Coronavirus Google Searches Could Save Lives
Buying ads in a pandemic
In 2019, I served ads to people in crisis to try to help them. This included people who told Google they were going to shoot up a school, commit suicide, use heroin, or join ISIS. From what I’ve seen, just as ads can get someone to buy a shoe or insurance, they can also convince someone to not carry out a violent or self-destructive act.
And yet I started to find an aspect of this work unsatisfying. Besides Google, I was the only one receiving the data of the people clicking my ads. The data wasn’t getting to those who really needed it: the experts in harm reduction, suicide prevention, counterterrorism, and so on.
The data that suicidal people leave behind when they call a helpline through an ad, for example — that’s valuable. Theirs is the data of people who were saved by the data that people like them left behind. Yet for all its value, it passes through to Google (where it’s lost to the machinations of their targeting tools), and it passes through to me. And that’s where its journey ends.
In a perfect world, the data would not wink out in Google’s server rooms or in my computer. It would be able to flow to Google, to me, and then be broadcast to experts through a new, third channel.
But there’s no way to do this. No way to create this channel and broadcast ad click data to the world. At least that’s what I thought. But then something happened in our world that made me think I really ought to try to confirm whether this is or isn’t possible. This thing that happened was the coronavirus.
Google is banned in China. Nevertheless, people in China use VPNs (virtual private networks) to mask their location so they can circumvent the Great Firewall and use Google.
While a VPN user’s location might be hidden from the Chinese government, it is not hidden from Google. As a result, Google has data on millions of people in China. That means those people can be served ads.
For the last few weeks, I’ve been serving ads to people in China who use VPNs to Google any phrase containing “coronavirus” or “COVID-19.” My ads appear #1 at the top of their results page, promising them access to current information on the virus. There are ads in Chinese, for users who have their browser set to Chinese, and ads in English, for users who have their browser set to English.
When someone clicks the ads, the clicker is directed to a Chinese or English U.S. Centers for Disease Control (CDC) page about the coronavirus. (One quirk of the Google Ads system is that you can send a click from your ad to any website you want.)
Then, the data passes through to Google, and to me: the word-for-word search entered, age, gender, income, and so on. And though the clicker is using a VPN, their actual location at the moment of the click is logged in the Google Ads interface.
With this data, we can create a heat map of the frequency of searches containing “coronavirus” or “COVID-19” based on the city where the searcher is located.
I created this map in Google Data Studio, a free tool. Data Studio syncs easily with Google Ads, so the click data imports directly into Studio and can be quickly plotted to charts and maps.
With the map, it’s easy to see where the searches about coronavirus are coming from. The deeper the blue, the higher the frequency of searches. We can see that, out of all the cities in China that can be served ads, about half of the searches over the last two weeks come from Wuhan, Shanghai, and Beijing.
In Data Studio, we can apply a date range to see how searches evolve day-to-day, week-to-week, or month-to-month — by intent, or geography, or by any metric we choose.
The click data we get from Google searches that contain the word “coronavirus” is not very useful because coronavirus is a broad topic. But we can zoom in on different aspects of the virus by choosing different keywords (in Google Ads) that we want our ads to show up for when people use Google.
For example, by swapping out the keyword “coronavirus” with the keyword “masks,” our ad will now show when people in China Google things like “out of N95 masks.” Because we have Data Studio linked up, now we have a heat map that reveals which city in China needs masks the most, in real time.
If we use the keyword “quarantine news” in Google Ads, we can see which cities need updates about lockdowns. We can swap “quarantine” with any type of news, to get a sense for which areas are most affected by censorship, by topic. If we want to compare reality with the narratives provided by the state media, we might serve ads to people who tell Google they want to report infections, and compare the heat map with official figures.
We can learn especially powerful insights by segmenting the data in Data Studio. For example, we can create pivot tables that show search frequency by user location, plus the searcher’s age, gender, or even the language they speak. For example, we can learn how many 29-year-old, single-parent women in Beijing who speak English need public transportation alternatives during the lockdown.
Of course, we should help searchers as best we can with the ads while we compile the click data. A CDC info page is not a suitable destination for certain searches. For example, pregnant women in China currently have great difficulty finding hospitals that still offer prenatal and postpartum care. The women clicking these ads could be directed to a website that compiles all hospital information in one place.
While all ads can be made more helpful by routing searches to appropriate destinations, the click data is always useless if it just ends up with Google and then sits in a chart in Data Studio. Click data must be made accessible to anyone who can use it. Here are a few ways to broadcast it to the world as efficiently as possible.
One way is to install a basic piece of code from Google Data Studio on a website. Through this setup, whenever someone clicks an ad, the click data will automatically populate on the website. Further, anyone can right-click on any report on the site to download it for free.
I run this very setup on a website I built to track coronavirus/COVID-19 searches and make the data freely available: www.coronasearch.live.
My website is a prototype. It’s meant to showcase the potential of what can be revealed with Google click data and how the data might be displayed for public access. I plan to expand the site in the coming weeks, bringing in data from new keywords and new countries.
That wasn’t always my plan. But now that coronavirus is officially a pandemic, it’s no longer appropriate to pull click data from China alone. Thankfully, in Google Ads, we can adjust the location targeting to get data from any country, city, state, borough, region, or zip code in the world.
The downside is that this data costs money. More specifically, the ads do. To pay for the ad clicks that fuel the data on my website, I spend about $200 per day. And that’s just to serve ads to people in China whose searches contain “coronavirus” or “COVID-19.” Expanding this targeting to the United States, where coronavirus searches are so numerous — and where Google has data on over 270 million devices — would cost ~$18,000 per day.
I think this is a small price to pay to connect searchers with information, while simultaneously building an indelible ledger of some of the most accurate real-time data possible to receive from societies grappling with the virus. Because with this data, we’ll be able to know the true story — told by people’s thoughts and identifiers routed away from a corporation — of how any aspect of the virus evolves over time and geography. And we’ll be able to forever look back and compare it to milestones during the virus’s development.
I am confident, from running my own tests, that this data will reveal heretofore unseen patterns and trends about the virus.
As I write this, when you Google coronavirus terms you will see a dialogue box pointing to the Centers for Disease Control (CDC) and the World Health Organization (WHO). They’ve taken other steps to control misinformation, such as banning ads for face masks. And yet they haven’t made this valuable coronavirus search data publicly available. Remember that Google isn’t against sharing this data. Because I pay for it, the click data passes to me. But it also passes to Google. And while I have data from just China (for now), Google has data from every Google user in the world.
Google could make the click data publicly available themselves; grant licenses to agencies like the CDC or WHO to use the technology; or offer grants to those who are paying out of their pocket to find the data (without all the limitations that advertisers usually face when they run ads through Google’s Grants program).
For now, few people (who aren’t marketers) know to look for this data, or that it even exists. But we don’t have to wait for Google to change that.
