a week I spend on groceries at Safeway is well worth the cost of not being around people who look like they want to marry their cousins, teenagers who think they're gangbangers, and 16 year old single moms. The Safeway is also cleaner and the staff friendlier.
Re: (Score:2)
Wal*Mart's low prices (Score:4, Interesting)
Never buy produce or fresh bakery goods at a Wal*Mart. The premium at the true grocery stores often corresponds to the produce & bakery goods actually being better quality.
Also, since you actually have a choice, try to memorize the routine sale prices at your other local stores. Sales tend to be cyclical. Wal*Mart has lower prices on the most popular items; for more obscure stuff they can go higher because those items are harder to find elsewhere, or fewer people are looking for them. I learned this when trying to buy a rare iron syrup (which could've had a proof number).
Wal*Mart is a good place to shop for low prices, but other places have different selection, and it's a good idea to give at least token support to its competition.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re:Quick - someone patent it ... (Score:5, Interesting)
Re: (Score:2)
Re: (Score:2)
I don't know what a coupon discount is I'm afraid .
Re: (Score:3, Informative)
Yes, I am a supply chain consultant...
Re: (Score:2)
Re:Quick - someone patent it ... (Score:4, Interesting)
A specialty shop with unique products can just slap a large margin on a product as long as the price is fairly well justified and do fine, but when catering to a clientèle working with higher volumes, low margins and aggressively price conscious consumers then wringing every last possible penny out of them works.
Everyone knows that by and large in the U.S. that Wal-Mart is the price discount king. And even if you are really watching prices and comparing a consumer will still believe that they are getting the best deal at Wal-Mart. But I know that a significant number of their products are name brands packaged into their own unique size. Of course at Sam's Club things have to be bought in bulk for a discount but comparing product X at Wal-Mart in a 13oz size verses the same product X at a local competitor in a 20oz size can appear to be the exact same product for significantly less than a competitor although the price per ounce is a better deal at the competitor.
I sell pet food and I'll say that Ol' Roy is one of the cheapest priced pet foods per bag. Sure there are some others cheaper per bag at feed stores which may or may not be a better deal. I know some feed stores that use loss leaders to beat Ol' Roy. But, when comparing Ol' Roy to premium pet food the price per feeding will beat Ol' Roy most every time. So, buying a $15 bag of feed per week is more expensive than buying a $26 bag of feed per month.
Re: (Score:3, Insightful)
"I sell pet food and I'll say that Ol' Roy is one of the cheapest priced pet foods per bag. Sure there are some others cheaper per bag at feed stores which may or may not be a better deal. I know some feed stores that use loss leaders to beat Ol' Roy. But, when comparing Ol' Roy to premium pet food the price per feeding will beat Ol' Roy most every time. So, buying a $15 bag of feed per week is more expensive than buying a $26 bag of feed per month."
It depends on which "Ol' Roy." If you buy the cheaper,
Re:Quick-someone patent it ...Airlines not WalMart (Score:2, Interesting)
A MUCH better example would be airlines; where what is practically the same product - a coach seat on the same flight might go for 5 to 8 price-points of $1
Re: (Score:2)
Almost but not quite the same (Score:2)
I thought their nutella came in a different size jar from everyone else's...
I shop at a lot of stores that do this. W
Yes, there are new things (Score:5, Interesting)
First, a disclaimer. I was employee #4 at KhiMetrics, the company founded by Ken and Tim Ouimet (employees #1 and #2). They're mentioned in TFA. SAP bought KhiMetrics in January of 2006. Ken had been my office-mate in grad school. That said, I haven't seen Ken and Tim in years, and I have no financial stake in KhiMetrics or SAP anymore (SAP bought out the KhiMetrics stockholders with money, not shares of SAP stock).
