LINEBURG


<< . .

 7
( 36)



. . >>

Infrastructure costs (technology and connectivity) are typically small and fairly pre-
dictable. The offshore unit may need to procure hardware and software. In some cases
clients have to purchase dual equipment and software licenses: one set for onshore and
the other for offshore. There are additional dif¬culties: software licensing issues may
be convoluted and more expensive offshore; and in India and in other developing coun-
tries, long lead times are required to procure some equipment.
The connectivity picture is better, costs are falling rapidly. International call rates
that were once signi¬cant are no longer burdensome, and voice over IP is beginning
to eliminate voice costs altogether. But voice lines and multi-party conference lines are
still needed in many cases. The annual leased costs of E-1 (2 Mbps) and DS-3 (45 Mbps)
circuits from India to US/UK are 43,000 and 780,000 USD, respectively (2004). Thus,
the cost of basic connectivity, an E-1 circuit, is roughly equivalent to the charge for just
one offshore programmer per year.
The most dif¬cult category to forecast is the cost of Knowledge transfer (KT), some-
times called technology transfer. KT is the notion of moving speci¬c knowledge and
experiences into the minds of the offshore developers. While some of the KT involves
well-understood skills and rules, much of it is in tacit knowledge “ knowing what is
38 The fundamentals


in-between the lines of a software speci¬cation. Packing up the software speci¬cations
and “throwing them over the wall” works for simple cases only. It does not work for
most cases. IT managers recognize this, although in practice they often fail to manage
the KT process properly. A more extensive discussion of KT is found in Chapter 7.
The largest KT cost may be the redundancy that is built into the project early on.
A typical scenario takes place when several offshore developers need to travel to the
client site for KT early in the project life cycle (which happened in the T-Corp case
described later in this chapter). At these early stages the client ¬rm is paying for double
staf¬ng for the same work, for its current and the offshore employees. KT shows up in
various other ways: for non-English-speaking countries, translation costs need to be
added. Another way to look at KT is to anticipate that things will go wrong: poor KT
early in the process leads to costly rework and repair. In the case of outsourcing off-
shore, some KT costs may be absorbed by the provider, such as with KT problems that
stem from high turnover rates, cultural training for provider employees, and some train-
ing related to transitioning.
Ef¬ciency is a productivity ratio comparing the onshore, “original” unit™s ef¬ciency
(the baseline) to that of the offshore unit. During the ¬rst few months the new offshore
individuals are less productive as they “go up the learning curve,” but over time their ef¬-
ciency rises as they master the knowledge and skills required. From a cost perspective,
however, software ef¬ciency is dif¬cult to measure, let alone forecast. While it is possi-
ble to measure ef¬ciency using objective measures such as Function Point per dollar, this
is rarely practiced. In contrast, measurement is relatively easy for offshore call centers
using measures such as number of calls per hour, duration of calls, and call satisfaction.
Some examples illustrate how the ef¬ciency item can be used in practice. Firstly,
how ef¬cient is the new offshore unit at the beginning of the project? Sand Hill Group
estimates 24% offshore ef¬ciency within 2 months; while US-based T-Corp (described
in the case later in this chapter) used a 50% ef¬ciency for an offshore maintenance
team after 4 months of KT before any production work actually began. Restating these
numbers, this means the offshore personnel are only 24% or 50% as ef¬cient as the
baseline (the original software personnel) within 2 or 4 months, respectively, from the
project start date. These low ef¬ciency ratios early in the offshore cycle are simply due
to the normal need to learn the complex knowledge.
Can the offshore units reach 100% ef¬ciency and when? Again, we use these two
examples. Sand Hill Group posits that ef¬ciency increases rapidly in the ¬rst several
months, but never quite reaches 100%. Similarly, T-Corp does not forecast reaching
100% of the baseline ef¬ciency. T-Corp assumes, based on experience, that the off-
shore personnel will reach 95% of the baseline ef¬ciency by the end of the ¬rst year.
Others claim that offshore ef¬ciency actually rises above the 100% baseline due to
quali¬ed people and solid processes offshore. This is the assumption used by Gartner,14
which uses the term “effectiveness factor.” This factor is a composite of several com-
ponents, namely technological expertise of the offshore unit, its project management
39 Offshore economics and offshore risks


