THE internet looks like an adman’s dream. Counting how many times an advert on a bus shelter has been viewed is impossible; counting clicks on a blinking banner ad is a doddle. But knowing where each click came from, and how many people are clicking, is harder than it appears.
Firms dedicated to click-counting put code on websites that reports the times, origins and frequencies of visits, or get consumers to install it buried in browser plug-ins or mobile apps. These record web-users’ digital calling-cards: the internet-protocol (IP) addresses of the devices they are using. But to assume that each IP address represents a single user in its country of registration is a wild oversimplification.
A new report published on November 4th takes a different approach. GlobalWebIndex (GWI), a market-research firm with local partners in 32 countries, surveys 170,000 consumers a year and recently began to ask detailed questions about internet use. It puts China and India in the top three for Facebook users. SimilarWeb, which does IP-based analysis, does not even put China in the top ten (see maps).
One reason for the difference is that in many developing markets devices are widely shared (for tablets that is true pretty much everywhere). Conversely, more than three-quarters of respondents in the GWI report said they used more than one device. Another factor is the spread of virtual private networks (VPNs) and proxy servers, which make it possible to surf the web through a foreign server.
Once restricted to the tech-literate, these are now common and easy to use. Chinese citizens who want to vault the Great Firewall to use Facebook (banned in China) can do so with a couple of clicks. Foreign fans of the BBC can use the same trick to watch its programmes via iPlayer, supposedly barred outside Britain. Since VPNs and proxy servers are clustered in countries with favourable rules, such as Sweden and the Netherlands, any count of visits to such sites will be skewed.
GWI’s picture, it should be said, is far from complete. It misses out Africa entirely, except for South Africa. Self-reported data also have their pitfalls: LIRNEasia, an Asia-Pacific IT think-tank, recently found that many Indonesians who reported using Facebook said they were not internet users—perhaps because they were not sure that one implies the other. And much of the world is going mobile-only, particularly in developing markets; preliminary GWI data suggest that a quarter of web visitors in Indonesia and Vietnam use only a mobile (from which VPN access is, these days, just as easy). Yet both surveys and click-counting software were conceived and optimised for desktop users. Uptake of mobiles is faster than the effort to capture demographic data from them.
More broadly, knowing who is online, and where, would benefit government policymakers as well as advertisers. Other figures on technology use are available from the International Telecommunication Union (ITU), a UN agency. But there are gaps here, too. It collates surveys from national census bureaus around the world, but cannot insist that they all pose the same questions.
Better figures would be useful, says Susan Teltscher of the ITU. They would help the agency fulfil its mission to ensure web content is available in the languages of its users. In nations where internet use is low, says Kojo Boakye of the World Wide Web Foundation, radio is regarded as the medium with the widest reach. Revised figures for web use would shift how public-service messages are distributed. And as internet use spreads, regulators will have to oversee competition between service providers, fight e-crime and plan investment in web infrastructure. But such efforts rely on the numbers that now seem so muddy, points out Geoff Huston, a researcher at APNIC, the internet registry for the Asia-Pacific region.
The ITU estimates that 4.3 billion people around the world are yet to get online, 90% of whom are in developing countries. But until data from surveys and clicks can be combined into a single picture, the map of internet usage will be little clearer than the viewership of bus-shelter ads.