Use Search's Fine tuning settings to refine how search works behind the scenes.
The Fine tuning search settings are broken into four sections:
- Search modifiers: Use these settings to choose autosuggest and search fuzziness:
- Control whether your search box autosuggests searches as readers type.
- Control how typo-tolerant your search.
- Search fields: Use these settings to control which article fields are included in the autosuggest and full search processes.
- Search field weights: Use these settings to boost the relevance of each field relative to the others. Use higher weights for more important fields.
- Preview field weight changes: Use a test search phrase before and after tweaking Search field weights. This lets you review the effects your changes have before you save changes to take them live--and see just how much that relevance score changes!
Refer to the more detailed instructions below for more information on each setting.
The Autosuggest behavior controls what kind of search is performed as readers begin typing in the search box, before they hit Enter or select the search icon to run a full search.
When autosuggest is enabled, readers will see up to 10 relevant results in a dropdown below the search bar, for example:
![](http://dyzz9obi78pm5.cloudfront.net/app/image/id/67ab8b3faa77475f330392c8/n/live-kb-sample-autosuggest-search.webp)
To change autosuggest settings:
- Go to KB settings > Search and synonyms. The General settings tab opens.
- Select the Fine tuning tab.
- In the Search modifiers section, under Autosuggest behavior, choose one of the options. Refer to the more detailed information below to understand the different settings.
- Be sure to Save.
No reindex required
This change doesn't require a reindex, so you can save and retest as much as you'd like.
Autosuggest behavior options
Choose from one of these three autosuggest behaviors:
Autosuggest articles to readers while they type
Use this setting when you want the search bar to autosuggest results as readers type but don't need it to tolerate typos.
For example, entering "bran" will return articles with the word "bran" in them.
If a reader enters multiple words separated by spaces, autosuggest searches each word independently. For example:
- If I enter "synonym", search autosuggests articles that contain "synonym".
- If I enter "create synonym", search autosuggests articles containing either "create" or "synonym". Results with both "create" and "synonym" display highest.
Autosuggest articles and include fuzzy search results
Use this setting when you want the search bar to autosuggest results as readers type and to tolerate typos.
For example, entering "bran" will return articles with the word "bran" and "brain" in them. This option will use your Fuzzy search settings.
With fuzzy search results turned on, the more precise match should display highest.
The downsides of fuzzy search
We recommend this setting with one caveat: for some customers with key search terms that are very close in spelling, it can increase the number of search results you get. Test it to be sure it's helpful rather than noisy.
Do not autosuggest articles
Use this setting if you don't want any autosuggestions to appear as readers type. Readers will need to hit Enter or select the search icon to complete a full search and see any results.
For more details on what happens under the covers in autosuggest, refer to Autosuggest search.
Fuzzy search controls how typo tolerant your search results are. The fuzzier the search, the more typo-tolerant your search results are. This can be a great way to handle regional variations in spelling, for example "color" versus "colour."
Using no fuzzy search can frustrate readers who aren't aware they're misspelling something. Using too much fuzzy search can make search results too noisy.
To change fuzzy search settings:
- Go to KB settings > Search and synonyms. The General settings tab opens.
- Select the Fine tuning tab.
- In the Search modifiers section, under Fuzzy search, choose one of the options:
- Automatically choose the fuzzyness level based on the search term length: This is the "smart" fuzzy option, where the number of typos to tolerate is determined based on how long the search phrase is:
- Search phrase of 0-2 characters: No typos or fuzzyness.
- Search phrase of 3-5 characters: One typo of fuzzyness.
- Search phrase of 6 or more characters: Two typos of fuzzyness. Refer to Typos in fuzzy search for more information on what counts as a typo.
- Allow for one typo: Handle one character difference between the entered and actual results. For example, if I enter "bran", I'll still get "brain" as a match. Refer to Typos in fuzzy search for more information on what counts as a typo.
- Allow for two typos: Handle two character differences between the entered and actual results. For example, if I enter "bran", I'll get "burn" as a match. Refer to Typos in fuzzy search for more information on what counts as a typo.