expertise, and its business-domain expertise. Gartner calculates that this effectiveness
factor is lower for the typical US “Fortune 1000” corporation than for the typical large
American and Indian providers. In other words, the providers are more effective than
the client. In fact, Gartner estimates that once a steady state is reached, they are 50%
more effective: a signi¬cant difference!
Due to KT and other needs, travel costs are signi¬cant. Yet, they tend to be under-
budgeted. For example, if there is a need for extended onsite work, an offshore team of six
developers visiting the client site onshore for 3 months will cost about 150,000 USD.15
On top of the direct cost is an obscured organizational cost associated with travel to far-
¬‚ung locations “ the opportunity costs of many wasted days in airplanes, jet lag recov-
ery, and sick days due to exotic food and water. Most offshore projects expect travel to
take place at the beginning or the end of the project life cycle, or sometimes both. Of
course, travel costs vary. For example, they are lower if the work is done nearshore.
The last three cost categories are Overhead, Governance, and Risk Mitigation.
Overhead allocation, an accounting item, varies from company to company and may
sway the economic bene¬ts of offshoring from positive to negative. Governance costs
represent about 5“10% of an offshoring contract. Governance costs include new posi-
tions to communicate with the provider and to carefully monitor the provider™s work
by collecting and analyzing data. Another buried cost item includes contract manage-
ment costs such as handling invoices and payments for outsourcing. Finally, risk miti-
gation is the investment in resources in case of failure, such as backup and recovery.
The last remaining issue is the overall impact. What is the total of all of these extra
offshore costs? Not surprisingly, the rough estimates (in Table 2.2) indicate a very high
variance of these extra offshore costs: from 12% to 57% of the contract amount.

Table 2.2 Extra offshore costs (%)

CIO magazine
16
(composite)17
Meta Group

Search and Contract18 1“2 0.02“2
Restructuring (layoffs and retention) 3“5
Communication 1“3
Process changes (KT) 0“10 1“10
Cultural differences (KT) 2“5
Transitioning the work (KT) 2“3
Lost productivity/cultural issues (KT) 3“27
Ef¬ciency 0“20
Travel 2“3
Governance 5“7 6“10
Turnover at offshore site 1“2
Total extra costs 12“52 15“57
Each item represents additional cost (%) of the overall offshore outsourcing contract.
The four items marked knowledge transfer (KT) include KT as at least part of that item.
“Lost productivity” includes turnover at offshore unit.
40 The fundamentals


In other words, if the extra costs are kept under control and managed closely, they will
be smaller than the wage savings, and lead to overall offshore savings. However, at the
upper range of these estimates, at 57%, the labor savings are wiped out and offshoring
ends up costing more.



What is the bottom line? Does offshoring lead to cost savings?

So far, we dissected the main economic offshoring trade-offs: the wage differentials
versus the extra costs of offshoring. This begs the important question: Are the extra
offshore costs indeed smaller than the wage differentials? After all, so many ¬rms have
reported offshore savings.
Four consulting ¬rms estimated this bottom line and concluded that offshore savings
are positive. Firstly, a study by US-based Deloitte Consulting19 ¬nds that in the best
case of offshore outsourcing savings are in the 25% range when considering all costs
and bene¬ts. This savings level can be achieved by a typical “Fortune 50” US
corporation that has been offshoring for 5 years. Deloitte calls this “as good as it gets.”
More typical success cases are Fortune 50 companies offshoring for at least 2 years and
saving 15%. The study author speculates that if an experienced ¬rm optimally com-
bined all best practices and processes that it would return a theoretical 47% total sav-
ings. A second study, by US-based Sand Hill Group, estimates savings for software
product R&D ¬rms at 40% of R&D budget,20 adding that Return-On Investment is
reached in a year. Third, Gartner estimates the typical offshore outsourcing savings for
large ¬rms to be in the 28“40% range. Fourth, US-based Magnolia Communications
surveyed New York City companies that offshore and reported that their savings were
44%, although the study authors noted that savings of nearly this total are possible by
simply moving to a less-expensive city in the state of New York, such as Syracuse.21
These estimates and surveys should be viewed with great caution because their
methodologies are not rigorous. We have not seen studies that have examined a broad
range of companies™ offshore strategies and produced a comprehensive comparison of
cost savings. We are skeptical that such a study can reasonably be done because of the
dif¬culty of standardizing assumptions and overhead rates.
In any case, the various studies have a limited bearing for the case of any speci¬c
company, because cost savings are not guaranteed by statistics. Nevertheless, the thrust
of this chapter should have made clear that the extra offshore costs are not hidden costs
at all. They are only hidden costs for those companies at early stages of offshoring with
little idea of what to expect. They are known costs which can be identi¬ed, decomposed,
and most importantly, managed. Managing the process is the key. If the process is well
managed the TSOS will likely be positive. One 2004 study conducted in Europe found
that 80% of companies “suffered problems” in offshoring.22 This is hardly shocking
since “normal” software projects suffer problems 64% of the time according to the
well-known Standish surveys of project success.23
41 Offshore economics and offshore risks


