Are Dealer Sites using Natural Language Search?

[ryan.gerardi]

Was doing some research today about Natural Language Search and got to wondering if any dealer sites are utilizing this?

NLS understands user intent and matches the same with attributes in your product catalog (vehicle inventory) and can provide high relevance.

Search filters are common on some platforms, I know, where users can turn dials and slide levers and input filtering criteria. NLS would accomplish the same thing through the search bar, via voice and text.

I did some test searches on Amazon and as expected, the more info you place into the search bar, the less likely you are to receive a result.

For instance, searching 'jaybird x3 sparta noise cancelling wireless' returns 0 results when in fact they do have such a product.

NLS is known to improve search results. Which got me wondering, could searches on dealer sites stand to be improved, or is it a non-issue?

Thoughts...

 

[Alexander Lau]

Was doing some research today about Natural Language Search and got to wondering if any dealer sites are utilizing this?

NLS understands user intent and matches the same with attributes in your product catalog (vehicle inventory) and can provide high relevance.

Search filters are common on some platforms, I know, where users can turn dials and slide levers and input filtering criteria. NLS would accomplish the same thing through the search bar, via voice and text.

I did some test searches on Amazon and as expected, the more info you place into the search bar, the less likely you are to receive a result.

For instance, searching 'jaybird x3 sparta noise cancelling wireless' returns 0 results when in fact they do have such a product.

NLS is known to improve search results. Which got me wondering, could searches on dealer sites stand to be improved, or is it a non-issue?

Thoughts...

Are you asking if dealer website providers (those that actually have a search mechanism) take this into consideration? I doubt it. <OR> Are you asking if the content that is placed on dealer websites qualifies or triggers such searches? I doubt it but it might be layered in sitemaps (XML). I could be wrong...

Sounds like a good idea though!

I posted this not too long ago: https://forum.dealerrefresh.com/threads/are-you-optimizing-your-content-for-voice-search.5849/, I think a bit of that was mentioned.

 

[Ryan Everson]

Dealer Inspire is currently rolling out Voice Search functionality on their websites!

 

[Alexander Lau]

Dealer Inspire is currently rolling out Voice Search functionality on their websites!

Boom! There you go, DI leading the way.

 

[ryan.gerardi]

Well, voice is only an input method. It isn't necessarily "leading the way" if the search technology isn't also.

Micro-specific searches like the example I provided above on Amazon are said to be more converting than generic queries, at least in the context of e-commerce searches. With current e-commerce search technology, such as what Amazon uses as do most dealer sites, the more specific a user queries, the worse the relevance is. This leads to sellers losing out on high converting shoppers.

For example...
On this Dealer Inspire site, there are 3 vehicles that match '2015' + 'bmw' + 'lexus' but if you enter '2015 bmw lexus' into the search bar you get 0 results.

Imagine wanting to search something like '2012-2015 bmw 5-series and lexus es below 30k with less than 50k miles' on a dealership (or aggregate) site.

Most aggregate sites offer a bunch of drop down menus checkboxes:

• https://www.autotrader.com/cars-for-sale/
  
• https://www.cars.com/for-sale/advanced-search/
  
• https://www.cargurus.com/

Google returns a list of results, essentially, some relevant pages.

So the question is, would NLS technology be a benefit for auto shoppers on dealer sites and/or the aggregate sites?

 

[Rick Buffkin]

I know DI has the voice search capabilities for inventory. Is it getting much action? I don't know. Just testing the demo site, once you enter one or two searches, you quickly learn how to speak your search so the system understands and get more results returned.

https://chevydemo.dev.dealerinspire.com/

Is this a NLS feature???? I think it's keyword optimized voice search. Not NLS. I think true NLS is more context focused instead of keyword / content focused.

Is this a pretty forward thinking and vast improvement for Dealership websites. 10,000% YES. If we are trying to compare DI inventory search platform to Google or Amazon Alexa, I think we're all in for a big disappointment at this current moment in time.

 

[ryan.gerardi]

@Rick Buffkin - DI is not using NLS. As you noted, NLS is more context focused. It uses semantic technology and machine learning to detect user intent. Whereas conventional search uses stringed queries.

Google uses NLS, I'm pretty sure. And e-commerce sites are beginning to. This is why I thought of dealers. Vehicle inventories have a ton of detailed information. With NLS, it seems like dealer sites could provide more robust search for users, for vehicle inventory and with parts inventory.

 

[Dave Rowe]

Correct - we're (DI) not using NLS - however, we do index quite a bit, which is making the new search interface much more robust (as you've seen from our Dev/Demo sites posted here ). For instance, combining fields provides more contextual relevant results - searching vehicles with specific accessories / options. While it's not perfect, it's something we continually refine as we see results from the analytics.

Something to note though, while we can push this forward - the consumers are far from using it this way. Our analytics are showing people searching with barebones text. On a Mercedes site - the top search query "glc" or "amg" - nothing nearly as in depth as "vehicles under $10k with less than 100k miles" - users just aren't searching like that.

This makes it crucial for us to have easy to use tools for building out listing pages matching terms they trust Google to find with the more contextual, deep, queries.