Google Now, Microsoft Cortana & the Predictive Search World – #SMX Liveblog
Google Now continues to grow and change, as does Microsoft Cortana and the other digital assistant rivals. It’s important for search marketers to keep up with these technologies that can present answers to people before they even think to search for them (aka, “predictive search”). In this session at SMX East, Cindy Krum gives us a tour of Google Now and covers what the latest changes mean for marketers.
Moderator: Danny Sullivan, Founding Editor, Search Engine Land (@dannysullivan)
Speaker: Cindy Krum, CEO, MobileMoxie (@suzzicks)
Predictive Search: How It’s Changing Our Interaction with Technology
We’re going to talk about predictive search (PS), and that’s an exciting area for search. Danny is jaded, yet this gets him excited.
Some questions that will be covered in this session include:
- What is the future of predictive search and where is it going?
- What are the signals used to predict searches?
- What do predictive search results look like?
PS is exciting, but it’s changing quickly.
What is predictive search and why is it important?
Search engines take all the information they can gather and give you an answer before you search. To predict a person’s needs, they use:
- User’s current context
- Historical personal behavior patterns
- Aggregated behavior patterns
- Active solicitation of information
This enables search on connected devices. This is important because technology is evolving and more of our devices are connected. Things like casting options in Chromecast are allowing more things to be web-enabled.
Predictive search has some strong use cases. There are lots of things you may want to search, but they don’t have a keyboard or even a screen.
Wearables, for instance, are a strong category for growth. Predictive search is the best answer on wearable devices because it means you’re not talking out loud, although voice search is the second option. The presentation layer (where the information is displayed) is very small.
What is Google Now?
Google Now’s two main characteristics:
- Card formatting – You’ll see items formatted in the shape of a card to provide one screen’s worth of information. Cards show up automatically based on your history.
- Voice search
Google Now is cross-device. We see it on wearables, mobile, desktop and TV.
What signals are used to drive PS results and what do those results look like?
Current signals contributing to predictive search:
- Where are you?
- How fast are you moving?
- What day is it?
- What time is it?
Google Now looks at what you usually do on that day and time (in the past) to give a prediction response without your asking.
What signals are used to predict searches?
Google aggregates the apps to see a combined “you.”
Google Now: App Notifications
There’s an option for showing you a notification in a Google Now result. Here are some examples of what these Google Now cards might look like:
Google Now email integrations:
If emails are marked up for an action behavior, Google can include that information in a Google Now card. It’s especially common in travel-type notifications. There’s a tool in Search Console that can be used to test the markup you’re developing to add to your email HTML.
Cross-device data collection:
You can see that Google connects a previous search from your desktop and brings it up on your phone. Even just typing “s” brings up a previous search query in the suggestions.
You never gave permission for Google to connect your mobile and desktop devices. It just does that because it knows you’re logging in on both devices. Because desktops/laptops don’t usually have GPS devices built in, Google Now will base your location on the location of your phone. If you leave your phone at home and do a Google search from another location, it will get your location data from your phone and think you’re in the wrong place.
YouTube and Google Play are other apps that Google Now will try to combine data from into your profile.
Those are examples of passive data collection. Google also solicits information from you, for instance:
There’s also an app called Google Opinion Rewards — you get asked your opinion on different things and are rewarded with points that you can use to buy things. #bigdata
Examples of Predictive Search Results
Here are some of the cards you’ll see:
- Stocks
- Weather
- Reminders/calendar
- Sports
- Shopping
- Travel
- TV shows
- Travel and transportation
- Parking locations
- Public transportation delays
- Traffic and alternate routes
- Flight and train tickets
- Meeting reminders based on travel time
- Activities
- Nearby events
- Nearby attractions
- Nearby photo spots
- And the list is growing every day
Google Now is predicting that when you’re traveling, you’ll need to know more things about where you are. It’ll also give you weather for where you are and your home town.
Controlling offline devices:
- Nest is the thermostat Google runs.
- You can use Google Now as your remote control.
- When you’re setting up your Google Now, it solicits information about your entertainment providers.
It pulls your ZIP code and all of your providers to make recommendations for you.
Voice Search and TV Audio Search
You can listen to the TV and get more information about the episode you’re watching.
Real-time TV suggestions and from-SERP reminders are based on what Google Now knows you like. Because you’ve watched this before, it’ll give you suggestions.
Reminders for Live TV and Links to Online Episodes
These types of reminders help Google sell more advertising. They also gain more traction for the Google Play service.
You can tell that the right-side image is a Google Now result because of the menu along the bottom.
There are relationships between similar shows derived from Google+ and Google Now.
Today, Google Chrome v2 was announced. It has more interactive and gaming options. Because so many media devices don’t have a keyboard, using voice search capabilities that they’re learning from Google Now is going to help Google web-enable more things.
Google Now and Android Wear
What is Cortana?
The main difference between Google Now and Microsoft’s Cortana digital assistant is that Cortana has less information about you. This is because Microsoft is not as integrated in people’s lives.
Cortana integrates with Windows 10. This might be promising, but it hasn’t taken off yet.
Because it doesn’t have as much information on you, it solicits information about you, such as restaurants you like or things you never want to see.
What is the future of predictive search?
Google Now on Tap = Mobile OS Central Interface
Google Now on Tap provides better and more predictive search results that integrate more seamlessly with your web and app behavior.
Lots of new cards in SERPs:
She thinks we’ll see a blurring of lines between Chrome searches, voice search, and Google Now (as well as between apps and the web). She thinks that Google Now will be the start screen, where Google has you begin. Google’s Android M(arshmallow) is sticky enough to bring together the app and the web world. The more they’re able to marry the app to the web world, the less work they’ll have to do to index it and get people to it. And the better it will be for users and to maintain.
When that happens, we’ll see more inter-app operability and deep link integration of apps in search results.
Verified App Deep Links
For Android devices, web links can automatically open apps. She gives these tips for handling app links (and suggests https://developer.android.com/training/app-links/index.html for developer-level information):
- Create intent filters within an app for website URLs.
- Configure the app to request verification of app links.
- Publish a digital asset links JSON file on the website.
Click a blue link and you don’t have to worry about whether its app or web — that’s what Google is going for.
There will be more casting functions, listening functions and music cards. Worldwide, Google Play Music is one of the most installed apps.
Sponsored Google Now Cards: Pay to Play
In one example, Cindy had been viewing a dress on Target, then she saw a Google Now card of the dress with a price drop.
More use of photo learning, like the current ability to find “visually similar products,” which is a way to get in front of people with similar products. Get your product that’s similar to a popular product in front of people with a similar photo. If the photo is on a woman, use a woman in your photo, and if on a mannequin, use a mannequin.
There are more interactive cards coming.
Google Now, Microsoft Cortana and predictive search are changing everything from how we find things to how we interact with our devices.