An important lesson is that cost savings are heavily dependent on time. In other words,
many of the extra offshore costs decrease with time. This is the positive impact of orga-
nizational learning illustrated in the “Stumble-and-then-Succeed” story of ValiCert ear-
lier in this chapter. There were many wrong turns, frustrations, and wasted spending.
Yet another lesson of the cost savings computations is that if your ¬rm is offshoring
it must ¬rst produce good internal benchmarks to determine if you are indeed
saving money. And you need good cost accounting to compute these benchmarks. As
you continue to expand offshore you need to show real cost savings in order to move
forward.
As a ¬nal note, all of these ¬nancial analyses ignore two vital issues: strategy and
risk. Firstly, the analysis, thus far, covered only costs and cost savings, yet the bene¬ts
of offshoring can be in less quantitative bene¬ts: in strategic and tactical advantages;
in speed; in quick ramp-up time; in availability of able resources; and innovation. The
strategic bene¬ts of offshoring are discussed in Chapter 5. Secondly, even a compre-
hensive cost analysis does not address risk. In other words, the cost savings projections
could show substantial savings, but the risk factors may be too high. Cost and risk are
not the same. Offshore risks are covered later in this chapter.




Case study Calculating the extra offshore costs at a giant American corporation
The case of T-Corp illustrates the process of offshore cost computation in detail. The case is
an actual case, but at the request of this large American company, all identifying details are
disguised.
The Finance Of¬cer took a copy of the offshore spreadsheet and computed the
¬nancial net present value (NPV) for the proposed offshore engagement. It exceeded
the 15% that the division sets as a minimum threshold for budgetary approval. “This
is great,” remarked the Finance Of¬cer, cheerily, and blessed the project.
Bobby Sanders directs the Global Services unit at T-Corp, a US “Fortune 500” technology
company. Global Services is an internal division tasked with matching the corporation™s
internal units to offshore resources. Bobby manages a network of six captive offshore cen-
ters (wholly owned by T-Corp); in India, Singapore, and several other locations. Much of the
software code developed at T-Corp is embedded software.
In 1998, Bobby developed an offshore spreadsheet template to assess the ¬nancial bene-
fits of offshore work and assist his internal corporate customers in making offshore decisions.
Between 1999 and 2004 he used the offshore template to assess 55 candidate engagements.
Bobby noted with pride that about one-third of the proposed projects were rejected by “run-
ning the numbers.” In these cases the “economies simply weren™t there” and the decision was
made to leave these activities onshore. Some of these were rejected because they were end-
of-life projects where the expensive knowledge transfer process was not justi¬ed.
42 The fundamentals



One of Bobby™s most important “wins” for the Global Services unit was in 2000,
in persuading T-Corp™s Strategic Software Division (SSD) to begin to peel off some
of its lower value work offshore.
As its ¬rst foray into offshore, SSD decided to consider the work of 30 engineers at
its Ohio engineering site. These were engineers that were immersed in their embed-
ded code. They knew it, they built it, and they maintained it. But every modi¬cation
request (MR) that came in tended to distract the engineers from their most important
task “ working on the next product release. Quite simply they were falling behind.
As they fell further behind, they became more attracted to the offshore pitch: “We
like this offshore idea,” Bobby remembers SSD™s Director saying.
Donald Robert, one of SSD™s product directors, visited Bobby™s of¬ce to begin
evaluating the offshore engagement. The two sat down in front of the computer
screen and Bobby pulled up his offshore spreadsheet template and began to explain
how it can be used to help in making the offshore decision.
The ¬rst hurdle in using the offshore template was lack of benchmark data. The
Ohio product group had no process data that could be used as a basis for computa-
tion. The group collected almost no metrics. Given this, how could they make the
¬nancial case for offshoring? Through dialogue Bobby and Donald found a reason-
able proxy. They examined the engineers™ time sheets and then estimated the
amount of time the engineers spent on MRs. This estimate came to 33% of their
total work time. Bobby and Donald used this ¬gure to compute the baseline ¬gure
in the spreadsheet. Thus, 30 engineers multiplied by the fully burdened cost of
150,000 USD per year 33%. This resulting number, 1.5 million USD, became the
current onshore cost. This ¬gure was entered as the ¬rst computation item into Part
1 of the offshore spreadsheet representing onshore costs.
The next important issue was making an estimate about knowledge transfer,
which knowledge transfer was going to be expensive and time-consuming because
the current 2 million lines of embedded code had almost no documentation. This
was taken into account when an important “plug factor” was used to drive the
spreadsheet “ which Bobby calls an ef¬ciency factor. Clearly, the offshore engineer
was not going to perform at 100% of an Ohio engineer™s capacity from Day 1.
Bobby usually uses a 50% ef¬ciency for the ¬rst few months of the engagement.
That is, each offshore programmer is only half as ef¬cient as his onshore counter-
part. But as knowledge transfer proceeds successfully, offshore ef¬ciency goes up.
In this case, Bobby used a gradually increasing ef¬ciency factor ending the ¬rst year
at 95% and continuing at 95% for subsequent years.
The other major offshore cost items are listed here:
— Onsite training involved bringing ¬ve of the Singaporean engineers to Ohio for

2 to 5 months for knowledge transfer. The costs of apartment rental, airfare, and
per diem for this period was not cheap: 92,000 USD.
43 Offshore economics and offshore risks



Fully burdened offshore labor costs were 5400 USD per month.


— Infrastructure expenses were broken down into three categories, some of which

are driven by tax considerations rather than “straightforward” economics. The
¬rst item is non-recurring infrastructure (an “expense-able” item), principally
software and hardware purchases, which usually include customs and tariffs.
When pointing at this, Bobby complained that software licenses at the foreign
sites tend to be higher than in the US. The second item for infrastructure is
mostly ongoing costs such as leased communication lines. The ¬nal infrastructure
item is infrastructure subject to depreciation which is bene¬cial with respect to
foreign tax treatments in some countries, such as Brazil.
— The last cost item was one of the largest, the local (onshore) resources. This is

the redundancy that was built in at the beginning of the engagement. The Ohio
engineers would have to continue working on the MRs while the new offshore
engineers were learning and acquiring knowledge in the ¬rst few months; in
other words, while the offshore engineers were still far less than 100% ef¬cient.
In the ¬rst few months this item, local resources, represented a substantial cost
item at over $50,000 per month, dropping quickly after that as the offshore
engineers become capable and more ef¬cient.
Bobby pointed out how well knowledge transfer was managed; it was budgeted cor-
rectly in the offshore spreadsheet and, more importantly, it was managed well. Four
full months of US engineers™ time were budgeted, at 45,000 USD, before the
engagement even began. The Ohio division paid close attention to the details of
knowledge transfer, such as on-going job enhancement of the offshore engineers,
helping to keep turnover at the offshore site to a manageable 10%.
In 2004, Bobby speculated that the offshore engineers, now with 4 years of expe-
rience, were actually operating at 120% ef¬ciency. In other words, they had become
more ef¬cient than the Ohio engineers who had trained them. He explains this by
pointing out that the Singapore (offshore) engineers, who do nothing but MR work,
stay more focused. But, Bobby noted that he has never gone so far as to use a
spreadsheet ef¬ciency factor above 100% for the offshore units. “It™s just too spec-
ulative” he said, as he shrugged his shoulders.
After some weeks of data collection and many telephone calls, the offshore
spreadsheet for SSD was complete. It showed a positive cumulative savings by
Month 18. Bobby™s offshore spreadsheet produced cost summaries and cost savings,
but no NPV. SSD™s Financial Of¬cers took a copy of the spreadsheet and computed
the NPV. The NPV exceeded the 15% that the division sets as a threshold. “This is
great”, remarked the Finance Of¬cer, cheerily, and blessed the project. Bobby
already understood the ¬nance game by then. He always avoided submitting proj-
ects in the third and fourth quarters because they could not show payback by the end
of the ¬scal year and were likely to be rejected by Finance.
44 The fundamentals



The offshore effort represented a long-term investment in knowledge transfer, as
can be seen in Figure 2.2, showing the actual forecast out to the end of 2004 “ an
almost 5 year timeline from engagement launch. The projection shows a total cost
savings of 2 million USD over 5 years. Four years into this engagement, the forecast
was deemed valid and SSD™s management was pleased with the offshore impact on
productivity. In early 2004, to validate the economics of offshoring, SSD collected
and analyzed data from a large number of MRs handled in either Ohio or Singapore.
The analysis revealed that at a US engineer™s rates (fully burdened), it cost 7000
USD to ¬x an MR, but only 4500 USD to ¬x an MR in Singapore.

2,500,000


2,000,000


1,500,000
Cost/saving (USD)




1,000,000


500,000


0


(500,000)
Ma rt
Ju 0




Se 4
Se 3
Se 2




De 4
Se 1




De 3
Se 0




De 2
De 1
De 0




04
Ma 3
Ma 2
Ma 1
Ma 0




Ju 4
Ju 3
Ju 2
Ju 1
r-0




0
0
sta




0




0
0




0
0

<< . .

 7
( 36)



. . >>

Copyright Design by: Sunlight webdesign