Amap and Baidu Are Again Pitching Maps as a Gateway to Local Services. Why Hasn't the Same Pitch Worked in Ten Years?

Finding good food and a restaurant still isn't a map's business.

On January 9, 2007, at the launch of the first iPhone, Steve Jobs demonstrated three internet services: video, maps, and music.

There was no App Store yet, and the iPhone could not install other apps. For more than a year afterward, apart from web pages in the browser, those were the only internet services an iPhone owner could use.

Maps were Jobs's earliest picture of local services — using where you are right now to connect you to the businesses around you. He searched the Maps app for a nearby Starbucks, and once the address and phone number came up on screen, two taps put him through to an employee.

"Good morning, Starbucks, how can I help you?"

"I'd like to order 4,000 lattes to go. Just kidding, wrong number."

Ten years on, carried by the smartphone, video sites had passed traditional television in both viewing time and advertising, in China and the US alike; China's online music industry was worth more than 15 billion yuan, ten times its 2006 value.

But map services are still telling the story they told ten years ago.

This August, Amap launched a feature called Geek Map, an entry point that looks like a news site. Tap a headline such as "China's Most Beautiful Universities," though, and what opens is not an article but a set of locations. Amap says the new product is meant "not only to solve travel problems, but also to solve life problems."

Open any map service — Baidu, Amap, Tencent, take your pick — and you will see much the same screen. There is always a "Discover" button below the map, through which users buy movie tickets, book hotels, buy group-buying vouchers, and so on.

At Apple's 2016 developer conference, Tim Cook demonstrated a new Maps app, this time with table booking from Dianping and Yelp folded in — no phone call needed, as there had been nine years earlier. Tap open the search box at the bottom and eight buttons appeared, all tied to local services: Food, Shopping, Travel.

Ten years apart, tech companies are again trying to turn the map into an enormous gateway to everything. That the same goal has been restated for a decade suggests it is still nowhere near reached.

Amap and Baidu Have More Than 500 Million Users Between Them, and Still Aren't a Gateway to Local Services

According to a report from the mobile data firm Aurora Mobile, one in every five mobile internet users in China now has a map app on their phone; the two biggest, Amap and Baidu Maps, have more than 500 million monthly active users between them.

A report Statista published in 2017 shows that the function people use most often on their phones is messaging (WhatsApp, Messages, Slack and other chat apps), and that maps rank above music and photos.

Google Maps has more than a billion users and, measured by traffic, is Google's second-largest asset after Search — worth more to the company than YouTube or Gmail. In 2013 it was, for a while, the most-used mobile app in the world.

Maps have also turned into something like infrastructure, reaching into services you never think about: ride-hailing, food delivery, group buying, bike-sharing — all of them need a location.

Over ten years maps have also become easier to use and more detailed. But people mostly open them during the morning and evening rush, which means the function they use most is still looking up a route.

Over those ten years, what trained users' spending habits was one specific website or app after another, not the map.

The map itself has not become a gateway. Or rather, it has yet to develop any character beyond that of a navigation tool.

Electronic Maps Reached the Public Through Google, More Than a Decade Ago

Humans have been drawing and using maps since the scratchings on Stone Age cave walls. Fifteen thousand years ago, in caves near what is now Lascaux, France, hunters painted the animals they hunted along with lines tracing migration routes. Counting from the oldest world map that survives, the Babylonian world map of the sixth century BCE, the history runs nearly 3,000 years.

The modern map is a product of surveying science. From measuring ropes and set squares to telescopes, theodolites, and aircraft, better instruments directly improved people's ability to render the real world. The materials that carried maps advanced too, from clay and tile to cloth and paper.

Maps only went electronic in the last half century. In the 1960s, computer hardware kept improving under the demands of nuclear weapons research, and computers became able to store and analyze data — able to "draw."

In 1967, the Canadian Roger Tomlinson built the first Geographic Information System (GIS) put to practical use, storing and analyzing geographic information about farmland.

Companies went into the map business too, among them Intergraph, founded in 1969 by a group of engineers out of IBM, NASA, and the military, whose main line was selling mapping software and hardware to enterprises.

Maps on a screen stayed in that shape for a long time: they took special equipment to use, the customers were mostly companies and governments, and the uses were confined to scientific research, construction, the military, and agriculture.

Not until the 1990s did a company called MapQuest offer online route guidance, putting an electronic map on the web for any internet user for the first time.

It looked nothing like a map of today, though — you never saw a whole map. MapQuest made you type in a destination by hand, then generated, from a street-level geographic database, one electronic map centered on that spot. Another function searched directly for a route from A to B, telling you how to get there and how long it would take.

It was Google that settled what an electronic map would look like. In the first iPhone Jobs demonstrated, the Maps app was Google Maps — Google supplied the service, Apple built the app.

It started as a rather boring little project.

