See the attached file.
A small firm operating in the U.S. prepaid-phone-card industry is considering a new $5 phone card offering 60 minutes of service to Mexico. The sales group at the company believes the card will be a market success. The brand manager, skeptical of the card's potential profitability, has been assigned the task of analyzing its costs. Three carriers have offered different pricing proposals for the new card.
Any spreadsheet model to evaluate the potential profitability and relative merits of the three proposals and single-point estimates of numerous inputs.
TAURUS TELECOMMUNICATIONS CORPORATION:
A NEW PREPAID PHONE CARD
Kyle Shilling loathed sales meetings. A brand manager at Taurus Telecommunications
Corporation (TTC), Shilling felt certain the sales team's job description was simple: "to fail to
sell perfectly viable products." To him, these meetings were just forums for the Sales department
to make excuses for its lackluster results.
Worse, Peter Francona?TTC's chairman?actively participated in the meetings. An exbeverage
salesman, he seemed remarkably sensitive to the alleged plight of the salespeople. On
joining the firm three months ago (straight out of college), Shilling had appreciated the
opportunity to work closely with the CEO of such a large company. But, as of late, his
interactions with upper management had become cumbersome.
Today's meeting was proceeding as he had expected. With growing frustration, Shilling
listened to complaint after complaint about his new product line. Never known for his adherence
to protocol, Shilling eventually lost his temper.
"Great Odin's Raven!" Shilling shouted at the lead sales manager, interrupting a tirade on
the product line's lack of competitiveness in Long Island.
We've created the last six products exactly to your specifications, because you
argued the sales team's knowledge of the market exceeded ours. And yet, week
after week, all we hear in these meetings is how non-competitive our products
are! The last five-dollar card we released offers forty minutes to Mexico; at that
rate, we are severely straining our profit margins. Why can't you do your job?
"Sit down, Kyle," retorted the sales manager, whose official title was vice president of
The fact is our customers don't believe our advertised rates. And even if they did,
the competition just released a card offering fifty minutes to Mexico. What we
need is a card that offers sixty minutes. We could sell a card with sixty minutes.
The current product is just unsellable.
This was the kind of insularity that drove Shilling batty. Current costs for a call to
Mexico were around $0.0845 a minute. At that rate, the company was very close to losing money
on every 40-minute card they sold (the common joke was that they'd make up any losses on
volume). Of course, Peter Francona would never accept Shilling's reasoning.
Just as expected, Francona chimed in: "Wow, sixty minutes. Now that's a great idea ...
we could sell a million of those cards a month. Kyle, why didn't you think of that?"
"Because we'd lose about a dollar on every card we sold, Peter."
"Kyle, you've got to think outside the box." Sadly, Francona was being serious and he
Don't worry about costs; we just got new pricing for Mexico from three different
carriers. It's pretty cool?they've given us some novel pricing offers. Run the
numbers, will you? I want an absolutely clean card, and I think we can get about
$5 million in revenue a month from this card.
Why Francona had waited until that moment to mention the new rates to Mexico was
beyond Shilling, but he would dutifully perform the cost analysis.
The Prepaid Phone-Card Industry and TTC
United States prepaid phone cards allowed users to place a call from any domestic
location to anywhere in the world. At the turn of the millennium, phone cards were particularly
popular among the immigrant populations, because phone cards generally offered cheaper
international rates than traditional service providers, and they eliminated the need for a home
telephone account, which many populations found unattainable due to poor or nonexistent credit.
In 2000, the prepaid phone-card market totaled about $3.4 billion in revenues. Ethnic minorities
and lower-income customers comprised roughly 85% of the market. Industry analysts expected
revenues to exceed $5 billion by 2005, primarily due to the opening of mainstream markets.1
Taurus Telecommunications Corporation had captured about $250 million of the total
market in 2000. The company found it difficult to maintain any single product for a significant
period of time. As a result, TTC's four brand managers constantly created new products and
adjusted old ones. Although the differences among these products were slight, the sheer number
of products?which was somewhere in the thousands?made management difficult. Mistakes
were quickly identified by the consumer population (and sometimes exploited; in one case, a
typographical error had resulted in the loss of over $100K in a single night).
1 Intelecard News printed 10/01/01, the online version at http://www.intelecard.com/
Prepaid Phone-Card Operations
The first key part of a phone card was its set of access numbers, which were printed on
the back of the physical card. A set of access numbers was unique to a product line, but not to a
single card. The user dialed one of the access numbers to connect to the phone card
manufacturer's switch (a telephone switch is simply a computer that handles telephone calls).
Manufacturers provided users with both toll-free access and local access numbers; local access
numbers were non-toll-free numbers with the area codes in which the card had been sold (e.g.,
212-578-7007 in New York).
The second key part of a phone card was its Personal Identification Number (PIN), which
was unique to each card. After the user dialed one of the card's access numbers, the switch
prompted the user to enter the PIN. The PIN was effectually the user's account; the switch
maintained a database of each PIN, the per-minute rates applicable to that PIN, and the dollar
balance remaining on the PIN. After the user entered his PIN, the switch checked that the PIN
existed in the database and had a positive balance.
The switch then prompted the user to enter the number he wished to call (known as the
destination number). The switch determined a per-minute rate for the call based on: (1) the PIN,
(2) the access number, and (3) the destination number. The switch then informed the user of the
maximum number of minutes the call could last, based on the PIN's remaining balance and the
per-minute rate. The switch then rang the destination number and, assuming the dialed party
answered, completed the connection.
Once the connection was completed, the switch began to decrement the balance of the
PIN based on the per-minute rate. If the call was terminated before the entire balance of the PIN
was exhausted, the remaining balance was stored in the switch's database; the user could re-use
the phone card until the balance ran down to zero.
The prepaid phone-card manufacturer (PPM) served as the middleperson for this entire
process. Some manufacturers maintained their own switches; nearly all contracted with other
firms to bring the call from the user's phone to the manufacturer's switch and to take the call
from the manufacturer's switch to the phone of the destination number.
The first part of the call was known as the "inbound leg"; the PPM contracted with an
inbound carrier. The manufacturer usually contracted with a single toll-free inbound carrier and
several different local inbound carriers (each corresponding to a different geographic location;
there was little competition among local inbound carriers). Local inbound agreements were
usually structured as flat fees for a certain amount of fixed capacity; if the capacity was fully
utilized, the per-minute rates tended to be far lower than those explicit in toll-free agreements.
The second part of the call was known as the "outbound leg"; the PPM contracted with
outbound carriers. A manufacturer contracted with dozens?if not hundreds?of outbound
carriers, as this leg comprised the majority of the expense of a call. Although there was
significant competition among outbound carriers, each carrier tended to specialize in a certain
region of the world (e.g., South America); some carriers specialized at the country level (e.g.,
Brazil). Often, this specialization represented the deregulation and liberalization of a country's
state-owned telecommunications operations.
The manufacturer's switch determined which outbound carrier to use based on a "routing
queue." The routing queue listed every materially different destination in the world, along with a
prioritized set of outbound carriers to use for each destination. Outbound carriers could only
handle a certain number of simultaneous calls?a maximum capacity. In the event the firstchoice
carrier's maximum capacity had been reached, the switch automatically routed new calls
to the second-choice carrier.
The determination of the routing queue was the key in cost optimization. A manufacturer
contracted with hundreds of carriers; each offered a different rate to each of the hundreds of
destinations worldwide. The simple method of using the least-cost route for each destination was
complicated by three factors: (1) scarcity of capacity for each carrier, (2) variability in quality,
and (3) payment terms.
Outbound carriers were often small operations; thus, they were rarely capable of meeting
a manufacturer's capacity needs at full-price optimization. The competitive environment among
outbound carriers generally created a positive correlation between carrier size and higher price.
Furthermore, outbound carriers were loath to commit all of their capacity to a single
manufacturer, due to a history of fly-by-night phone-card operations. Manufacturers did not want
to depend on a single outbound carrier, due to the similarly short life spans of many low-cost
firms, so capacity-per-carrier was severely limited. If a carrier provided a least-cost route for
more than a few destinations, the optimization algorithm had to take into account that carrier's
inability to handle traffic to all of those destinations.
Many low-cost outbound carriers were either running traffic through "gray routes" by
using connections that avoided government monitoring and the concomitant fees, or running
traffic over a decrepit telephony infrastructure. These factors contributed to significant variability
in the quality of an outbound carrier, which was measured by the clarity of the connection and
the length of the time a connection would remain open, etc. A wired telephone call was made
over an electrical circuit; poor wiring at any point along the circuit significantly interfered with
normal processing. Phone card users were willing to accept relatively poor quality, because perminute
rates were at such a discount to traditional carriers, and international infrastructures were
known to be poor. Nevertheless, low-cost carriers frequently fell below the threshold that
customers were willing to accept?or that manufacturers were willing to accept?and thus pure
cost optimization was often not of practical use.
