The future of search is conversational
A brief reflection on the future of search as conversational search, and the benefit of a healthy relationship-network-data loop. By Omri Klinger, Hadar Landau, and Guy Adam Ailion
Search 1.0, The Engine
Along came the internet and everything changed. A forever expanding database of information or depository, disparately stored online. But we needed to access this library. The ‘Search Engine’ was the first winning use case of the internet. An agnostic and accessible consumer-facing interface to find what you were looking for; Google, Yahoo, IE. But when you throw the net wide you catch a lot of rubbish and a lot of the wrong fish.
Inch wide peep-holes a mile deep.
The next winners were the vertically-focused mini search engines — the industry-specific aggregators; Zillow, Glassdoor, Ticketmaster, Expedia, Lastminute.com. These websites were large search portals with a narrow focus. They were inch wide peep-holes, a mile deep. Offering you more options but in one subject.
A search bar is a blunt tool.
Even with filters, the current state of Search is dogmatic and presumptuous. This is a fait accompli when the only data a person can provide the interface is shallow and reduced to predetermined filters and keywords.
Simple data in = simple data out. Search is therefore limited to a simplistic understanding of your motivation. It’s blind to solutions outside its small point of reference, blind to your intrinsic motivations, blind to your emotional compass and knows nothing of what we call ‘gut feeling’. It is just a blunt tool smashing through content.
Not only a logical problem but a psychological one.
The next winning use case of Search will be one that understands the context and meaning behind both the enquiry and the enquirer. A machine that understands you and your motivations can help you find the most relevant result quicker and without unwanted clutter. This search tool will be sharper and smarter to help you search difficult queries with more accuracy.
Searching for a home, for example, is not just a logistical task but a big emotional decision to make. A decision where we try to balance our fiscal and logistical priorities with our reality, our aspirations, our dreams and our emotions. A true conundrum for many of us! Therefore, the machine that searches for your home has not only a logical problem to solve but a psychological one too.
Conversations with the internet
This new interface will be conversational — Conversations have vast and expressive powers and are imbued with meaning, context, and data — A conversation is both the ‘signal and the noise’. By engaging in conversation the machine gets deeper and better quality data about the enquiry and the enquirer. Most importantly, the machine will have the ability to also ask questions. The machine no longer waits for input; it seeks it.
Good data in = good data out. Once the machine knows you well you can guide it and use it to consider all results specifically tailored for you.
Our instinct to speak.
Language is so natural to us that we can easily forget how truly miraculous it is. Language is the original medium for humans to exchange information, coordinate action and share thoughts with each other. We are born with the instinct to speak before we learn to write. This is why it is the natural evolutionary interface for the future of search. When something is native to us, we feel more comfortable and we share more.
Conversational Search is like a Q&A with a search engine. This format gets you closer to what you want not only because you guide the engine, but because you learn more along the way. Conversational Search is essentially a data feedback loop. Capturing conversations as the data input drives faster improvement of the machine and more personalized service.
The heart has a reason of which reason knows nothing.
However, human decision making is more often than not irrational and unpredictable. This is because we are social primates who are deeply contextual and driven by our emotions. The machine must understand that providing more data does not always mean better decisions. For example, just showing more house listings does little to enable the enquirer to make a smarter decision. It actually makes it harder to decide because of the paradox of choice.
Complex questions, such as ’is this a good deal?’ will not yield a linear path to the answer. In fact, logic may not always apply because the enquirer may not always know what they want, or say what they mean. Or as Blaise Pascal wrote, ‘’The heart has its reasons which reason knows nothing’’. In our example, the real question the enquirer may be asking is, ‘is this a good deal for me?’.
As we mentioned before, there are logical problems and there are psychological ones. Language is therefore just the tip of the iceberg of the non-linguistic systems at work under the hood of the new Conversational Search. The machine must seek quality input such as context and emotional motivations by earning user trust. Meaningful data in = meaningful data out.
A relationship-data-network loop.
3 steps to a relationship-data-network loop.
Step 1. Through conversations one can build rapport and rapport creates a relationship.
Step 2. When you have a relationship with someone or a thing, you can start to build a foundation for earning trust.
Step 3. Trust is earned when you can provide value or expertise that is relevant, accurate, honest, unbiased and meaningful.
When you have a good relationship and trust is established a magical thing happens — you share more, and you share openly. By focusing on the relationship and earning trust, users provide more unique and personal data that is better quality and more meaningful. This allows the machine to paint a contextual picture to better understand them and what their needs are.
The machine can now better serve the enquirer, guiding them to the most appropriate results with a better service and experience. A better service and experience means a better product. This creates a flywheel that brings more users and more users mean more unique data: a relationship-data-network loop.
