Conversation Mining: a UX Writer’s Secret Weapon Against the Coronavirus

Kevin Pichinte
UX Planet
Published in
6 min readMay 18, 2020

--

The worlds in sync. Everyone’s worried. Yet, readers received a notification asking them if they “want the coronavirus” in March. That’s a risky click!

Do you remember when residents of Hawaii received an AMBER Alert notification about an incoming ballistic missile in 2018? That single false notification claimed a life.

To empathize with people, I started conversation mining. This method offers a front-row seat to understand the way users think, talk, and engage with digital products.

A Pandemic Message Going Wrong

Before I continue, I want to EMPHASIZE that nobody knew the coronavirus would be a pandemic. The Seattle Times was doing their best with the information they had. So, let’s continue.

A push alert (notification) is an opportunity for news organizations to reach readers, according to the University of Texas at Austin.

The Seattle Times reached many readers with this notification after it went viral on social media:

Did your gut tell you why? Hint: the CTA.

Many questions popped into my mind when I saw this. Was it the lack of an editor? Skipping Quality Assurance? Or the design? Maybe it was all three, or not, but we’ll get back to it soon.

So here’s how I used conversation mining and a couple of other tools UX Writers and design teams’ can implement to tackle any challenge.

Step 1: The Plan

I started with some desk research. I quickly found data from Media Solutions about the Times audience. I formed a hypothesis, business goal, and, you guessed it, mined conversations.

  • Mission statement: “An important part of The Seattle Times’ mission is making sure our storytelling and our newsroom is inclusive on issues of race, gender, age and more.”
  • Business goal: increase [5%] of users receiving health news updates.
  • Media Solution: readers 25–39 yrs, higher than average income, $150,000+ household income, etc..
  • Hypothesis: readers are more willing to “receive updates” about the latest health news to make the best decision for their health. It’s also commonly used in news.
  • Conversation mining: to gain a deeper understanding of how people (users) feel about the notification and the language they use.

Step 2: Conversation Mining

To get a 360-degree view, it’s good practice to gather as much as you can by looking at different social media sites, online forums, product reviews, and anywhere you can think of to get close enough to your target users.

Get positive, negative, and neutral comments to have a well-rounded idea of the way users think and how they interact with a product.

So I started by looking at the way users were interacting with the Seattle Times (the product):

The response is positive, and the majority are enjoying the free information the Times is offering since taking down their paywall (I’m not sure if they still are). That’s altruistic of them.

But just like any talented journalist, conversation mining requires a full circle report. Positive, negative, and neutral feedback from users.

I headed straight to Facebook and Twitter to see the issues users’ had with the notification. They quickly unveiled the elephant in the room: the microcopy.

Not only was I able to zero down on the negative, neutral, and comical feedback, but I noticed a common word in users’ vocabulary: “update”. Bingo!

Mining conversation also allowed me to understand the way readers think. They viewed this single notification as an “irresponsible” and “epic fail”.

THE MICROCOPY CANVAS

As you can see, I have an enormous responsibility to readers. To do them justice, I used the microcopy canvas and filled it out with data from Media Solutions:

This artifact can help writers explain the thought process to stakeholders, designers, and anyone else on the team; and validate the words in the UI, down to the last drop of digital ink.

If you want a more detailed explanation about how to use this artifact, look at Jane Ruffino’s article.

COPY EXPERIMENTS

COVID-19 is a hot talking point, so naturally, media ratings are rising and more people are reading news articles. But it’s important to be responsible because with a large audience comes great responsibility.

So, with all the data from Media Solutions and conversation mining, I started with the headline: “Sign Up for Coronavirus Updates”.

While the Seattle Times original headline is eye-popping, the coronavirus is a subsection of health news. Plus, its times like these that health is a top priority for people. Instead, I wrote:

“Latest Health News Updates”

The headline lets readers know the notification is about health news. Not just coronavirus. It’s clear and concise. It doesn’t beat around the bush, and it’s honest. And it incorporates the user’s language: Update.

I then played with the body text and narrowed down to two possibilities:

“Get tips to help you stay healthy and the latest coronavirus news from the Seattle Times.”

AND

“Get the latest coronavirus news updates from the Seattle Times.”

The language is simple, it’s empathetic to the situation, and the second body text is the same except for “update”(looking back at it now, it’s redundant, but that’s why iterating on your copy is part of the process).

Now, to the most controversial part: the CTA. Asking readers if they “want coronavirus”.

I changed the copy to “RECEIVE UPDATES”. And if readers want more information about privacy concerns or how it works, “Learn more” offers that option (changed to sentence case post-launch).

Visually Communicate Your Ideas

Visually presenting your work is key. So learning how to use design tools is important. But if you want to mock up something quick, use Whimsical.

I mocked up two similar examples:

And

I did this so people can visually see my work. Instead of writing:

<H1> Health News Updates

<Body text> Get the latest coronavirus news updates from the Seattle Times.

<CTA> RECEIVE UPDATES

I think you get the point.

USER TESTING

For this mini exercise, I A/B tested both versions with UX Designer Mar (I tried reaching actual readers but I had no luck).

My question was simple (business goal), which notification intrigues you the most to click “receive updates”?

Mar said, “I don’t really want ‘tips to stay healthy’ (but another user might think differently). I want the latest coverage on the outbreak.”

Another pro tip is to ask an expert. I received feedback from some of my lifelong journalism mentors.

I then launched my write up on LinkedIn and Twitter. And the response was positive. Professionals celebrated. Like. Loved. And even shared.

Product Designer, Stephanie Louw, re-shared and said, “Yay, well done UX Writer!”.

And Service Designer, Phillip Frost, said, “One of the little gems you’ll get if you have a UX Writer on your team.”

Oh, and how can I forget Mar’s retweet:

With this simple, humble mockup, I hope teams see how they can implement conversation mining to empathize with the person in front of the screen and understand how they think.

And using artifacts like the microcopy canvas and Whimsical can help any writer explain their thought process behind each word and visually communicate with team members and stakeholders.

It’s our job as UX Writers to not only write contextual copy, but also think about the “what ifs?” By thinking ahead, writers can point out some idiosyncrasies that may arise and avoid them from happening.

--

--

Strategizing copy with a splash of personality. (portfolio in the works)