10 realized that this was actually a technical
and human-centered design challenge that
presented an incredible opportunity for us
For instance, people in India are not accustomed to paying for water. Even if they are
willing to pay, cash collection is an incredibly
expensive task in villages that are far apart.
And, even if you could get the cash, a distant
company selling a locally available natural
resource is simply untenable.
Our solution was that local water should
be treated and consumed locally. Villagers
would buy the Sarvajal service from a local
franchise, not directly from Sarvajal. And, to
best support a franchisee model, the technology needed to be a turnkey solution.
We had early success. Of the communities in the Shekhawati region of Rajasthan
where we first began our operations, roughly
20 percent of households started buying
from Sarvajal within weeks. Franchisees
started making money.
But we ran into problems. Water had
to be available as we promised—daily and
dependably—and we quickly learned that
it is an unimaginably difficult task to keep
hundreds of complex machines running in
dispersed villages with unskilled operators,
unreliable electricity, intermittent mobile
phone service, with franchisees who are
likely to blame you for any and every failure.
Although our business was fundamentally
hardware, the challenges of distributed
infrastructure for water in India began to
seem a lot like the challenges of distributed
infrastructure in information technology.
To live up to our promise of water for
all, we had to think through major technical and engineering challenges. Those
challenges ranged from industrial design
to thermal management to sensors barraged by power outages, operator error,
and people trying to game the system,
while at the same time keeping a tight
rein on our costs. Frugal innovation was
the only way we could fulfill our fundamental promise of affordable clean drinking water to our customers.
When we started Sarvajal six years ago,
we knew very little about water filtration
machines. The machine for our pilot effort in
Rajasthan came to us in a truck full of crated
parts: motors, pumps, switches, membranes,
tanks, and so on. The truck was followed by
a welder, a plumber, a filtration expert, all of
whom built the filtration machine onsite. It
was a custom-built machine that took several days to assemble inside the room where
we installed it.
We realized that this process was highly
inefficient. Experts told us that machines
had to be custom built based on the water
source and output requirements. When we
asked another supplier to send us a pre-built
machine to a particular specification, and it
came to us in a wooden crate that took up a
lot of space and the crate alone was 20% of
the cost of the machinery.
There was no way this was going to work.
Our initial team was full of tinkerers who
were obsessed with Internet technologies.
We dreamt of solutions that came more
clearly from our experience with web technologies, not with clean water policy per se,
giving us a fundamentally different perspective on approaching the problem than others
with strictly policy backgrounds.
We had to figure out how to reduce
the all-in costs of the machinery, includ-
ing time, installation, transport, and so
on, and we wanted to standardize and
modularize the machinery to make it easy
to deploy. We were confident that the
math around customizing machines could
be translated into a basket of products
built around a standard design, especially
because water quality was reasonably
predictable if mapped by locality. Our goal:
Largely out of ignorance, we started trying
to build machines in various configurations.
We needed to reduce space, maximize the
number of machines we could fit in a stan-
dard 24-foot open-top truck, create quick-
release connections, pre-calibrate and prep
membranes, among other issues.
We came up with a standardized
500-liter-per-hour vertical membrane
machine built inside a frame that would
also protect it during transport (thus requiring no expensive packaging material) and
that could be installed within minutes of
arrival. Moreover, the standardization of
the machine helped us reduce the costs of
machinery by 75 percent without significant
re-engineering of the components. It wasn’t
perfect, but within a year and a half, after
much trial and error, we had a machine that
was cheaper and more efficient than those
of our competitors—and nearly every one of
them eventually began using our machine
template and/or our supplier for their own
Once we started deploying our machines,
we realized that we were still quite far from
our goal of reaching those who had the least
access to clean water. Our goal was to get
water to the last mile, but even our small
and standardized plants required around 800
customers to make it financially feasible for a
franchisee. And that fact kept us from reaching the small hamlets where water was an
even more acute problem.
Our search for a solution to this problem
found inspiration in a piece of technology
very common in the developed world: the
automated teller machine. For a “water
ATM” to work, it would need a number
Frugal innovation was the only way we could fulfill our
fundamental promise of affordable clean drinking water
to our customers.