Yes, it's true that humans doing pricing try to do the same things. But the thing is that software can do things a human mind cannot. Yes, the opposite is also true, but here software has a lot of advantages. In the case of the KhiMetrics (now SAP) software, it works on the category level, optimizing profit for the category as a whole, which can include taking losses on individual items. The software never makes the common mistakes human beings make. For example, different "flavors" of the same size package of the same product should come out at the same price, and the unit price of a given item should go down as the amount bought increases. I can tell you that I have seen examples where humans have screwed this up this week. When there are two sizes of a given product, let's say a certain laundry detergent, then the price per weight of the larger package better be less than the price per weight of the smaller package, or there's never any incentive for the customer to buy the larger package. Still, I see examples where the pricers have gotten this wrong. I've even asked people at the stores if they were trying to move the smaller packages because of having too much of that size in stock or something, and they told me that no, they had no such problem.
The other thing is that the KhiMetrics software uses actual sales data to determine how sensitive the customers are to the price of a given product. This can be done down to the SKU (individual item) level in the product dimension and down to the level of customers of a specific store in the geographic dimension. In other words, the KhiMetrics software is capable of determining the sensitivity of the customers of each individual store to the price of a specific product. No human being could do that at all, much less in the time the KhiMetrics software can do it. Even with a pricing team for each category in each store, which would end up costing a fortune in human resource costs, the result would not be as good as what KhiMetrics can deliver. Additionally, since the Ouimets "grew up in retail," the KhiMetrics software, since the beginning, has been compatible with things like Category Management and Efficient Replenishment, and able to take into account things like having different goals for different products in a category (loss leader, profit generator, traffic generator, etc.). The software takes into account complex factors like seasonality, promotion, and product visibility. Since I have a reasonably good idea of the internal workings of the software, I can tell you with some confidence that I, a Ph.D. in theoretical physics, would not even want to try to tackle the problem of optimizing the prices for a subcategory of 20 products in a single store, much less the dozens of categories and tens of thousands of SKUs in the dozens of stores in a retail chain. KhiMetrics can do all that, basing itself on years of actual sales data, before breakfast.
There are experienced people in retail who are good at such things, but the software was created with people that have the same level of understanding of retail pricing, plus it has all the advantages of being able to do high-speed computerized analysis of huge amounts of price and sales data. I don't work for KhiMetrics anymore, nor for SAP, but I can say that if I were working in a retail company, I would definitely want us to be using software like this for pricing. And experienced retail people agree with me. One thing we saw back when I was with K
Re: (Score:3, Insightful)
"The software never makes the common mistakes human beings make. For example, different "flavors" of the same size package of the same product should come out at the same price, and the unit price of a given item should go down as the amount bought increases. I can tell you that I have seen examples where humans have screwed this up this week. When there are two sizes of a given product, let's say a certain laundry detergent, then the price per weight of the larger package better be less than the price per
Re: (Score:2)
That's a good point. Also, as I mentioned in the grandparent post, sometimes a retailer might do that in order to move the smaller ones if they had too many units of that size in stock. But what's really cool is that the software can actu
Re: (Score:2)
The problem with that, of course, is going into a different store from the same chain and finding out that your usual store is ripping you off on certain items.
This end up hurting "the brand", so franchisors won't do this at the store level, but at the chain level. Several chains tried doing this in the 60s by charging more to customers in the slums than in the suburbs, where there was more competition. Word of the practice got around, and people started voting with their feet - making the trek to the 'b
Re: (Score:2)
"can afford to buy a year's worth at a time" perhaps. A years worth of mustard is great, unless all you get to eat all year is mustard. Like I should talk though; after spending maybe, oh, total 45$ this year for groceries total in the past two days I splurged and just spent 30$ on food(some of it was better than sex though, and I don't regret it, although I'm not going back for another meal there, ever, due to the cost).
What about small businesses? (Score:2)
As an example - my girlfriend is an optometrist with her own, single practise selling glasses, sunglasses and contact lenses. They don't sell anything in great volume and generally just make up prices as they go along.
She would love to have software like this to give her some pricing guidance, but I personally don't see how it could work effectively without a lot of sales data to crunc
How does it handle ... (Score:2)
What is the critical mass of sales data required to get good results? Has there been any effort or success in aggregating data from a number of small retailers to give them the results to match the information power of the larger retailers? Obviously, there would have to be secure handling of the data so the risk of letting out your sales data would be outweighed by the rewards of the improved pricing data.
I'm also w
Re: (Score:2)
The technology is that good. Retailers have reported huge gains in both revenue and profits from im
Re: (Score:2)
Sell at a loss... (Score:3, Funny)
Re: (Score:1)
Re: (Score:1, Interesting)
Actually that is a price optimization software I keep expecting to be put in use secretly by Walmart's competition. Walmart has "the guaranteed lowest prices every day", but only if you challenge them on it with an advertisement for lower price then they have listed. Most of the time when you do this they will match the lower price on same brand and often will on a store brand switch in for grocery store private labels advertised prices. Walmart's however does not i
before all the "duh" responses (Score:3, Informative)
Re: (Score:2, Insightful)
Interested parties will consider entering the market and driving down the cost, that's what. Or maybe consider selling different versions (more price discrimination)
Re: (Score:1)
The is basically "Cost Accounting". As the first poster pointed out, this has been done for centuries. Using computer models will make it better and/or faster, but that is the same reason we use computers for just about any computationally boring process.
Re: (Score:2)
Exactly, though I'd argue that even straight economics has a large behavioral component. What this does is let organizations maximize profits by tuning price to de
Re: (Score:1)
To expand the LCD TV example used above, let's say the original, static price was $700 (with a cost of $500). Before price discrimination, they make one sale to Mr. MBA and receive $20
Re: (Score:2)
Typically, people only have a limited tolerance for "perceived value", especially on commodities. One might occasionally go to the corner market and pay 50% more an equivalent product, but if people h
Perceived value (Score:2)
Why this is good for everyone (Score:5, Interesting)
The amusing fact is that this is nothing more than a capitalist version of taking from the rich (those are willing and able to pay more) and giving to the poor (those aren't willing or able to pay more).
Re:Why this is good for everyone (Score:5, Insightful)
It's called coupons!
The product is priced on the shelf at the price most consumers are willing to pay (say, about 60%). The coupon discounts the product to a price the other 40% are willing to pay. Now you get to charge two prices for the same product! Woohoo!
Re: (Score:2)
Re: (Score:2, Interesting)
We aggressively pushed store brands, the profits were much greater.
Re:Why this is good for everyone (Score:4, Insightful)
Re: (Score:2, Informative)
Re: (Score:2, Insightful)
Re: (Score:1)
Re: (Score:1)
p2 CS = p2 - p1
p2 == p1 => CS = 0
p2 > p1 => no transaction, no consumer surplus.
Perfect prize discrimination means p2 == p1 for everyone. SUM CS = 0
Re: (Score:1)
The first line is supposed to read p2 > p1 => CS = p2 - p1
price optimization vs. market segmentation (Score:4, Insightful)
The classic example, that I was thinking of, is the revenue management strategies of airlines that attempt to optimize prices across presumed underlying segments of price-sensitive leisure travelers versus price-insensitive business travelers. Technically, it's the identical seat that's being sold for radically different prices (up to 10X different) depending on issues such as how the customer buys the ticket, when they buy the ticket, whether they book a saturday-night stay, etc. The result is that the business customers pay for the plane and the vacation travelers only pay for the fuel and variable costs. The ability to differentiate does benefit the business customers because the added volume of travelers means a more frequent schedule of flights, larger planes, and more destinations.
You are right, though, that true market segmentation involves much more than just price optimization.
Re:Why this is good for everyone (Score:4, Insightful)
Re: (Score:2)
Re:Why this is good for everyone (Score:4, Insightful)
Some people might think you are kidding there, but there really are cases where everybody wins from differential pricing, and businesses really do take from the rich so they can afford to sell to the poor. Let me add an example to make it clearer:
Imagine you want to build some sort of clever new software. You see that 10,000 people would pay $100 for it (as a fun toy, say), and 5 companies would pay $1m for it, because they can each make $3m from using it commercially.
If the software costs $1m to develop and you sell all the copies for $100, your profit will be $500. Nobody's going to go to all that trouble for $500, so you wouldn't make the software. And if you did, you'd be steamed that these companies made millions while you got pocket lint.
