💬 Smarter Chat, Better Data

How I Designed a Messenger Flow That Boosted Engagement and Surfaced User Preferences

My Role

Lead UX Designer. I was responsible for strategy, user research, conversational flow design, content tone, logic mapping, segmentation and analytics.

Problem To Solve

The team needed a way to capture user preferences and segment audiences without relying on clunky email forms. We wanted a playful, seamless, and measurable flow to drive engagement inside Facebook Messenger, a place users already interacted with us.

Project Constraints & Context

  • Brands: VersaChalk, Silver Phantom Jewelry, Ballet Bracelet

  • Platform: ManyChat on Facebook Messenger

  • Goal: Collect usable segmentation data while increasing open and click-through rates

  • Audience: 7,000+ engaged users


My Process

1. Started with Behavior, Not Buttons

Before building anything, I mapped out how the conversation should feel, not just what it should do. The goal was to keep things fast, friendly, and frictionless. I quickly realized that most people don't want to type in Messenger, they want to tap. So I leaned hard into pre-set buttons and minimized free text wherever possible.

 
 

2. Made It Sound Like... a Person (But a Polite One)

I experimented with tone and quickly saw a pattern: users were blunt but casual, think more “help” than “hello.” So I kept the copy short, warm, and direct.

To make sure no one got stuck, I added “listening” triggers for common phrases like “unsubscribe,” “support,” and “talk to a human.” I also built in logic to catch frequent typos so if someone typed “custmr suport,” the bot would politely clarify: “Did you mean customer support?”

It wasn’t about pretending to be human, it was about keeping things simple, responsive, and a little forgiving.

 
 

3. Tested with Real People, Not Just Wireframes

To validate the flow, I launched it to two small user cohorts (about 250 people each). This helped me catch issues early, like overly long questions or missing menu options. I added a persistent menu to make it easier to jump to shop, support, or browse at any time.

 
 

4. Scaled It Without Losing the Human Feel

Once the basics worked, I expanded the flow to over 7,000 users. I added smart questions that captured users' preferences, like industry, shopping habits, or content interests, and used tagging to group them automatically. Behind the scenes, everything was built in ManyChat with logic jumps, tags, and auto-export to spreadsheets for marketing to use later.

 
 
 
“
“When users realized they were talking to a bot, they tended to be more direct, use keyword-based language, and avoid politeness markers. This type of language is generally more successful than the convoluted, indirect language often used in normal conversation.””
— Nielsen Norman Group Study
 

Outcome

The chatbot flow significantly outperformed our typical email campaigns:

  • 83% open rate

  • 32% click-through rate

But metrics aside, it also made things easier for users. Some were genuinely surprised by how seamless the experience felt, being able to buy art supplies, get help from the support team, or just browse the website, all within a simple Messenger interface.

With fewer distractions and a more focused flow, users didn’t have to navigate tabs or dig through emails. The experience was direct, intuitive, and required very little effort.

Overall sentiment ranged from neutral to slightly positive, no hype, just a clean, low-friction way to engage. And on the backend, I was able to capture valuable user data and tailor future content and offers to match individual preferences.

Reflection

This project reminded me how important it is to meet users where they are, both literally (Messenger) and emotionally (direct, no-frills communication). The bot didn’t need to be clever. It just needed to be clear, fast, and helpful.

If I were to do it again, I’d build in a few more safeguards:

  • A simple way to reset the conversation if users got stuck or clicked the wrong thing.

  • Gentle prompts when someone paused too long, to guide them forward without being intrusive.

  • A clearer intro that set expectations, letting users know right away they were chatting with a bot.

I’d also explore more personalized follow-ups based on user preferences. If someone identified as an artist, for example, the bot could follow up with early access to new products or invite them to review the latest chalk markers.

Small details, but they add up.

Why This Project Matters

This wasn’t just a chatbot, it was a way to create value on both sides. Users got a faster, simpler way to find what they needed, and the business got clean, actionable data to personalize outreach without relying on guesswork.

It showed how much can be accomplished with a lightweight, well-designed interaction. No complex UI. No heavy dev work. Just a clear, focused flow that respected the user’s time and delivered results.

It’s a reminder that small, well-executed tools can have a big impact when they’re built with intention.