Looker Semantic Search Block: What it is & Use cases
The Looker BigQuery Semantic Search block is a new feature that Looker has added to its open-source Looker x GenAI solution. In this blog, we will delve into what Semantic Search Block is, and how it speeds up business users in the decision-making process.
I. What’s Semantic Search Block?
With the Looker block, customers can directly utilize BigQuery’s semantic search capabilities within their Looker instance. End users can dynamically create searches and perform similarity matching in real-time.
For example, starting with the search prompt “coachella clothes” and the number of matches “3000” we can easily have the list of top-matched products. BigQuery will scan through the dataset and pinpoint products that closely align with the prompt. The results are ranked by relevance, enabling us to refine or broaden the search as desired.
At first glance, you might think: I can still use the contains filter in Looker.
However, the contains filter will find data based on the exact wording or patterns but cannot understand the meaning or context of the search query and might return some irrelevant outputs.
A semantic search block offers a more powerful and user-friendly approach. It identifies semantically similar documents, even if they don’t contain the same words.
II. What problems does it solve?
In many business teams, there is a high demand to analyze a specific but not-yet-categorized item to respond promptly, either capitalize on opportunities or resolve issues.
- E-commerce use case: Assume you work in the Marketing department for a fashion company. There is an upcoming concert, which provides an opportunity to promote fashion items that people will purchase for the event. You want to have information on audience size, total generated revenue, forecasted revenue, and audience type driving revenue. By gleaning these insights, they make quick decisions on targeted promotions, and inventory adjustments. However, the products are not yet categorized in your data, so you cannot immediately filter out the products and make a quick analysis.
- Food-delivery use case: Or imagine you work in the Strategy & Planning of a food delivery firm. You want to increase breakfast orders. To accomplish this, you must first identify a list of food items appropriate for the morning event, conduct an analysis to evaluate the promotion plan, and then provide the list to the Product team for display in Banner Ads on your app. However, this is an initiative and the data is not available. You cannot immediately filter out the food items and make quick analysis.
You request assistance from your data team. And here’s what happened…
3. How Semantic search block solves
With semantic search, business users can directly input in natural language and get insights instantly, which eliminates the long lead time and dependency on the data team to write complex queries.
In short:
II. Use cases & Demo
1. Use cases
1.1. Analyzing trending products: This scenario applies to businesses like e-commerce and food delivery, where new popular products or food items emerge frequently.
- Identify trends & gain audience insights: audience size, total and forecasted revenue
- Make Data-Driven Decisions:
+ Targeted Promotions: Run special offers to capitalize on the trend and maximize sales.
+ Inventory Adjustments: Ensure you have enough stock of trending products to meet demand.
+ Temporary Menu Additions: For food delivery services, add trending food items to menus for a limited time.
1.2. Analyzing customer requests on emerging problems: This use case benefits customer service or support teams:
- Surface Customer Concerns
- Identify Emerging Issues
- Take Prompt Action: Address these emerging problems swiftly to minimize customer frustration and maintain a positive brand image.
2. Demo
Let’s move on to an example from the e-commerce sector: With the Coachella festival close at hand, a fashion e-commerce website needs to decide whether to and how to increase the amount of visibility for those products.
Thanks to semantic search, they now can have a customized list of products and quickly examine the insightful information, such as:
- Insights 1: How this custom product segment has performed historically? And how the products will perform over the next 6–12 months?
- Revenue: 10% of revenue comes from the custom segment, suggesting that it’s a promising area.
- Top-selling brands and product categories associated with this segment.
- The seasonality of sales for this segment: As the Coachella music festival takes place in April each year, upward trends from January to April are often positive indicators. Also, we can utilize the forecasting function in Looker to predict the sales in the next 6 months.
2. Insights 2: What are the characteristics of the audience size?
With these insights, the company can leverage the identified audience and optimize the product mix to enhance the campaign’s effectiveness.
HOORAY! Now you can see that Looker Semantic Search can help the business team simplify data analysis & speed up the decision-making process
If you have any questions related to this setup guide or want to know how to fit this into your organization’s use cases, please leave your contact info HERE and we will get back to you right away!
Besides this, we also offer other extensions in the Looker x GenAI package, which are Data Actions, Explore Assistant and Dashboard Summary.
You can also read our other blogs on Medium and visit our website to know more about us.