However, if you sell the first 5 copies for $1m each (with, say, some fancy documentation and a support contract), you can then go on to release a consumer version to get everybody else. You get $5m in the bank, so you're happy. The companies netted $10m, so they're happy. And everybody else got a fun toy at a reasonable price.
Note that although your average price per copy there is $600, you couldn't get the same effect by charging $600; none of the consumers would pay that much for a toy.
Re: (Score:2)
Your reasoning only applies to a non-free market.
In a truly free market the people who bought the software for $100 would be able to on-sell it to the business for $110, meaning the original vendor would be out of a sale.
The fancy documentation and support contract is a different product that some companies are willing to pay more for. It's also bundling. It's not differential pricing on the same product.
I'd like to see both differential pricing and all on-selling restrictions made illegal. This would
Re: (Score:2)
How would that be more optimal? The result would be that the hypothetical company sells the software for $1 million only, and the 10,000 other potential customers never get to see it.
Yes, but... (Score:1)
Re: (Score:1)
The amusing fact is that this is nothing more than a capitalist version of taking from the rich (those are willing and able to pay more) and giving to the poor (those aren't willing or able to pay more).
No, it's getting the maximum amount of money from each customer. They're not giving anybody anything - they're charging the rich more.
As a side benefit, it can make a product more profitable and promote more development, so I guess the poor benefit from that... Kinda like people who buy version 1 of $NewStuff. (As a Version 3+ guy, thanks!)
Re: (Score:2)
One example was a store that had a line of paintbrushes and was selling too many of the cheapest paintbrushes. Solution for store? Discontinue the cheapest paintbrush, since the customers tend to buy the paintbrushes at the same time as the paint and therefore won't shop around.
This benefits no one but the store.
Re: (Score:2)
This benefits no one but the store. ''
The problem is that customers are not stupid, and eventually they realise that there is likely a reason why you can't get cheap paintbrushes at this store. So either they just realise that thi
Re: (Score:2)
Re: (Score:2)
(You mathematically prove that this increases what is called consumer surplus which is the equivalent to the consumers "profit" on the purchase and the seller's profit)
This cannot be true. Take it to the limit, where the vendor can set a different price for every customer. If it's a monopoly they set the price to a tiny fraction less than whatever it's worth to the customer and all customer surplus is close to zero. If it's a competitive market with no price fixing going on then they must set the price
Re: (Score:2)
http://web.mit.edu/14.271/www/hio-pdic.pdf [mit.edu]
game the system (Score:4, Interesting)
There was a quite fascinating article published the other day in a Digg linked blog that I am sure many here read (I don't have the link unfortunately). What is really interesting is that by knowing the system and subverting it you can make HUGE savings in your shopping. The layout of the store is carefully crafted to expose you to the products they want to push. Color schemes and shelf placememt are designed to confuse or lead you to select certain products. Prices and product sizes are carefully designed to make comparative math very difficult to ordinary folks. Bargains are placed outside the normal lines of sight.
In other words, the very existence of a cold and calculated system is what enables you to game it.
Some bits of strategy I remember:
1) Make a list and stick to it. Impulse purchases account for a huge amount of profit and the stores rely on you buying things you do not need.
2) Never look at the products at eye level, they are the most expensive and worst value.
3) Move as fast as you can to the back of the store. Start at the back of the store and work your way forwards.
4) Do not stop unncecessarily. Deliberate impedences are put in isles to slow you down.
5) Don't take a cart or basket unless you really can't carry what is written on your list.
6) Use the bathroom before you go shopping. They place the restrooms to make you walk as far as possible past tempting impulse products.
A couple of my own...
7) Eat before you shop, never go to a grocery supermarket when hungry.
8) Take cash, just as much as you need and no more, and use the cash only fast checkout.
Perhaps someone who knows the systems they use in detail should write a piece of open source software in their spare time to calculate the optimal path through a store
Re:game the system (Score:5, Insightful)
That is really not a valid statement, for a couple reasons. The first error is the last two words 'worst value'. Only the customer can determine what the value is. If I'm looking at a condiment section and at eye level is a name brand catsup, and below it is a private label equivalent, the price per unit will probably be lower on the private label. That doesn't necessarily make it a better value. If I think the private label tastes rotten and wont eat it, the more expensive catsup is a much better value.
