Autosuggest is a means of guiding your site visitors as they create search queries, maximizing the relevance of items in search results.
Autosuggest eases the task of information discovery by helping your site visitors express their intent. It shows the most relevant queries, exposes related queries, and helps reduce bad queries like misspelled queries. Autosuggest reduces the overall effort in completing a search task.
As visitors type their queries, autosuggest displays a list of likely queries terms or phrases. As they continue to type, the list narrows to what they're looking for. Visitors can stop typing their queries and select an autosuggested query or product from the autosuggest list.
Autosuggest is interactive. Visitors can select a query or product from the suggestion list as soon as they see one that fits their needs. The list relies on what customers are currently typing and predictions based on past general traffic flow.
These suggestions help visitors discover products they're looking for even when they use terms that your site might not be using to categorize or keyword items. For example, Sophie is a customer who is looking for headphones at an electronics retailer. She thinks of the particular kind of headphones she's looking for as headsets, so she starts to use the search term, headset. The retailer's site autosuggests headphones to her after she's typed just hea. She sees headphones in the suggestion list and selects that suggestion instead of continuing to type her own query.
Autosuggest examines your site visitors' queries while they type them, then suggests queries and products. Autosuggest predicts the intent of your site visitors.
Let's consider this query: red shoes.
Every Little Thing is a general retail site that sells all manner of merchandise, including shoes and movies. Kate is a customer who wants to buy a pair of red, strappy sandals. She enters the query, red shoes, in the site's search box.
BloomReach suggests queries while Kate types. Here are some of these suggestions:
- The Red Shoes
- red shoes
- red heels
- red-soled shoes
- red sandals
- red boots
The last five suggestions make sense in the context of Kate's intent. The first suggestion only makes sense in the context of Kate's typed query. However, The Red Shoes is a classic movie, not an actual pair of shoes, much less Kate's red, strappy sandals. In fact, the likelihood that any particular customer typing red shoes in the search box is looking for the movie rather than a pair of red shoes is low.
As a merchandiser at Every Little Thing, Calvin knows that customers looking for the movie are going to find it. They know that red shoes is more likely to find pairs of red shoes than a movie. They're likely to restrict their search to a particular category like Movies than to simply expect The Red Shoes to be at the top of a search results page for a red shoes query on a site that sells both shoes and movies. Calvin is much more concerned about customers like Kate. Nearly every customer at Every Little Thing who enters the query, red shoes, is looking for a pair of shoes that are red.
Calvin's solution is to create an exclusion rule for the query, red shoes. He specifically excludes the movie, The Red Shoes, from the list of autosuggestions for the red shoes query.
All non-alphanumeric characters (any character other than 0-9, and A-Z) are considered special characters. Bloomreach treats special characters as blank spaces by default.
Bloomreach picks up the terms and queries that your customers use when they search for items on your site, then refreshes regularly to provide current and appropriate autosuggestions. Traffic is important – Bloomreach only creates autosuggestions when terms are commonly entered, not if only a few customers use a term once or twice. The suggested queries are ranked in descending order of user engagement.
If your site doesn't carry particular items, then Bloomreach doesn't add those search terms to autosuggestions. For example, if you don't sell Gucci products, then Bloomreach doesn't add Gucci to the autosuggestion list even if Gucci is a commonly entered search term on your site. If a valid search term becomes a zero result query due to products going out of stock, such search terms are also removed periodically.
Bloomreach also considers the bounce rate of queries to determine what to autosuggest. Queries that have very high bounce rates will be removed to ensure that suggestions are high quality.
Bloomreach's semantic engine also generates structured queries, which combine product attributes and product types to create suggestions. This can help with "cold start" problems when you introduce brand-new product lines into your catalog by helping those products get discovered more easily. For example, "dinner set" could have a structured query of "red metallic 12-piece dinner set".
Here are some rules that determine how suggested queries are ranked:
- Prefix-dependent ranking: For a given prefix, such as "sh", suggestions are ranked based on what users are more likely to click on. If users are more likely to click on "shoes" than "shirts", then "shoes" will be ranked higher, even if "shirts" is searched more frequently as a query.
- Frequency: Queries that have been searched a higher number of times will be ranked higher in suggested queries.
You can create and modify exclusion rules to change autosuggestions. An exclusion rule excludes or blocks a suggestion from appearing in the dropdown list. For example, an art print retailer carries prints of paintings by Pierre Auguste Cot. Some of these paintings feature nudity, but this particular retailer strives to appeal exclusively to museum patrons and other customers interested in classic art. Therefore, this retailer decides to exclude terms like "naked" from its autosuggestions. Please note that exclusion rules apply globally that is excluded queries will never be suggested.
Autosuggest exclusion rules for Multi-sites
Autosuggest does not respect the site selected in the top-right merchant dropdown, hence implementing a multi-site Autosuggest blocklist is not possible.
However, the Autosuggest blocklist can be managed by language. To configure an Autosuggest blocklist rule for a specific language, select the language from the language-based dropdown.
If you have multiple view ids but a single language, the Autosuggest blocklist behaves the same as it does for a single site.
The merchandising rules like boost, bury, include, exclude etc. have a direct impact on the way products appear in the product grid on your site. When it comes to product suggestions (products surfacing based on the top query suggestions), Bloomreach applies the same merchandising logic. This means that the merchandising rules also apply to product suggestions in the same way as it does for product search.
Product suggestions are not impacted by feed drop and is refreshed every 2 hours provided a query is available to create a relevant suggestion.
What is the caching refresh schedule for product suggestions. How is that different from product search?
The caching for product suggestions does not change by any feed drop or rule change like it does for product search.
Updated 9 months ago