So why are the ATM fees rising?

atm-You likely do not know it or see it but for those who use ATM that is not their bank’s, because they need cash for transactions the fees are approaching $5.  The Journal reports possible reasons why this is happening. I would like you to consider these and what questions you would pose.



That is overall fee revenues are shrinking due to regulations and there is a need to make up for that. Number of times people visit ATMs are down. And more people find ways to avoid paying fees. Left with the need to make up revenue targets from fees and facing higher cost of operations the banks need to allocate these costs and revenue expectation across fewer customer base in the form of higher prices.

Do they need to?

A moment’s thought should show you the problems with this reasoning.

As 75% of customers find a way to avoid fees – be it to not use cash, get the bank card, get cash in advance from bank branches – those who are left are clearly those

  1. Who cannot afford the bank card because of other fees and credit history prevent them from it
  2. Who cannot visit branches to plan ahead – that is these are likely the very people who have a high opportunity cost, lack transportation to visit banks or spend money in places where the bank won’t open a branch to do business.
  3. Who need cash and cash alone for the purchase occasions

In other words, the informed and high value customers stepped back and the ones left standing in front identified themselves as those who have no option but to pay higher prices.

Yes this is a kind of price discrimination.  Is this good or bad? It can be argued that if not for the service offered, the said customers will have no other option to turn to for for their transactions. From that view it is likely good price discrimination.

But the challenge I have in declaring this outright good price discrimination is its origins. Banks see fees as a major revenue source – fees from bounced checks, maintaining accounts, sending paper statements etc. That revenue stream is down to $34.1 billion from its highs of $41.5 billion. Since other fee avenues are shut down they see ATM fees as knob to turn. Why see fees as revenue stream at all?

Shouldn’t pricing be tied to net new value create for customers? Fees are ways to extract value from those who do not have options than as share of value created. After all business model is creating net new value and finding a fair way to share that with customer. Pricing is a way to get your fair share of the value created.

ATM fees are rising not because of costs but because banks have outdated business model. This will be disrupted.

Version Design – How to keep customers from being tempted by your lowest priced version?

Here is what a cleaning service offered me as my options

  1. Weekly cleaning schedule at fixed time at my convenience,  $60/per visit
  2. Biweekly cleaning schedule at fixed time from available choice of time slots, $90/per visit
  3. Monthly cleaning schedule, no fixed time, will call days before to set day and time, $130/ per visit

I am not going to reach to big conclusion here on carefully designed price discrimination and versioning design that we all can learn from. It could be that case but it is more likely done due to some operational constraints.

But the bigger point is, if you offer multiple versions of your product you need to understand first what your different customers value and setting a price based on this value distribution. Value to customers comes not just from the product but from many related factors like easy of buying, convenience, ease of consumption etc.

It may very well be the same product but you sure can offer multiple versions based on these other dimensions.

Now if only the dry-cleaners switch from price discriminating on gender to offering versions based on convenience, time to clean and pickup options.

You go into Apple store and they charge you a price between $499 and $929

ipad_air_pricesJimmy Kimmel, in his Late Night Show on Halloween, was making fun of iPad Air pricing. It went something like this (you can Google the video),

“iPad Air costs from $499 to $929. You go in the Apple store and they randomly charge you a price between $499 and $929”

Isn’t that close to a profit maximizing marketer’s Holy Grail? People walk into the store and each one pays a price that is specific to them – their price at which they are happy to get the product over keeping the money.

Granted we would like unbounded upper limit – not constrained at $929 and we do not want it to be a random price, we want each customer to pay precisely the exact price they are happy to pay to get the product.

That is impossible in practice and likely is illegal. You wouldn’t like it that I get to pay only $199 for iPad Air (because I am far too happy with my iPad2) and you have to pay $879 because you just pulled up in your Tesla.

I also will not like it when they pick a random number  and ask me to pay that price. Except of course if it is like a Becker-Degroot-Marschak method where they ask me to first write down my price between$499 and $929, then they pick a random number between these two.

If the price I wrote down is higher than the random number they picked, I get the iPad Air at the number they picked, otherwise I walk out losing my chance to be “awesome” and damned to eternity in Luddite world.

Since forcibly charging different prices for different customers or randomly asking them to pay different prices is not possible, Apple has the next best thing in pricing – making customers willingly self-select themselves to pay the price they want by choosing the particular product version they want.

That is why Apple offers 8 different iPad Air versions to spread the price spectrum from $499 to $929. In some sense Jimmy Kimmel is not all wrong in saying, “randomly pay”, you could argue arrival rate of customers is random and the version they will like and buy is also random. Only thing is customers willingly pay that price and the marketer enables this with multiple price points.