The second issue is that quite often what you see at eye level is determined by who payed for the placement. It may not be the highest margin item for the retailer on its own. But it is their because the vendor payed a royalty to have it where they want it.
3) Move as fast as you can to the back of the store. Start at the back of the store and work your way forwards.
That doesn't make a whole lot of sense. If you are talking grocery, very few stores are laid out the same way. There's no way this can be a 'rule' that will help you when what is at the front or back will vary from location to location. I think a better way of looking at this might be - don't buy what is on end-caps and floor displays until you have looked at the prices for comparable items. This means, not running to the back, but going to the aisle where the item is normally located.
In larger stores this really doesn't make sense. If I go to Fry's Electronics and run to the back, how does that help me? If I go to Best Buy and hustle right back to home appliances, I'm not sure what I've done to help myself out.
The psychology of all this is over rated. A little common sense - like many of the other suggestions in the list, will go a long way. That's not manipulating the 'system' it's just using your mind and operating above a visceral level.
Re: (Score:3, Insightful)
You assume that the vendor always charges the same price. Often, the vendor will pay for good shelf space and lower the price. It's called a promotion.
The vendor uses this as an investment to encourage new customers to try the product or people who've forgotten about the product to start purchasing it again. Some of those people then begin to like the product and purchase it when or where the promotion is not available (perhap
A better way to game the system... (Score:2, Interesting)
Re: (Score:2)
Of course, that metro area has several related local grocery chains. Do they count?
And a few suggestions of my own... (Score:2)
A lot of stores, such as Safeway for example, include the unit price on their shelf tags. This is very convenient when comparing the cost among brands
Re: (Score:1)
You clearly have never been to the grocery store while high.
Re: (Score:2)
Re: (Score:2)
Luckily, all the grocery stores that I frequent have the price/oz printed on the price cards on the shelf edges. I wonder why they'd do that if they want to make it hard for shoppers to compare? Could it be that the advantage of making customers more willing to shop at their store (vs. a competitor) outweighs the disadvantage of being less able to sneak in poor deals? Or do, perhaps, the lower priced s
Re:someone here reads digg (Score:1, Interesting)
Re: (Score:2)
Re: (Score:2)
That's not my experience. Even if it was, I'm willing to jump to the credit lines if the line is too long anyway. If they want to pay that extra ~3% fee to take a charge after I've already made my purchase choices whereas taking cash with a couple anti-counterfeit measures costs a lot less, then that's their prerogative.
Re: (Score:2)
Most grocery stores have express lanes for n items or less. Sometimes they're faster, sometimes they're slower (no conveyer belt & too many people with less than n items), and these days they're often self-serve. But nobody has ever stopped me from using a normal aisle for less than n items, even when the express lane is open.
Worst value? Can't be right. (Score:1)
2) Never look at the products at eye level, they are the most expensive and worst value.
If that were the case, shouldn't candy, cookies, ice cream, and action figures be spread out over the entire store on shelves about 2-3 feet from the ground?
Re: (Score:2)
The candy is in two sections, the checkouts and an in-store shelf; the checkout candy will usually be a "worse value" than the in-store candy, and the candy that's easily visible will be a "worse value" than the candy that's buried in the skinny shelves below eye level.
Cookies have their own section for the most part, but the "worst values" will be at eye level. Below eye-level, there are no cookies--only crackers.
Ice cream will have its own sectio
Paco Underhill (Score:1)
A cruise thru "The Power of Persuasion" by Robert Levine
might also be enlightning.