Finally, outbound carriers required varying credit terms of their customers, spanning the
range from prepayment to net 60 or even 90 days. Massive power at lower points in the phonecard
distribution network often forced manufacturers into difficult cash flow situations; thus, the
manufacturer was sometimes forced to use a higher-cost carrier that offered delayed payment
These three variables complicated an optimization algorithm, but by no means made such
an algorithm impossible. Unfortunately, organizations were not sophisticated enough to develop
or implement a workable algorithm. Given the level of traffic handled by TTC, the inability to
develop an optimization scheme resulted in tens of millions of dollars in wasted profit over a
four- to five-year period. Moreover, routing inefficiencies undoubtedly contributed to a longterm
inability to compete in the marketplace.
Sales and Distribution
The PPMs' distribution scheme consisted of regional master distributors, sub-distributors
to those masters, and small endpoint stores (e.g., bodegas). Most PPMs retained an in-house
sales force to manage relationships with the regional master distributors. Significant interrelationships
among the large distributors resulted in a high degree of cross-shipping, so regional
promotions and margin differentials were difficult to maintain. A master distributor usually
purchased a phone card at a 32-38 percent margin (on face value); of this margin, about 22-25
percentage points were left to the store owner.
Endpoint stores sold every phone card for cash; as a result, the entire network operated on
a cash basis. Single transactions between the regional distributors and manufacturers would often
exceed $300K in cash. With so much cash exchanging hands, it should not come as a surprise
that internal and external procedures often strained the limits of legality.
Master distributors were often immigrant-owned, ethnically specialized organizations;
furthermore, these distributors implemented significant barriers to entry (which were not capital
barriers but were far more demotivating). Combined with the heavy competition among
manufacturers, this resulted in tremendous power for the distributors. The primary result was
distributor demands for onerous credit terms and margins from manufacturers.
The cash nature of the business and the complexity of the distribution system were the
primary reasons that larger, incumbent domestic telephone firms did not enter the international
prepaid phone-card market. Although many well-known providers did offer domestic cards, the
space in which TTC competed was primarily composed of immigrant customers making
National and regional advertising were generally eschewed in favor of word-of-mouth
and channel-saturation techniques. Immigrant communities tended to be very close-knit; thus, a
well-performing card (either in quality or minutes) did not go unnoticed for long. Master
distributors were powerful enough to block the sale of a good product should the manufacturers
have failed to offer sufficient margin or terms; there was much debate about whether the same
distributors were capable of producing a similarly powerful positive effect on sales. Brand
loyalty was exceedingly low; the general consensus was that heavy competition among
manufacturers had forced commoditization of the product.
PPMs advertised the rates for each product using point-of-sale posters, which often
displayed the price of the card ($5, $10) and the corresponding minutes the user would receive
on a single call to every nation worldwide. If a customer purchased a card, he would find the
same rates quoted by voice-prompt on dialing the PPM switch.
The phone-card industry grew quickly and without the burden of federal regulation. Stiff
competition and increasingly robust switching software led to industry-wide acceptance of false
advertising. PPMs' quoted per-minute rates (both on posters and by voice prompt) were
substantially increased by unquoted surcharges (per-minute percentage increases) and one-time
fees of various incarnations (weekly fees, per-call fees, etc.). At times, these additional
surcharges and fees were so high that the charged rate was more than twice the advertised rate.
The practice of false advertising became so pervasive that the average consumer assumed
that most products offered far fewer minutes than they delivered. Consumer perception was
strongly tied to the manufacturer's label, and so new products generally suffered from the same
The actions of manufacturers only exacerbated the situation. Often PPMs released a new
product that delivered exactly as advertised; however, because of perceptions of false
advertising, the PPMs were forced to advertise untenably low rates in order to generate even
baseline sales. Sales would quickly balloon as word spread, and the product would cannibalize
existing, profitable products and quickly become significantly unprofitable. The PPMs were then
forced to increase the unquoted fees on the products to stem the outflow of cash, which further
strengthened perceptions of false advertising.
A card that offered exactly its advertised rate and imposed no additional fees beyond the
per-minute rate was known in the industry as a clean card. Clean cards had become exceedingly
The New $5, 60-Minute Phone Card
Kyle Shilling thought TTC's economic analyses of new products were generally
inaccurate and simplistic. In particular, he was concerned that they failed to take into account
uncertainties the firm knew existed. Shilling was convinced that the limitations of TTC's
economic analyses were responsible in part for the firm's inability to resist the temptation to
deviate from clean cards. For this reason, Shilling decided to use the analysis of the proposed
new card?which Francona had specifically said he wanted to be clean?to think more carefully
about how TTC performed analysis of this type. He started where he always did, by considering
the three main components of the economics of any card: revenue, inbound cost, and outbound
cost. He decided to use the two-week period as his unit of analysis.
