Home < Technical < Product Overview < Related Searches
Related searches
Related Products and Searches is a product that recommends popular search pairs.
The insight it provides is incredibly valuable as it’s based directly on user data
and tells you exactly what your webshop’s visitors want to see and how they think.
Related Products usually appear on 0-result pages so users can easily navigate to
other products that fit their needs, but can also be configured to appear when results for a query are found.
Related Searches encompasses both search suggestions and product suggestions and can be
displayed on search result pages (SERP) and 0-result search pages. When a user searches
for a product and results are found, Related Searches places smart keywords above and
below the results. For difficult, long tail searches when the search engine cannot
display results, Related Searches places keywords and product suggestions
(with images) on the 0-results page to help users find products that match their initial intent.
Related Searches mines and refreshes content based on user behavior; since it is continually learning,
recommendations are always up to date and improve over time. For example, in the summer,
it would learn that “air conditioner” is becoming a more popular search and start to
recommend it. In the fall, as searches for “air conditioner” decrease, it would no
longer be a recommended search or product.
These product and keyword suggestions help users easily navigate to other products, serve as an effective
way to disambiguate user intent, refine searches, and recommend related products.
If you are not ready to replace your entire search engine just yet, this is an easy improvement
that yields quick conversion gains and saves shoppers from a bad search experience.
0-Result Page Keyword and Product Suggestions
0-Result Pages are one of the worst experiences users can have on your site,
especially when you sell the product they are looking for.
Typically, web stores have a 30% 0-Result Rate, which is high. And we mean really
high - 1 in 3 searches won’t display a result at this rate and this
often causes customers
to leave the site, resulting in missed business opportunities.
When no results are found, Related Searches suggests keywords and products to users
so they can refine/reformulate their query in order to find the initial
product or another that suits their needs.
The suggested keywords are reformulations of the initial search, which serve as an
alternate way to express user intent in case the product users are searching for
is listed under a different name or the initial query contained a typo.
Related products are displayed if the search engine fails to find results
when attempting to match the query text to product details. The main search
engine tries to match queries exactly to products, while related products
broaden the search in order to recommend products the user may have intended
to search for. For example, if someone searches for “10GB USB stick”, which
the shop doesn’t carry, there would be 0 results, but the reformulation
would provide alternatives like 5GB USB
Search Engine Results Page Keyword Boxes
Related keywords on the Search Engine Results Page (SERP) help users
refine searches and discover more products related to their initial
query. For example, if someone searches for “door”, Related Searches
may recommend “front door, wooden door, sliding door, door handles,
door paint” as keywords and products. For example, if a user visits
www.auchan.hu (a grocery store), searches for “tomato”, and adds that
to their cart directly from the SERP, Related Searches will suggest
similar keywords, such as potato or pepper, so the user can quickly
navigate to other products. This increases the average cart value by
directing users to
more products; it appears even when a user’s original search yields results.
The keyword boxes on the SERP help users reformulate their initial
query. Since the software learns from user behavior (based on click
through rates) the most relevant results are displayed first. Without
these suggestions, users have to manually type in a new search, sort
through attribute filters in order to find a more specific product,
or page through seemingly endless results to find what they’re looking
for. This isn’t very effective because people give up looking for a
specific product if they don’t see it on the first page - ideally as
one of the top results - they will look elsewhere.
SERP keyword boxes help users reformulate their search with just one click.
It is a good practice to put these navigation boxes both at the top and
at the bottom of the SERP. We recommend only placing a few suggestions
(4-5) at the top of the SERP and many at the bottom since this is where
the user will end if they don’t
find what they’re searching for and this gives them another chance to find a product that fits their search.
Calculation
Lists of related keywords and products are generated daily based on user behavior. If a user searches for X,
but doesn’t click on a product and then searches for Y, X and Y are then considered related or tagged as synonymous.
If someone clicks on a related search suggestion, that product/keyword’s relevancy
would be boosted (by an amount you set) and would therefore be more likely to appear
as a synonym for the original search. You can set how often
those 2 terms search pattern must occur before it appears as a related keyword/product.
Unlike our other products, the more recent actions are not
weighted more heavily than those occurring farther in the past.
Ranking
Simply because one person executes a search pair does not mean it’s automatically added to the related searches list.
You can set the number of times a sequence must be searched before it appears as a related keyword/product.
Our Related Search boxes can be placed at the top, bottom, or to the side
of the SERP. You can opt to have all these boxes, 2, or just 1 and can set these preferences in the portal.
Settings
When determining which product/keyword to display first on the list of related searches,
the frequency of the search pairs comes into play.
If people search for “computer” then “laptop” as well as “computer” then “Macbook”, and the “computer-laptop” sequence is
more frequent, then the list of related searches for “computer” would be “laptop” then “Macbook”.
Related Products pulls data from keywords and clicks when determining
suggestions. Keywords that lead to 0-results are automatically not included.
If someone searches for “computer” then “laptop” and clicks on a specific “Macbook” product, that “Macbook” would linked
to the search term “computer” and would appear as a recommendation on subsequent “computer” searches.
Keyword suggestions are based simply on searches, while
product suggestions are calculated by searches and product clicks.
Insights
If a certain query frequently leads shoppers to a 0-results page,
you can find out what users search for next and rename products
accordingly, so they appear for the
first query shoppers usually type – this will help drastically reduce your 0-result page display rates.
Mobile Navigation
When implementing Related Searches on your mobile site, we recommend
displaying fewer related products and searches than you do on your desktop
site. For example, desktop sites usually have 7 keywords on the top and
11 at the bottom of the SERP, but on a mobile site you should only have 4 at the top and 7 at the bottom.