- Automatically choose the fuzzyness level based on the search term length: This is the "smart" fuzzy option, where the number of typos to tolerate is determined based on how long the search phrase is:
- You can choose to Disable fuzzy search for full page search results. This setting can be useful if you're using Autosuggest behavior with fuzzy search results and don't want your full search results to include the same fuzzy behavior.
- You can also select a Fuzzy starting point. This sets how many characters to ignore before fuzzy rules apply. This can help ensure that fuzzy search isn't applied to acronyms or super-short entries (which decreases how "noisy" it can be). Use this setting if:
- You aren't using Automatically choose the fuzzyness level based on the search term length and
- You want to establish a minimum length of characters before fuzzy search is applied. We recommend using at least two characters. For example, if you use a lot of three-letter acronyms that are heavily searched for, you may want to set a three-character fuzzy starting point.
- Be sure to Save.
No reindex required
This change doesn't require a reindex, so you can save and retest as much as you'd like.
Typos in fuzzy search
Fuzzy search handles four types of typos:
- Changing a character (box → fox)
- Removing a character (black → lack)
- Inserting a character (sic → sick)
- Transposing two adjacent characters (act → cat)
Use the Search fields section to control which article fields are factored into your autosuggest and full search results (search results displayed once a reader enters a search and hits Enter or selects the search icon).
To change your search fields:
- Go to KB settings > Search and synonyms. The General settings tab opens.
- Select the Fine tuning tab.
- In the Search fields section, use the Autosuggest searches boxes to choose which fields your Autosuggest search uses.
- In the Search fields section, use the Full searches boxes to choose which fields the full search results page uses.
- Be sure to Save.
No reindex required
This change doesn't require a reindex, so you can save and retest as much as you'd like.
Search field weights determine the relevance each article field has in search. Our search relevancy scoring algorithm will adjust the order of search results based on the weights you set here. Changing your search weights allows you to boost the relevance of individual fields, sort of like putting your thumb on the scale for that field.
All fields start with a weight of 1 and have equal relevance in search.
For example, say that you set article Title to a weight of 10 and keep the other fields at 1. The search relevancy scoring algorithm weights matches in the title as ten times more important than other matches. Articles with a matching title will appear higher in your search results.
To view and update your knowledge base's current search relevance weights:
- Go to KB settings > Search and synonyms. The General settings tab opens.
- Select the Fine tuning tab.
- In the Search field weights section, use the dropdowns to view or change the search weight for each available search field: Title, Body, Permalink, PDFs, Meta description, and Search phrases. If you've unselected certain Search fields, adjusting their relevance shouldn't impact your search results.
No tags
Tags use a different search mechanism and don't impact search weights or relevancy scores. Refer to Tag Searches for more information. - Use the Preview field weight changes to test your changes by running a search for a term or phrase before and after adjusting the weights. This lets you test without actually saving changes, and to see the change in the relevance score as you make changes. Refer to Preview field weight changes for more information.
- Be sure to Save your changes.
Looking to tweak your search results even more? Refer to Optimize search for more help.
Wondering why a certain phrase or term is returning the search result order it's returning, or how to change that order?
Use a few test words or phrases in the Preview field weight changes section and run a search before and after making changes to review how those changes impact your search results.
The preview displays the top 20 search results, including their search relevancy score, so you can run it before and after making knowledge base-wide or article-specific changes to review how those changes impact search results.
This preview can be used in two ways:
- Before and after changing Search field weights.
- Before and after making changes to a given article, by editing any of its Search fields.
No reindex required
These tests don't require a reindex and don't impact your live search results.
To use Preview field weight changes, we recommend beginning testing BEFORE you make any changes to your search field weights or articles:
- Go to KB settings > Search and synonyms. The General settings tab opens.
- Select the Fine tuning tab.
- In the Preview field weight changes section, enter a Test search phrase. Hit Enter or select the magnifying glass icon.
- The Top 20 search results section updates to display the top 20 results and their search relevance score. Take note of the order of results and the scores or grab a screenshot for comparison.
- As you make changes to your Search field weights or to a specific article, rerun the same test search(es) to review how your results and the relevance scores have been impacted. You don't have to save changes to your Search field weights to test them, but you will have to save changes to individual articles to test them.