Shilling knew that TTC's master distributors had recently been satisfied with receiving a
32% discount on the face value of phone cards. This meant that TTC would receive as revenue
68% of the face value of the new $5 offering, a sum of $3.40. Total revenue would then be
determined by three key factors:
Period 1 sales (in number of cards): The phone-card market was extremely efficient in reacting
to new product offerings: A card would realize its market potential almost immediately, and
retain that level until a new, clearly dominant card would typically cause its immediate and
complete demise. Peter Francona had thought (somewhat optimistically in Shilling 's opinion)
that the new card would sell one million units in the first month, which translated into a best
guess of 500,000 cards in the first two-week period, or $1.7 million in revenue.
Period sales growth (percent of period sales): Although a competitive new card offering would
go from "zero to sixty" in a hurry, there was almost always some residual growth, as long as the
card stayed competitive. Although changes from fortnight to fortnight varied, they did not do so
dramatically. Also, the size of the initial splash clearly influenced the subsequent growth rate.
For anything in the 500,000 card range, however, Shilling decided it was not unrealistic to
assume, on average, a 2% per two-week period growth rate for as long as the product stayed
Product lifespan (in bi-monthly periods): Due to heavy competition and a commoditized
industry, products rarely experienced gradual declines in sales; more frequently, a product
dropped out of the market in the span of only a week. How many periods the card remained
viable on the market was the direct result of how quickly a competitor could come up with a
more attractive option. Sometimes, cards were "one-hit wonders," selling well for two weeks and
then being completely cannibalized by a better product. Yet, a good card, often based on a
special, particularly favorable one-off cost structure, could last for as long as a year before
becoming cannibalized (sometimes by the very company that issued it in the first place). The
lifespan of a product also depended on how well the product was received by the company's own
sales team, the distribution network, and consumers. Shilling decided as a base case to assume a
lifespan of 10 bi-monthly periods, about average for TTC products.
Inbound carrier costs
Salespeople at TTC often had an annoying tendency to overlook inbound carrier costs,
because the per-minute costs were generally low compared to outbound rates. As outbound rates
continued to drop, however, the inbound leg of the call became more important. The product was
going to offer both toll free and local access. Local access rates were currently fixed at $10,000
per two weeks per card. Toll-free costs were charged at per-minute rates and varied minimally
with the location of the origination number. Currently, the average per-minute cost to TTC for
toll-free access numbers was $0.0080. While it was certainly possible that this figure could
change?up or down?Shilling could not conceive of it changing by more than a few percentage
points in either direction.
Because the local-access charge was fixed per two-week time period, the more customers
that used the local-access numbers, the lower TTC's total inbound costs. Shilling had some direct
experience with Mexican consumers on previous cards; he estimated that 75% of total minutes
on the new card would arrive over toll-free access numbers. He could not imagine being off by
more than a couple percentage points in either direction.
Outbound carrier costs
Shilling talked to the routing department and found that three potential outbound
carriers?two fairly established, the other brand new?were indeed offering novel pricing
schemes to Mexico. Shilling understood that carriers traditionally offered a single rate to a given
country; however, recently some carriers had begun offering separate rates for different
destinations within a country. Furthermore, some carriers had started charging different rates at
different times of the day. At TTC's behest, the carriers were also offering volume discounts,
contingent on exclusivity. Shilling also knew that carriers were infamously undependable. They
would not allow prices to be locked in under a contract beyond the first couple of weeks and
routinely adjusted rates irregularly as their own costs changed. Also, carriers occasionally
experienced service interruptions, which forced PPMs to scramble and use expensive back-up
providers to meet demand until service could be restored. Shilling would need to account for all
these factors in choosing among the three carriers.
Arbitel was the most progressive and largest of the three carriers, bringing with it a
reputation for relative dependability. On its offer sheet, the firm had listed hundreds of
destinations within Mexico, but had also categorized each destination as either urban or rural.
Calls to all urban destinations were charged at the same rate, as were calls to all rural
destinations. Also, Arbitel offered a significant discount for calls placed from 12:00 a.m. to
11:59 a.m., a period Arbitel labeled "off-peak." Arbitel did not have a history of large
unannounced rate increases or service interruptions.
Shilling had also heard of BC Tel; the firm had very humble beginnings but had
established itself as a low-cost provider over the last few years. Although BC Tel charged the
same rate for every destination in Mexico, it did offer a 50% discount for calls placed during offpeak
hours, which it also defined as 12:00 a.m. to 11:59 a.m. BC Tel was known as one of the
only carriers that passed cost decreases as well as increases along to its customers and had
offered TTC a substantial discount, if the company managed to hit an aggressive traffic target.
