We sat down with Jason McClelland, Algolia’s CMO, to discuss the importance of search. Jason explains why search is much more than just a box in the upper right-hand corner of a website or application. Businesses should strive for adopting what users increasingly expect: the most relevant and personalized search and discovery experiences.
Throughout this article, learn why Algolia takes an API-first approach, how you can get best-in-class search up-and-running (and more quickly than you may think), and how Algolia approaches personalization with machine learning.
Why should businesses be thinking about search?
We see search as the front door to web and mobile applications. It tends to be the first part that users interact with. For example, 43% of users on retail websites go directly to the search bar, and a recent Forrester blog reported that 92% of B2B purchases start with search.
Search has become central to many online services beyond just eCommerce, including media, B2B, and SaaS, to mention a few. Every time your customers interact with your application or webpage, you’re making a query for some content. Those queries should behave like a search engine, dynamically ranking and serving the most relevant experiences, personalized for those specific customers. Conversion rates go up, customers engage more. People move from searching to browsing and discovery. Companies merchandise and curate content using search.
And now we’ve added AI-powered features like personalization, product recommendations, and dynamic relevance based on trends. With Algolia, all of this comes without needing to manage a team of developers, or to hire the expertise required to build it yourself.
Why is it so important for Algolia to provide an API-based solution for search?
Our API-first search provides the most relevant content to your users from the moment they start typing, with mere milliseconds of latency. We provide the infrastructure, and we’re really good at hosting a search service with high availability and low latency. The API provides a great developer experience to index and search your content. It also gives direct access to configure the engine’s settings and build advanced features.
What does it take to get Algolia Search up and running?
Start by thinking about what sort of search and discovery experience you need for your application. Does your end user’s interaction start with search? Then you’ll want a prominent search bar with strong autocomplete capabilities. Do your users tend to browse your site? Then you’ll want a faceted index that you can use to build category pages. And these are only some of the questions to ask at the outset.
Once you’ve designed your search experience, the next task is to use our APIs (in all popular programming languages) to turn your catalog into a searchable index, hosted on our global servers. Then you start building the search user interfaces that your business needs. You choose the components (we call them widgets) from our front end javascript libraries or framework integrations.
One thing customers forget to do is include analytics into their search from day one, to track how their users are interacting with the search. Analytics gives access to a number of advanced features like click and conversion insights, Personalization, A/B testing, and Recommendations.
Algolia has a second product, Recommend. Explain why recommendations make sense as a second product for a search company to offer?
It’s a natural progression of our focus on increasing relevance for end user content. Our Recommend service uses the same event data that we use to personalize search results to train machine learning models. Algolia manages these models for you and retrains them against your event data on a daily basis. Then you can use the trained models to inject relevant results anywhere you need to make a recommendation.
Can you give an example of a Recommend model and how you would use it?
Sure! Two models that are currently available are “Frequently Bought Together” and “Related Products.” Both use a rolling sample of the last 30 days of end user event data to find patterns between products. For instance, if an end user has a ski cap in their cart, our model might suggest a pair of gloves to go with it based on the buying habits of previous users.
What do you see as the future of search?
A few things come to mind:
- Personalization will be crucial. As applications add machine learning, users will expect search and discovery crafted to their personal preferences everywhere. Collecting high quality data around user activity and maintaining well-trained machine learning models is critical to providing those experiences.
- A greater variety of AI models will enter the market to help solve all the complexity of search, while keeping flexibility for developers. AI will help to improve developers’ productivity.
- The potential usage of synthetic AI-generated data as an alternative to indexing, ultimately solving scalability problems.
- New input methods ‘beyond the search box’ such as scanning, visual search, and NLP/Voice have changed how users produce queries and interact. Companies will need to adapt with APIs that can support these new input methods.
Where can users learn more about Algolia?
We invite you to join our active developer community to talk more about building with Algolia. When you are ready to try it out, you can sign up for free here.
You can also find us on GitHub, Twitter, Facebook, YouTube, and Linkedin.
About Algolia
More than 10,000 companies including Under Armour, Lacoste, Birchbox, Stripe, Slack, Medium, and Zendesk rely on Algolia.
By the numbers
- 1.5 trillion search operations / year are powered by Algolia
- 400k developers in the community
- 10k+ customers
Additional resources:
Related Categories