Google paid for the final design of the stations and is covering the monthly fees for satellite bandwidth. The company has also invested in O3b, a start-up that hopes to deploy a constellation of satellites over Africa by the end of next year.

“Building infrastructure is not necessarily Google’s objective, but if you look at all the areas that Google has gone into, in many cases it has been to fill a gap,” said Joseph Mucheru, who heads Google’s East Africa office. “The market should see the opportunity.”

Just how much opportunity there is remains unclear. Google is uncertain whether such satellite stations can pay for themselves in rural areas, given the cost of equipment and bandwidth. Communities may well benefit from the connection, but they do not all have the means to afford it.

Bandwidth fees for stations like the one in Entasopia could cost as much as $700 a month, though slower ones cost less, said Wayan Vota, a senior director at Inveneo, a nonprofit that works to disseminate Internet technology throughout Africa and the developing world. As these connections are introduced more widely, which is O3b’s goal, the price could fall, Mr. Vota said.

Here are the summarised results from the telecenter operator survey done by LIRNEasia at the weCan workshop in October 2008. Sample was not representative, but large enough to get a general idea about the telecenter operations in Sri Lanka.

Out of a total of 147 operators surveyed, the bulk, 101 were from Nenasalas, the 500 odd telecenter network created under the World Bank funded e-Sri Lanka programme. 10 were from Sarvodaya multi-purpose telecenters and 6 from others (eg. public libraries) 30 have not specified the type of the telecenter.

Do telecenters in Sri Lanka make money? Yes. They report an average monthly income of Rs. 22,119. (=USD 201) This is associated with a relatively large standard deviation of Rs. 21,714 (= USD 197) indicating a variation within a wide range. Not a surprise since some telecenters are running at a loss (presumably temporarily) and few reporting a monthly income of over Rs. 100,000 (= USD 900).

However, providing Internet services ranked only third among telecenter income components (16%). The key sources of income are education and training (43%) and providing fax, photocopy and printing series (21%). They also make money from VoIP (4.5%), bill payments (2.5%) telephone calls (2.5%) selling other goods (2%) and VCD/DVD rentals (1%).

Asked for the preferred income profile, the results were not too different. They still want 33% income from training, 21% from fax photocopy and print services and 17% from Internet services. Is this an indication of being more realistic or less ambitious? You decide.

Average monthly expenditure of a telecenter is Rs. 15,837. (USD 144) This may not reflect the real costs as the Internet charges for Nenasalas are borne by ICT Agency from the money comes from e-Sri Lanka programme. Salaries is the highest cost component (39%). Then comes electricity (25%). About 10% each for telephone and rent.

On average basis a telecenter has 5 PCs and employs four staff members – two permanent, two temporary. Correlation between the income and the permanent staff strength was 0.56; the income and the number of PCs, 0.62.

Do telecenters make a profit? Yes, but perhaps may not in real terms. They record a monthly average profit of Rs. 6,735 (=USD 61) with a large standard deviation of Rs. 9,504 (=USD 86). This indicates the loss incurred by some of them. This is again without considering the cost of the communication link. (The monthly average cost of a 2 Mbps business broadband connection is USD 46 in Colombo. This might be slightly high in rural areas.)

Telecenters operators are rewarded in different ways. Only 33% are salaried. 22% receive a share of profits. 13% receive an allowanced based on performance. 32% receive no personal income. How they prefer to be rewarded? 51% wants a monthly salary; 26% a share of profits and 18% a performance based allowance. Doesn’t sound too entrepreneurial but in Sri Lanka culture job security plays an important role.

The presentation that Helani Galpaya and I did (Sujata and Chanuka ran a parallel session) included components on innovation in service industries, the external environment that made innovation so important for telecenter operators, and systematic learning from failures. Because we had to work with multiple languages, it was not possible to cover all the slides, which are here.

One of the things we noticed was that there appeared to be two different kinds of problems: the first kind could be fixed through process innovation; the second kind was structural and required remedies that were outside the scope of an event like this. An example of the latter is location. When the telecenters under e Sri Lanka were first planned, the principle was that they could not unfairly compete with existing unsubsidized communication bureaus and that there had to be no other telecenters near their locations. Yet we found that this principle is being violated in the rush to put as many telecenters in place as possible. In some cases location decisions had been made on the basis of promises of connectivity, which had not been fully met.

So with the next offering, we will say upfront that structural problems are outside the scope and focus the participants on the solveable issues.