Interesting ... (Score:5, Funny)
So what is new? (Score:3, Informative)
I have used a prize optimization solutions based of MS SQL Server back in 2005
http://www.microsoft.com/industry/retail/solution
Watch out for customer "value" cards (Score:3, Informative)
In an idea scenario, they set all prices on the high end - but then give the customers "value" cards that offer varing discounts and rewards so as to optimize sales and profit. For example, if they know you won't pay more than $2 for a soda, then your soda will always be $2. For example, they might do something like use buying habits track your period. If you buy tampons on the day of your period - you will get reamed hard because they know you need them right now, but if you buy them in the off cycle then you get a good deal. If you buy just milk in the early morning, you will get reamed hard because they know you might need it for breakfast right then, but if you buy it later on you will get a competitive discount. If you buy a phone today, but the last phone you bought was two years ago and had a two year average lifespan, then you get reamed hard because they know you need a replacement right now. Otherwise you get a deal. If you buy condoms on friday night, you get a nailed hard, but if you buy them on wednesday morning you get a great deal.
Re: (Score:1, Insightful)
In fact, the scenario you described with the cards is exactly the same one you said customers wouldn't tolerate. It is also illegal in many places.
Re:Watch out for customer "value" cards (Score:4, Informative)
If the gallon of milk is marked $4.99 on the shelf, the customer who is unwilling to pay more than $3 is not going to put it in his cart. That the store plans to discount it to $2.99 at the register won't change that. And if the customer is willing to pick up the $4.99 milk, what incentive is there for the store to charge less for it?
Re: (Score:2)
When there is a lower price for people with value cards and a higher price for people without them, the store covers the area on the shelf listing the normal higher price with a large card giving the lower price (for people with value cards) in huge numbers.
People with value cards know what they'll pay immediately and will be attracted to the sale items.
People without value cards can look under the big card to find the real price.
Pe
Re: (Score:1)
Sounds like you're getting a great deal either way!
Re: (Score:2)
I'd say it's my girlfriend who get's nailed hard.
*DaaDum Crash! Thud.*
And if you spent a second day in that class... (Score:2)
Re: (Score:2, Insightful)
The real issues are getting the data and interpreting the results
I fail to see the point... (Score:2, Insightful)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Surely a store that prominently advertises "Rollbacks" is not using a flat price mark-up!
Re: (Score:2)
We were doing this years ago (Score:4, Interesting)
The original pricing theory in traditional auto parts stores was based on four "turns" per year: after opening a parts store and filling it up with stuff to sell, you needed to sell each stock number four times, at 35% gross profit, to make an adequate gross profit to cover your expenses -- and pay off your inventory in twelve months.
This resulted in some items being priced much higher than at mass retailers, and caused stores to lose sales on popular items: people would go elsewhere for oil, antifreeze, and the most common spark plugs, brake pads and filters, because they were much less expensive at places like Pep Boys.
The GROI method constantly recalculated sales and adjusted the prices downward on popular items, thus increasing sales and lowering the prices still further. For example, the best-selling oil filters would sell at under 10% gross profit, but we would sell out our inventory of those items twelve to fifteen times per year... thus making a larger profit on the initial purchases we had made to stock a new store. By setting a minimum gross profit percentage, runaway sales on an item would result in higher profits instead of ever-lower margins.
This was all calculated by an incredibly expensive 200 MHz Pentium Pro box, running proprietary software atop SCO Unix.
(I was the only one who could work the thing... which led to me running a newly purchased Netware 4.1 network in the chain's offices... which led me out of that filthy auto parts business altogether, thank Jeebus).
Does amazon use it? (Score:2)
Hardly new (Score:2)
Airlines have been doing this for years. And hotels, and other industries. Apartment complex management is getting into the game as well.
I have a friend who works for a rental car company, he does this sort of thing in his head at
And this is surprising? (Score:2)
Of course, if you're a company selling stuff, evolution favors the intelligently designed.
What comes around, goes around (Score:2)
I mention Target because they give me the creeps. I leave with the feeling that I'm never really getting a deal and every item of inventory and model of that item has been efficiently spreadsheeted for optimum pr
iWon spam site? (Score:2)
Who the hell are these people?
Re: (Score:3, Interesting)
Don't care. While the individual case can't be modeled the group case can be. Plus a WM can price a DVD at one price in LA and another in SF, seen the trend, reverse it next week, see what happens then and adjust the price in all of the stores accordingly. Do variations of the above for 1,900 stores nationwide. See what happens when you advertise price A vs. price B and seel how many are sold. Do variations of the above for 1,900 stores