skip to Main Content

Autocomplete, also called autosuggest or incremental search, improves the user experience by making it easier to execute searches by suggesting words and phrases that a matching algorithm determines are appropriate based upon the characters entered into a search box. Google has had autocomplete in their search box since 2008. Following in their footsteps, any company that values its customer experience has implemented autosuggestion in their onsite search. But there’s a new trend in the type-ahead game that we’ve begun to see cropping up on more and more company websites: Instant Search.

I’m not sure if we’ve landed on the commonly accepted term for this function but I’ve chosen Instant Search because that’s what Google called their “search while typing” feature. Basically, it works like this: You start typing in a search box. Instead of providing you a list of keyword suggestions, the search engine starts returning search results. Actual search results are generated by the search engine. This happens in the same window where you’d expect to see keyword suggestions.

Is this a good thing? I’m not sure. First, it’s a bit of a leap cognitively. Google and practically every search engine out there have trained us to see keywords when we type ahead. So when something other than keywords appears it takes a moment to figure out exactly what we’re seeing. If you choose to display search results you have to help me understand that I’m seeing something different and help me figure out how to use it. You have to help me make the cognitive leap without adversely affecting my journey. Screw things up and I go back to Google. That’s where your competitors live. Don’t do that.

Instant Search in the wild

One of our clients, a B2B company, briefly adopted this function with their new search engine. Unfortunately, they had two things going against them. First, the out-of-the-box design wasn’t very good. It only displayed titles from the search engine. Second, it displayed the top ten search results, so the relevance of most of those links was low. The cognitive disruption from using the function was high. But was it just design?

Search results have to do two basic things:

  1. Provide a list of highly relevant links.
  2. Give me a hint as to which one I should click to get to my answer

The Instant Results that our client presented had titles that weren’t very illuminating. They have lots of detailed technical documents with similar titles. Without snippets, there were no hints about what which of those results are right for the searcher. Ten links were presented all of which weren’t helpful. The default action of a user was to click through to the actual results page where they’d get enough information to make the right decision. This problem isn’t unique to our client. many companies have equally challenging content. We’ve seen a handful of websites using Instant Search with equally disappointing user experiences.

We worked the client to deploy a much better solution that autosuggested keywords and keyword phrases weighted for search success. (If you want our help too, reach out for a free consultation). Their search success is trending in the right direction. But that left me wondering if there was a way to make Instant Search an effective user experience. It took us a while but we found one.

Instant Search done well

On your desktop (this doesn’t work on mobile devices), go to and start typing the word “migraine” in the search box. Here you get the best of both worlds. You get instant search and suggested keywords. This design feels effective for several reasons

  1. The search results are good. High relevance.  Quality content. Call to action.
  2. The design choice is excellent – they chose the top three results, not 10 results. The choices are limited but because the results are quality I don’t feel like I’m missing anything. You can only get away with this if you have confidence in the quality of your search results. By the way, with this sort of confidence, they’re probably well suited for the voice search world, but that’s another topic.
  3. The design choice is excellent – they left the traditional autosuggest in place with three flavors: 1) a few faceted search suggestions, 2) a standard dictionary type-ahead, and 3) popular (trending) related search terms
  4. They link to the product page if I want to know more.

The big question is whether, even with an excellent design, this will be an effective user experience in the long run. I’m sure CVS is measuring these results. They may even be A/B testing this experience against a traditional experience to determine which delivers the best results. Running fact-based experiments will help them tune this experience over time to be the most effective it can be for the user. That should lead to better business results.

Boon or Bane?

One of the reasons that Google abandoned Instant Search in 2017 was that the capability didn’t exist on mobile devices. With more users defaulting to their mobile devices Google decided to discontinue to develop the desktop Instant Search function. That made sense for Google. Will it make sense for all companies? They all face the same device trends but perhaps some find the economics compelling

In a B2C experience, I like the fact that the company is trying to get me where I want to go as quickly as possible. I also think that it’s relatively easy to know what the most popular results are for a broad range of products. One of the challenges with this sort of experience is that it makes it difficult for the consumer to discover new products. Whether those three products are chosen by popularity, relevance or merchandising (or some combination of the three), it does limit choice. But that’s not a new problem nor one that’s going away soon. For now, I like what I’m seeing out of companies like CVS. Help me get where I’m going quicker. It’ll be good for both of us.

Next week we’ll talk about the sins of tiled search results. Yes, I have an opinion.

Have a great week.

Back To Top