The carrier also had a relatively good track record of limited service interruptions.
TTC's director of Carrier Relations admonished Shilling about placing too much stock in
Course Communications because it was a very new operation and very little was known about its
size, origination, or business practices. The carrier was not offering any time-of-day discount but
did offer amazingly low rates to the urban areas of Mexico, which it defined in basically the
same way as Arbitel. Given the director's concern, Shilling believed Course Communications in
terms of continuity of service and constancy of rates was the least reliable of the three carriers.
The carrier's offer sheet included a nominal discount.
A summary of the carrier's pricing schemes is given below. The discount volume is given
in total minutes over a billing period, which was the industry's standard two weeks.
ArbiTel BC Tel Course Communications
Peak - Rural Cost $0.0693 $0.0600 $0.0788
Peak - Urban Cost $0.0452 $0.0600 $0.0294
Off Peak - Rural $0.0534 $0.0300 $0.0788
Off Peak - Urban $0.0348 $0.0300 $0.0294
Discount Volume 32,500,000 33,500,000 27,500,000
Discount 5.0% 10.0% 2.0%
To perform an economic analysis of the three pricing schemes, Shilling knew he would
have to make assumptions about what percentage of customers would call during the peak and
off-peak hours, as well as what percentage would call rural versus urban destinations. Even
though he had experience with Mexican consumers, he had little feel for this aspect of the
analysis. Because carriers had not historically differentiated rates based on these categories,
Shilling had no reason to pay attention to them. Extrapolating on his very modest knowledge of
other countries, Shilling decided to assume that 80% of all card minutes would be incurred
during the peak time period, and 25% of all minutes would go to rural areas. He was very aware,
however, of how limited his knowledge was in this area.
As mentioned previously, outbound carriers were free to raise (or lower) rates, even on
existing contracts, as market conditions changed, and the current offers retained that right. Thus,
to be on the safe side, Shilling allowed for the probability that rates would change (probably
upward) if the card survived its initial launch. Even though rate changes did not occur at regular
intervals, Shilling assumed a steady rate of increase (or decrease) captured long-term trends in
rate adjustments and reasoned that it would be a conservative approach to protect TTC from
unanticipated increases. For Arbitel, he decided to assume an average 1.0% increase per twoweek
period. For BC Tel, who seemed to more successful at holding rates steady (and also to
pass on decreases as well as increases), he decided to assume no change (0.0% increase). But
because of its newness, Course Communications was more difficult. Given the carrier's
aggressive rates to urban areas, Schilling decided to assume a 2.5% increase per two-week
period. To simplify his analysis, he further assumed that all rates would be adjusted by the same
percentage; this was also consistent with his experience with other countries, that carriers found
it easier to justify overall rate increases (as being beyond their control), as opposed to changes in
the relative rate structure.
This was the trickiest part of the analysis, as far as Shilling was concerned. Service
interruptions could occur as the result of foreign government intervention, infrastructure
problems, or poor business practices (such as overselling capacity). Larger, higher-cost carriers
tended to be more stable than cheaper ones, meaning fewer service interruptions. The cost of an
interruption depended generally on its length, which was a function of how quickly the carrier
could restore service. Again, the more established carriers seemed to be more efficient in
Shilling knew he wanted to correct the failings of previous analyses, which ignored the
costs of the inevitable service interruptions. He also believed that Course Communications had to
be the most at risk in this regard?more than either Arbitel or BC Tel. Based on his general
knowledge of the Mexican market, he decided to set aside $1,000 per two-week period for
Arbitel and BC Tel for unexpected service interruptions. And influenced by cautious attitude of
the director of Carrier Relations regarding Course Communications, he decided to set aside
$5,000 per two-week period. If Course Communications could overcome that handicap, Shilling
reasoned, he would feel comfortable going with them.
Shilling was happy with his base-case analysis and his incorporation of the important
factors (for his model, see associated spreadsheet file). But he knew he had made a lot of
assumptions, and he was convinced that TTC's failure to account for the uncertainty in key
inputs was hurting its ability to come up with popular and profitable products. Unfortunately,
fully grasping the complexities of the situation and building the base-case model had taken a full
day of work, and Peter Francona expected a complete analysis from him the day after tomorrow.
Shilling knew that the CEO regarded economic cost analyses as a highly questionable use of
time. If he was going to have any hope of changing Francona's point of view, he would have to
do so quickly. This meant he had just tomorrow to go out and find data that might help him come
up with better estimates of his input, as well as to determine how much actual uncertainty to
allow for. Shilling wondered how best to spend that time.
The solution discuses a new prepaid phonecard for Taurus Telecommunications Corporation.