This is the third post in a series about Magento EE’s integration to Solr for search. If you have not read part one and part two, I would take a minute (or thirty) to catch up. This post simply ties a bow on the series by walking you through, visually, how each feature works.
There are six products that are set up to be searchable. Pay close attention to the “Name” and “SKU.”
MarketingSherpa data reveals 43 percent of visitors who land on a website go immediately to the search function.
There are no search terms yet in the system.
When searching for the word “shirt,” there are three results returned.
When searching for the word “one,” there are two results returned.
After our two searches, there are two search terms in the system.
Now that there are search terms in the system, auto-completion is available. After typing two letters, Magento will show recommended search terms that begin with those characters.
If we want a different word (“top”) to bring back search results for “shirt,” we can use synonyms.
Synonyms also work for auto-complete recommendations.
Although the word “top” is not in any of the product attributes, because it is a synonym of “shirt,” it returns the same results as “shirt.”
The count for “Number of Uses” will increment for “shirt,” even though we are searching for “top,” as long as “top” is not added as its own search term.
You can remove synonyms if they are not producing the expected results.
You can manually add new search terms directly into the tool without searching for it on the front end. Here we add “shirt” as a related search term.
The search for “top” yields no search results, but the word “shirt” is shown as a related term.
Solr will return spelling suggestions that can be offered as possible search terms.
If you do not want to return search results, but instead would like to redirect to a specific page (any page), you can use the Redirect URL field.
Searching for “shirt 3″…
…will redirect directly to the Shirt Three product page.
Although this search term has the word “shirt” in it…
…it will not show up in the suggested terms in auto-complete because it only will return suggestions that begin with the text in the box.
We have configured partial SKU searching. Solr returns results for the SKU “ShirtOne” when searching for “ton.”
Changing the SKU to “TheTee.”
Because we have partial name configured in Solr, a search for “irt” will return any product with the name “Shirt.”
When searching for the word “on,” both the spelling suggestions and related search terms kick in.
After clearing out the search terms…
…only the spelling suggestions will displayed.
The reason that “none” is returned as a suggested spelling is because the “Tax Class” attribute is indexed into Solr. All of the products are set to have a tax class of “None.”
After changing the tax class on one of the products…
…only five results are now returned.
Each search term can only have one synonym configured. However, multiple search terms can have the same synonym.
If the search term is “tee,” “top,” or “tshirt,” it will bring back results for the synonym “shirt.”
The management of search terms, and the refinement of search results is not a “do it once and you’re done” type of process. It is involved. It is tedious. It is never-ending. But it is worth it.
“MarketingSherpa data reveals that 43% of visitors who land on a website go immediately to the search function. In addition, customers who use the search box on e-commerce sites convert at nearly three times the rate of general browsers. Let’s take that data at face value and run some numbers. An expected, conversion rate from most lead generation activities runs in the 2% to 4% range. On a site with a smoking site search capability that shows the visitors what they want to see will nearly triple that rate to 6% to 9%. Let’s say the revenue accrued from the normal lead generation visitors was $500,000. The incremental lift in ROI from your effective site search capability would add another cool million to the bottom line. True, you would have to go back to the attribution model and figure out what percentage of that would be attributed to site search … if your company even included site search in those calculations. Most don’t.”
Note: Much of this article was supplemented by Magento’s actual User Guide.