Another month, another new addition to OnVoard's product suite. Here's a short backstory on how I came up with this new product.
Before I started OnVoard, I was working on a food review startup called Eatasy to help make food search easier than flight search. I will leave the part on why I decided to move on from that in the future (probably in another post) but one of the key design is to automatically split long-winded reviews into individual phrases highlighting key points.
How a long-winded review looks like.
Most people wouldn't even bother reading past the first paragraph. Putting myself on the reader's perspective, I know that most visitors would only read up till the first 3-4 lines of a review. They want to know in the most concise manner why a place is recommended. And no, they don't care about how you and your spouse first met at that cafe in downtown.
So this is what I did is to create a summarized version of each review.
- Split reviews into individual sentences
- Measure emotions and sentiments of each sentence
- Extract sentences with high emotion index or extreme sentiment values
- From extracted sentences, our system will automatically pick up to 5 sentences to display based on the calculated score.
Here's how the final summarized version looks like on Eatasy's app. Notice how each point highlights key features of the place. Not too shabby right?
Extracting key highlights from a review makes information much easier to grok.
Back to OnVoard
One of OnVoard's key market segments is local service businesses. While prospecting for new customers in the air-con servicing space, I noticed that many of them are social proof notifications to show stats for recent purchases/visitors.
A stroke of brilliance came to my mind. Why not extract key highlights from existing customer reviews and use it as social proof? After all, the whole concept of social proof is to build credibility and nothing beats authentic customer reviews in convincing visitors that your product is tried and tested.
By adding a popbox snippet to your page, OnVoard is able to automatically extract positive quotes from your existing reviews and render them as social proof.
PopBox supports 2 types of notifications.
Rating Notification - Show listing's overall rating details in notifications to convince visitors that your brand is tried and tested by other users.
Quote Notification - Show quotes extracted from reviews to highlight key aspects on why your brand is chosen by reviewers.
We know that predictions may not always be accurate in sentiment analysis and there may be negative quotes that are incorrectly classified. Instead of leaving the grunt work to machines, you can opt to only display quotes that are manually shortlisted. This will give you full control over quotes that are shown on your site.
We want to keep things simple and easy to use for the start. There are some ideas on how to take things further for the future.
One of the initial concepts was to be able to route and show review quotes or listing details for outlet nearest to the visitor. For example, we know that Pizza Hut has hundreds of stores in many cities. Instead of displaying reviews from a random store, why not find the store nearest to the user's location and select review and listing from there?
We didn't implement this concept because:
- It will obviously be much more complex to implement and setup
- We don't have enough knowledge as to whether third-party social proof notifications would attract sufficient interest from large multi-chain businesses. From what we know, most sites that use social proof notifications are run by small businesses or are in the e-commerce space.
Another concept was to consider adding frequency and recency cap to notifications. However, this will introduce tracking concerns that can be easily avoided.
Of course, there's a possibility that none of these concepts will be implemented due to negative benefits-to-complexity tradeoffs.
For now, take a look at a live demo for OnVoard PopBoxes and let us know what you think.
For the next post, I'll be laying out the technical implementations of OnVoard PopBoxes and some of the decisions made in building our Widgets library.