Do not for a second think the prices are different because of the cost. You do not believe it costs Apple $300 or even half that to go from 16GB to 128GB do you? Compare this to cost of going from hard drive to flash in non-Retina MacBook Pro and you will understand.

The prices are different because we are different and we all value products differently and willingly pay higher prices than others for the same product.

That is just versioning done right.

In fact the price spectrum for iPad is wider than $429 to $929.

If you include iPad2 it is $399 to $929.

If you include iPad mini and mini Retina is is  $299 to $929.

There is no one price to rule them all. You build and deliver products at multiple price points to let customers choose the price they want to pay.

Before you go and introduce versions at dozen different price points, see also 4 costs of versioning.


The most beautiful price fence

Fence a3

Fences are never beautiful. May be the picket fences are. But when designed to keep two sides from moving easily from one side to the other they are not usually described as beautiful. Price fences are as bad as say a fence between two countries. But they do serve the same purpose – to keep the two sides apart. Here is how a research article from 2009 defines price fencing,

market segments should be kept separate to prevent demand spillover from high priced segments to low priced segments and the associated revenue loss. Tools to restrict customer migration across segments are referred to as ‘fences’

 Price fences are key components of segmentation and revenue management. They are designed such that those who can afford and willing to pay higher prices are not tempted by the lower priced versions.

Let us take an extreme example from the early days of railroad transportation. Railroads are a high fixed cost business and the marginal cost of adding one more passenger was practically zero. They could have set the ticket price really low to fill every seat but that would be not capturing value from those willing to pay higher prices.

So they offered classes of service – small number of super premium first class service, moderately priced second class service and really low priced third class service. To prevent those who can afford second class service from being tempted by the low price of the third class service railroad operators removed roofs from the third class cars.

That price fence was not beautiful.

But here is some really beautiful price fence – comes from your favorite brand that excels in product design, Apple.


It is three different price points with right mix of features so carefully selected to let those who can and willing to pay higher prices from choosing the cheaper version. You may not see how impervious the fence is as you admire the beauty of the MacBook Pro. Let us dig deeper.

Between the $1,299 and $1,499 versions the differences are only in the RAM and flash capacity. Say you like the $1,299 version but just need more flash capacity. They are designed such that those with higher willingness to pay will choose the $1,499 version and pay the $200 extra.

You want the lowest priced version and try to customize your MBP with higher flash capacity. But guess what? There is no option available to increase just the flash capacity of your MBP. You can increase your RAM from 4GB to 8GB (the same level as the two versions on the right) for $100 more but cannot do that for flash  capacity. It is not hard for you to see that if RAM difference is priced at $100 the disk difference should also be priced and offered at $100.

Do not think this is a technical challenge. It is not, and it is offered as an option for another MacBook Pro – the non Retina version.

flash-upgradeThe MacBook Pro without Retina ships with hard-drive (the spinning platters kind) and if you want to customize it with SSD you would pay $200 more for 128GB and $400 more for $256GB. That is they want those customers to pay $200 for the same 128GB to 256GB upgrade.  So offering  just flash upgrade for Retina version  (for $100 as we saw above) would pose challenge to that $200 extra they charge for non-retina MacBook Pro version.

To state in simple terms, Apple’s price fences are not some isolated chain links but an integrated system of impenetrable walls that are passable only if you are willing to pay the same price where ever you decide to cross the fence.

They want $200 additional price  and they make sure they do it with price fences.

And go ahead and try to tell me that is not the most beautiful price fence you have seen.


Snail Slime Facial Gender Based Price Discrimination

Let us recap the news item. A spa in Japan offers Live Snail Facials – For $350 you can have five snails slink all across your face. It is a painful process even though it lasts only five minutes.

Who is the customer? Almost all women. Here is what the spa says about catering to men,

The service attracts a mostly female clientele in their 30s to 50s, and although men can also opt for the live snail treatment, none have done so yet. “I’ve observed that women will endure more pain and fright than men, if it’s for beauty,”

Let us set aside the fact that this non-scalable , one customer per day, service is not really the product. Let us ask, if this were a real product, if the spa would like to expand its market to men, what should it do?

Say men have less tolerance for pain when it comes to beauty. Could the spa offer lower price to men to help slime grease the wheels? That is overt and egregious price discrimination. A no no!

While price discrimination is good, gender based price discrimination is NOT. So how can it put to practice an effective and fair price discrimination?