In 2003, Google set up an internal project called Search by Location. Enter a zip code or an address and Google returned a set of search results tied to that place. It was still search at bottom, search over verified yellow-pages listings, and the results still came back as a series of web pages.

"To be honest, it was a completely useless project," recalled Bret Taylor, its product manager at the time. Search by Location went whole days without a single user. Taylor later became Facebook's CTO, and now runs the collaboration tool Quip.

In 2004, Google bought three map data providers: Keyhole, Where2, and Zipdash. Keyhole's core technology was stitching countless fragments of satellite imagery together and matching them onto a map.

On February 8, 2005, the first version of Google Maps was posted to the tech community Slashdot. "A map you can drag with your mouse. Cool!" "Best web app I've ever seen." The early reviews were good.

It looks crude now. Places on the map could not be clicked and opened no page — it was a paper map put on the web. Its widest coverage was the Americas, only the continental US had real geographic data, and Canada and Mexico next door were outlines.

Because nothing else worked that way, Google Maps drew more than 10 million visits on its first day online. Within a year it was the largest map service in the world and Google's second-biggest source of traffic after Search.

China's internet map companies appeared at about the same time. The earliest was Tuxing Tianxia, founded in 1999 and later bought by Sogou and renamed Sogou Maps; Amap Software was founded in 2002. Their main customers were still governments and companies, and they sold map data to carmakers and telecom operators.

The company playing Google Maps' part in China was Baidu Maps. When it launched in 2005 it too was only a simple web page, but ordinary people could now browse a map without leaving home.

Before smartphones spread, Google Maps — which then had a Chinese version people in China could reach normally — already had a team working on mobile, but the mobile version ran on a narrow range of handsets, BlackBerry among them. Mark Crady, an engineer there at the time, recalled that because almost nobody used it, the mobile app's traffic numbers looked bad and downloads were few.

The first iPhone was where Google Maps really went mobile. Before the launch event, Apple and Google struck a deal: Google would supply the underlying data, Apple would handle the interface and the interaction, and Google Maps would ship preinstalled on the iPhone.

In the years the iPhone and then Android phones spread across the world, the map also became a critical piece of technical groundwork — travel services, restaurant reviews and the rest all needed a map interface. By ProgrammableWeb's count, by 2012 there were already 240 map-related services, and the number of developers building on them had grown 243% over 2009.

In the years after 2008, Baidu and Amap also released Android and iPhone versions with navigation.

The maps themselves got better in this period. In 2007 Google's Street View cars set out around the world to photograph the ground. By 2010 the map covered all seven continents.

The Chance to Become a Gateway Went to Local Services Like Dianping, Meituan, and Yelp

At that first iPhone launch, the scene Jobs sketched for maps was about local services from the start.

Unfortunately, while the map companies were busy refining their data and fighting big rivals, "calling Starbucks from your phone" was claimed first by group buying and review sites.

In 2003, Zhang Tao, who had spent ten years in the US, founded Dianping on the model of the American Zagat Survey, offering business listings built mostly around restaurants; users could rate them, leave comments, and talk to each other. Yelp launched the following year, founded by two former PayPal engineers, Jeremy Stoppelman and Russel Simmons.

Unlike map products, these third-party review sites cared about operations. They accumulated data written by users and listings uploaded by merchants, and became communities with a distinct atmosphere. As Yelp's Stoppelman put it, "Yelp pursues the experience of the review itself. It's more like a blog with a fixed structure."

The street-level promotion that group-buying sites would later make familiar began with Dianping and Yelp. Their teams went out onto the sidewalk and handed out small gifts in exchange for sign-ups.

Each time it entered a new market, Yelp hired a local community manager whose job was to cultivate reviewers and coax new users into writing.

In four months alone, from April to July 2011, five group-buying sites — Meituan, Dianping, Wowo, Manzuo, and F-tuan — raised close to $500 million between them. The money went to subsidizing users and merchants.

None of this was the sort of thing map companies did. They were busy buying data and building new features.

As the iPhone and Android phones spread, Dianping and Yelp put out mobile apps of their own. The phone made their use concrete: you could look up businesses near you, based on where you stood.

Both grew into very large businesses. Yelp began expanding fast into overseas markets after 2010 and listed on the New York Stock Exchange in March 2012.

Dianping started group buying in 2010 and added reservations in 2012. The group-buying war that followed ended in 2015 with the merger of Meituan and Dianping. The combined company is now a giant valued at $25 billion to $30 billion.

After more than a decade of accumulated content and data, the actions that make up local services — is this restaurant any good, can I book a table online, is there a discount — now belong to services like Dianping.

The map's role in all of it is foundational: no group-buying or food-delivery service can work without the location data maps provide. Google, wanting more of this everyday information, paid $115 million for Zagat to help users find the best places nearby.

None of that changed much either. The map's original vision of becoming a gateway to local services was never realized.

