Overview of Search Results
What You’ll Learn
In this section, you will learn:
- What is a Vertical
- Vertical vs. Universal Search
- How Answers uses Federated Search Architecture
- What is a Direct Answer
WWGD: What Would Google Do
You’ve used Google before, right? It’s safe to say that’s a given. Have you ever really thought about how it works though or why it looks the way it does? So much of the interface is second nature to us, which is why Google is the search engine king that it is. They know how to make the most intuitive interfaces and they have billions of users helping them improve every day. It’s no wonder that every other successful search engine out there uses many of the same principles — including Yext Answers.
Let’s run a search together. Go to Google.com and search for “Sicily”.
This seems really standard, right? Let’s break it down:
- Prominent search bar at the top to modify or run more searches
- All your results listed in an “All” tab
- Vertical search engines (also known as vertical searchers) appear dynamically in tabs, like “Maps”, “News”, “Images” or “Videos”, and are ordered based on their relevance to the query.
- Clicking on a vertical searcher tab will re-run the query on that tab
- Results grouped together in what we call Verticals, like “People also Ask” and “Things to Do”
- A Direct Answer card for Sicily showing more information about the place
- Link results at the bottom taking up the rest of the SERP with about 102 million results (would hate to be on that last page)
Well, Answers is architected very similarly to Google.
Let’s run a search on Yext’s Answers for something simple, like “phone number nyc”.
Disclaimer: We’re making improvements to our own search experience every day — and even moving offices soon! So, the results you see when you conduct the search might be different from the ones you see here. Don’t panic! You should still see the similarities.
Let’s deep dive into some of the key components that make this possible, namely:
- Federated Search
- Universal Search
- Vertical Search
- Direct Answers
Before we can go any further, it’s important to define what a Vertical is in the context of Answers. When we refer to a Vertical, we’re talking about a class of results, often synonymous with an Entity Type (though you can choose to have multiple entity types in a given vertical if you’d like those results to be combined, like Branches and ATMs).
Examples of popular verticals include:
- Case Studies
And so many more…
When you build an Answers Experience, you’re actually building out and defining each vertical individually — what it should look like, what entity types and fields should be considered, any boosts or blacklisted intents, how you want results sorted, and so on.
For example, in the “phone number nyc” search, there are actually 3 verticals returned:
- Offices: powered by the Knowledge Graph
- Help Center: powered by Zendesk
- Links: powered by Google Custom Search Engine
So why do verticals matter and how do they turn into search results? That’s where Federated Search comes in.
Federated Search is a technology framework where the search engine queries multiple data sources and aggregates the results in one unified experience for the user. Yext Answers uses a Federated Architecture to search across verticals within your Knowledge Graph, all from one search query. Whether the user search is for products, professionals, or locations, this architecture will identify the context and return results across the verticals. Answers can also integrate with third-party systems (e.g., Zendesk) to return even more results to users — all in one place.
When a user runs a search on Yext Answers experience, it looks like this:
Answers routes the search to each vertical in the Experience, including those that are using Knowledge Graph data and any Third Party Verticals (which you’ll learn more about later). It aggregates the relevant results, figures out what order to put them in (using both the Answers algorithm and the logic from the Search Configuration), and passes the information via the Answers API to the frontend for display.
Federated Search is important because it allows us to tailor search logic per vertical, both universally across all Experiences but also individually to each Experience via the Search Configuration. It also provides a scalable, decentralized architecture to keep results quick.
Federated Architecture is a common practice in Search technology. If you’re interested in learning more, here’s some additional reading:
- Understanding Federated Search and Why It’s the Key to Unlocking Your Information
- Information Management
- Federated Search
- Federated Search 101: What, Why and How
Universal Search vs. Vertical Search
There are two types of search that you’ll become very familiar with in Answers:
- Universal Search - searches across multiple verticals
- Vertical Search - searches across a single vertical
Universal Search is like Google’s “All” tab, aggregating all of the different types of relevant search results in one screen, with the opportunity to deep dive into any given vertical with a quick click. Universal Search is designed to give users a preview of each vertical, but is not designed for deep or detailed search of any one vertical.
For example, if you run a query on Yext.com for “Answers” here are the results you’ll get:
You see Vertical result sections for Products, FAQs, Webinars, Publications – and there are even more if you keep scrolling. If I wanted to read more FAQs, for example, I would click on the “View All” link for FAQs, which will run the same search but just for the FAQ vertical. You could do the same thing by clicking on one of the tabs below the search bar, similar to how other consumer-facing search engines work.
Vertical Search, on the other hand, is meant to be purpose-built for a given vertical, with much more ability to refine or browse your results. Vertical searchers can and should look different, depending on the Entity type.
For example, here are 3 verticals in the same experience: Jobs, FAQs, and Locations. Jobs has faceted search, FAQs have an accordion style card, and Locations are accompanied by a map.
WWGD? Well, a very similar paradigm.
Since Universal and Vertical search are different they support different features. While you’ll learn more about each of these along the way, here’s a reference table for you to keep in mind:
|Universal Search||Vertical Search|
|Single search bar searching across all the different verticals and providing one unified search results page. This would typically be the main search bar on a website.||Searches only a single vertical, which is typically limited to a single entity type or group of related entity types (e.g., locator, FAQ search, event calendar).|
|Supports Knowledge Graph & Custom Backends|
|Allows results from multiple verticals|
|Supports Facets and Static Filters|
|Supports Hardcoded Prompts|
|Supports Popular Queries|
|Supports Backend Sorting|
|Supports Frontend Sorting|
|Supports Vertical Intents|
Not sure what Facets or Vertical Intents are? Not a problem - we don’t expect you to, you’ll learn more about them as we go.
Direct Answers with Structured and Unstructured Data
You ask a question, you want an Answer! And, ideally you want this Answer as fast and as straight forward as possible. That’s where direct answers come in.
When we are very confident that a search has one answer, we will surface a Direct Answer box at the top of the results. We only do this when we think we know the exact entity and field content that the user is looking for. We will also surface any other results that may be relevant below the direct answer.
A Direct Answer can be surfaced from both structured and unstructured data. With Answers, you can set up the backend to surface Direct Answers from structured data in the Knowledge Graph (think: phone numbers, addresses, titles, etc) as well as unstructured data (think: long form content on help articles or blogs).
For example, if someone searches for “Phone number NYC office” on Yext.com, we know that they are looking for:
- The Yext NYC office
- The phone number
This data is structured and lives within a Yext Knowledge Graph, and the Direct Answer search result would look like this: