Google’s BERT algorithm, released last week, is the company’s latest effort to understand the intent of search queries. For ecommerce sites, the unique details on product pages could now drive more organic search traffic with high purchase intent.

BERT Algorithm

BERT — Bidirectional Encoder Representations from Transformers — is an open-source algorithm from Google to process each word in a search query relative to other words in that query, versus one-by-one in the order they appear.

Among other things, it does a better job of evaluating prepositions — such as “to” and “with.”

BERT doesn’t replace RankBrain, Google’s 2015 effort to understand searchers’ intent. Instead, both RankBrain and BERT, applied simultaneously, can help decode intent and deliver the most relevant search results.

Google offered a couple of examples, including the one below for the query “2019 brazil traveler to USA need a visa.” Previously, Google would have shown organic listings for both U.S. and Brazil visas because it didn’t understand the importance of the preposition “to.” But now, the results show only info for travelers to the U.S.

Before the BERT algorithm, the query "2019 brazil traveler to USA need a visa" would have shown organic listings for both U.S. and Brazil visas. But now, the results show only info for travelers to the U.S.

Before the BERT algorithm, the query “2019 brazil traveler to USA need a visa” would have shown organic listings for both U.S. and Brazil visas. But now, the results show only info for travelers to the U.S. Source: Google. Click image to enlarge.

With BERT deployed in the U.S. by Oct. 25, Google’s organic search results and featured snippets are closer to providing searchers with relevant results for more complex queries, such as long-tail ecommerce searches.

Product Pages

Product pages and filtered product grids tend to rank better for long-tail queries. Thus those pages have the most potential to benefit from BERT. The importance of prepositions applies especially to product attributes, such as color, size, and material.

Imagine all of the product details that shoppers search for. Those details frequently include product names and attributes connected by prepositions, which I’ve italicized in the examples below.

  • “plantation window shutter with solid oak wood;”
  • “beaded 10-foot curtain with rainbow real crystal beads;”
  • “diamond solitaire engagement ring with rose gold band leaves.”

How your product pages express these details is even more important with BERT. It’s an opportunity to optimize the descriptions of your best-selling products with content that no other site uses.

Based on keyword research, you already know some of the product attributes that your searchers are interested in. Shoppers are telling Google what they want to buy from their search queries. Google shares that data in its Keyword Planner.

All you have to do is include attributes on your product detail page descriptions. For example, to win a search for “plantation window shutter with solid oak wood,” a store could make clear in the description that its plantation shutters are made of wood.

Moreover, the store should list every available finish for those shutters. How it does this is critical. Color-swatch images don’t help search engines. And listing only color names doesn’t help shoppers. What’s needed is both: a visual for shoppers and a text label for search engines.

Another BERT-friendly strategy to target long-tail queries involves harnessing filtered site-search — those thousands of filtered product grids stemming from product attributes, such as “rose gold” and “solitaire” for a ring, as in “solitaire ring in rose gold.”

Manage the strategy carefully, however. Index only pages with high-value filters. For example, according to Keyword Planner data, ring filters of “yellow gold” and “rose gold” have much more organic search value than a price filter of “under $1,000.”

Relevant Traffic

For search queries containing both a product and attributes, Google vacillates between showing product detail pages and filtered product grids. As queries become more complex — such as those with attributes — the search results could contain both types of pages. But product detail pages tend to rank higher.

Google’s search algorithm still doesn’t fully understand the nuances of human language. But the BERT algorithm helps. With its ability to evaluate the entire query, including prepositions, BERT provides ecommerce sites an opportunity to drive more relevant organic search traffic to product detail and product grid pages.

Source link

Leave a Reply