Structural content strategy and taxonomy can support, and are impacted by, voice search. By considering search terms, SEO, taxonomy, and the on-page structure of content, we can improve the voice search user experience.
Some definitions are needed for this topic. It’s important to be clear with the terms we’re using so we can build and support the right thing. For your edification:
- Voice activated search: Using search by saying, “OK google” or “Hey Siri”. Using your voice to start the search functionality on a device such as Google Mini or Amazon Alexa or on your phone. Say “OK google” and then say your search term.
- Voice enabled search: Using search by activating the onboard microphone manually (in other words, by pressing a button), then saying the search term.
- Conversational search: A progressive search where each search term builds on the last, such as, “How old is Keanu Reeves?” (54 years old) “Where was he born?” (Beirut) “What’s the political situation like there?”
- Natural language processing: How technology processes unstructured content and is focused on enabling computers to understand and process human languages, to get computers closer to a human-level understanding of language.
- Natural language understanding: The post-processing of text, after the use of NLP algorithms (identifying parts-of-speech, etc.), that uses context understand the meaning run-on sentences or sentence fragments.
How NLP Plays Into Your Website
It’s most likely that your website will take advantage of Google’s NLP, Amazon Web Services, or you could install Standford’s CoreNLP. However you manage it, the NLP application works its own magic separate from the search indexing and it works on top of your website.
Some scenarios you would need to consider:
- What happens when a person uses a voice-activated devices Google Mini and asks a question? How do we get the content on our site into the answer that the devices might give? (Off-site search)
- What happens when someone uses voice-enabled search (a microphone in the search box) on our site ? Do we display text-based search results? Can our NLP application read out an answer? (On-site search)
The two scenarios are important because the first one happens entirely off your website and the second one happens on your website and is more within your control.
Off-Site Optimization for Voice Search
Standard SEO practices are still the main way to improve your content for search engines, regardless of whether the search term is typed or spoken. However, when using voice search, people tend to have longer search phrases.
Instead of “Keanu Reeves age,” they might ask, “How old is Keanu Reeves?” Instead of “Keanu Reeves films,” they might ask, “What was that film where Keanu was the voice of the cat?” (FYI – the answer is Keanu.)
Review your search logs for searched terms on your site to see what long tail search terms are and if there are any themes to the questions. Then optimize your content for these terms, if applicable. Add these keywords and phrases to your taxonomy. Ensure your NLP application can access your taxonomy.
On-Site Optimization for Voice Search
Short paragraphs, plain language, and inverted pyramid style: these have always been important for writing on the web. This becomes even more important when voice-activated search devices are looking for an answer to a question.
Update each page with a brief, 30-word or less, summary of the content on the page that either answers a question that someone might ask or defines a term. This text should be at or near the top of the page. Both on-site and off-site NLP can leverage this short piece of text by either displaying it as a top result or reading it aloud.
Ensure your content is properly tagged with taxonomy to help the NLP application’s indexing of content and the relation of synonyms and keywords to taxonomy terms.
In addition to this, the NLP application you use may have some requirements for content.
As a note, currently there is no metadata schema for voice search that you can integrate into your website metadata.
Impact on Structural Content Strategy
Supporting voice search should ultimately be driven by web and content strategy. Your structural content strategy can support on-site voice search.
- NLP does separate search indexing from the text-based search on your site, so your structural content strategy may need to be assessed to ensure the taxonomy and metadata needed from the NLP application is available.
- Taxonomy can be enhanced with synonyms and keywords or keyphrases to improve the match of searched term to the content.
- Taxonomy can be further expanded into a knowledge graph to enable the NLP application to better understand your specific domain and return (and recommend!) more relevant results.
- Voice search can support accessibility by providing alternatives to users who have problems typing, who experience low-literacy, or are poor spellers.