Advertising Can't Turn a Profit, So the Map Companies Return to Local Services

How Dianping and Yelp make money is easy to grasp: charge merchants for advertising and put them somewhere more visible. The home page is a billboard too, able to show different kinds of content in different slots. It is the same form of advertising Baidu, Google, and today Toutiao all sell.

Map products have a hard time doing that.

Data costs a great deal to build. Google has never said what it spends on Maps, but according to The New York Times, its spending in 2012 ran between $500 million and $1 billion a year, about a fifth of Google's R&D budget.

Take Amap's financial reports from before the acquisition as a guide, and a map provider's revenue comes from navigation deals with carmakers; from location-based services (LBS) for mobile apps, meaning interfaces sold to other developers; and, for companies with surveying and mapping licenses such as NavInfo and Amap, from selling data to the map operators downstream.

Google Maps has never done much about making money. You can find restaurants and shops on it, even book a hotel room, but the bill that follows has nothing to do with Google.

In 2016 Google added search-result ads to Maps. It was a restrained form of advertising: an offer tied to a place only appeared once you searched for that place. Search for a Walgreens pharmacy, open a particular store, and you saw whatever promotion was running.

That amounts to pushing the flyers a shop would otherwise pay someone to hand out on the street straight to a passerby's phone. In the official blog post introducing the new mobile ad format, Sridhar Ramaswamy, Google's senior vice president of ads, wrote: "With online ads bringing more people to your storefront, think about how you'll measure the impact."

"On a shrunken phone screen it's hard to weigh a paid ad against a genuine ranking, because there's so little screen," Brian Smith, founder of the startup Placeable, wrote in an article explaining why pay-for-placement in a search box is naturally at odds with a map. "It's like voice search. Would you listen to Alexa slowly read out a dozen results?"

This August, Baidu Maps was accused of putting Putian-network hospitals — private hospitals run by operators from Putian — near the top of the results when users searched for a children's hospital. Baidu replied that this was not advertising, only a mistaken alias match. At the least, the episode shows that any ad appearing on a map becomes an interference with serious consequences.

The Same Attempts Are Still Going On

To plug back into local services, the map companies have made their products steadily more complicated since 2015.

Baidu has been the more aggressive of the two. In 2014 Baidu Maps began pushing an "O2O strategy," a plan for Baidu Maps to do for Baidu what QQ and WeChat do for Tencent: bring in Baidu's other services and convert them into sales.

Baidu Maps was rebuilt in 2015, with dining, ride-hailing, hotels, travel, and car-owner services all stuffed into one large "Local Services" button, entry points that fed traffic to Baidu's own products. The desktop version had no such conspicuous door.

"In its development ahead, Baidu Maps will keep innovating and keep opening up, enriching its core functions and its O2O services, to build a multidimensional platform combining information and services."

At Baidu Maps' tenth-anniversary event in 2015, Li Dongmin said the routes, traffic information, shop locations, sights, commercial districts, even the flows of people and vehicles the map had accumulated were all assets to be used in the future. They just had never worked out how to make money from the information.

Baidu Maps also ran more kinds of search advertising than Google Maps. Type certain keywords and the results would include entries marked as ads.

By this year Baidu had cut back on that and made it subtler. Search for something like "dental hospital" and a suggestion such as "Jiamei Dental" appears under the search box.

Next to Baidu's own search ads or Toutiao's feed ads, advertising on a map is plainly a much poorer business. In an interview this August, Li Dongmin also said that "no result of Maps' commercialization will ever affect the organic ranking."

In January, after Lu Qi took charge, Baidu reorganized: the O2O services that kept burning cash were cut, and the maps business was recast to fit the same artificial-intelligence story.

In July, Baidu Maps settled on a brand line, "technology makes travel simpler," described itself as "the water, power and gas of the AI era," and stopped mentioning O2O.

Food delivery is gone from the Baidu Maps app now, but you can still use it to open Nuomi and buy movie tickets and group-buying vouchers.

New services keep getting stuffed in all the same. The newest version of Baidu Maps added Baidu's voice assistant, Xiaodu, so you can ask out loud for a route.

Services of the moment, such as bike-sharing, have been wired into the map too: you can use the map app to find a bike nearby and scan the code to unlock it.

Amap, Baidu Maps' main rival, said flatly after Alibaba bought it in 2014 that it would "do no commercialization for five years." Even so, pull open "Search Nearby" at the bottom of the Amap app and you will find the same recommendations for food, hotels, sights, and movies. Behind them sit Alibaba's various services — open a shop's information page, for instance, and there is a button that goes straight to Alipay.

Amap carries ads as well, buried in the news articles below its local-services section. Unless you treat it as a news app, it is hard to imagine finding the motivation to scroll that far.

Geek Map is another experiment in becoming a gateway. Amap wants you to use it for more than navigation, and ideally to come to depend on it.

Year after year, feature after feature, recommendation after recommendation.

Except that if it never occurs to users to tap the map's icon when they are looking for a business, how would they ever use any of it?

Cover image: stocksnap, Unsplash