Two options – none of these need any gimmicks – both are tried and tested and rooted in economic principles

  1. Bundling – Offer couples package
    Remember bundled pricing works when customer preferences are negatively correlated. Set a price that represents the sum of the higher and lower value perceptions of women and men. Say a $550 couples package could work. The man would think his cost is $200 while the woman would think her cost came down from $350 to $275.
    Besides the couples package will help drive more customers as women bring their partners, like this Candy store CEO says.
    Sure you give some consumer surplus in cases where both people value it at full price. As long as that lost revenue is smaller than what is gained from serving more customers with the bundle you should do it.
    For complete fairness, you should offer the couples package to any couple.
  2. More Product Versions
    Create an express version – 2 minutes for $200. That helps assuage men’s concern for pain and fright. And the lower price is attractive too. Brand it to signal something other than the best so women are less likely to choose it.
    As long as you are creating versions, create another premium version, there exist customer segments willing to pay more and you should capture that.

Both these options implemented together would work as well.

There you have it – Live Snail Price Discrimination done fair and square.

If you are going to charge for content

If you are you going to charge for content it pays to start to with the customer segments that value your content (more than other alternatives available) and are willing to pay for that value. Recently Mathew Ingram of GigaOm wrote about a proposal that involves dynamic pricing

A model could work like this: The piece costs $1.99 for the first 5,000 articles sold, garnering $10,000 in revenue [and] once that threshold hits, the price adjusts dynamically to maintain at least $10,000 in overall revenue, but adjusting downward against the paying population as more and more readers commit (which also earns Esquire additional advertising revenue). A ‘clearing price’ is set, perhaps at 50 cents, after which all profits go to Esquire.

Let me tell you at the outset that I do not like this for one primary reason – it ties price to your revenue goals vs. value to customers. That is as bad as cost based pricing. I does not merit further discussion just for that reason but let me dissect it anyway.

For the sake of this article let me call this proposal as Group Buying Pricing.

This is not new. In fact it is almost hard to find anything new in pricing. Recently a variation of Matthew’s proposal was studied by researchers from Kellogg School of Management. And that method itself is a variation of really old Dutch Auction. Kellogg study called their pricing Purple Pricing. This fixes two flaws in Matthew’s proposal,

  1. Those who wait (till others buy) for the price to drop (sideline issue)
  2. Those who paid higher price than those who came behind them (fairness issue)

Purple pricing works like this

  1. Start with limited number of units to sell or set a time after which the product has no value
  2. Start with a set high price and periodically drop the price
  3. Anyone who buys the product at a given price is guaranteed they will only have to pay the lowest price the last unit is sold for

So if you like the product at a given price you should pay. You may not have liked it at $2.99 but when it drops to $1.99 you might find it attractive. You should agree to pay that price. The scheme guarantees you will not pay more than the lowest price someone else who buys the product after you pays. If the last price is $0.39, that is the price you will pay even if you bought in at $2.99. So it does not suffer from sideline or fairness issue.

In many ways a magazine fits well for Purple Pricing. While there is no limit on number of digital copies, its relevance is lost after the week. So Purple pricing will work and it is not operationally any more complex.

In addition Purple Pricing helps us find the distribution of customer willingness to pay (and hence the demand schedule) while Mathew’s Group Buying Pricing says nothing on that front. Customers have no reason to reveal their true willingness to pay with Group Buying Pricing.

If you are asking, “doesn’t that compromise revenue goals?”, you should recognize that you will not be dropping prices if what you get from new sales is not enough to lose the price premium you captured from all the previous buyers. In fact the proponents of Purple Pricing recommend that only as a method to find the demand schedule and use it to set a price that maximizes profit.

Both schemes do suffer from one common problem – What if the value for the content increases as more people read it?

As long as we are coming up with proposals for pricing for content let me also tell you about another old method – Becker-DeGroot-Marschack (abstract only) method. Here is how you do it

  1. At the time customer tries to read the magazine ask them to quote a price they are willing to pay – give them a range low and high
  2. Let them know right there that you will pick a random number between low and high right after they state their price
  3. If the price they quoted is greater than the random number you picked they only have to pay the lower number you picked and happily read the magazine.
  4. If the price they quoted is lesser than the number you picked, they don’t get to read (I assume your engineers can figure out how to enforce it so that people don’t try many times until they get lower price)

It is in everyone’s best interest to quote the price they are willing to pay – not high-ball (that would be silly) not low-ball (No magazine for you!). And you get to find the demand distribution as well.

All is well. But imagine all the complexity in the buying process when a customer goes to read a magazine. Think of all the cognitive costs. You surely do not want this to be the buying experience every time someone wants to read a magazine do you?

By all means do pricing experiments but use those experiments only to find the demand distribution. Then set a price that will maximize your profit and make it very easy for your customers to do business with you.