The First Five Seconds

I built a bot last year for a real estate client. Spent two weeks on the flow logic, the property database connection, the follow-up sequences. Launched it on a Tuesday. By Friday we had 143 conversations started and exactly 2 qualified leads.

I was furious. Then I looked at the data.

You have about five seconds of goodwill when someone opens your bot. My bot spent those five seconds asking for a budget.

Users were landing in the bot, getting asked for their budget, their preferred location, their timeline, how many bedrooms, whether they had a pre-approval letter. Question after question. And they were leaving. Not abandoning at question four or five. Abandoning at question one.

The bot was doing exactly what I had built it to do. It was qualifying. And the users were rejecting the entire process before it even began.

This is the tension every chatbot builder has to face. When a brand new user lands in your bot for the first time, they do not know you. They do not trust you. They clicked a button or scanned a QR code or followed a link, and they are giving you exactly one chance to prove this interaction is worth their time.

If the first thing you do is ask them a question, you have already lost a percentage of them. The question is how large a percentage, and whether you are okay with it.

I spent the next month running tests on that same bot. Same audience source, same landing point, but different onboarding approaches. Here is what I learned.

The default approach: qualify first

The direct qualification approach is the default for most builders because it feels efficient. Ask the hard questions first. If they are not qualified, you have not wasted their time or yours. The problem is that qualification questions demand something from the user before you have given them anything. You are asking for their budget, their phone number, their availability. You are asking them to invest in a conversation that has not yet proven its value to them. From the user's perspective, this feels like being interviewed by someone who has not even introduced themselves.

I tested a bot that led with three qualification questions before showing any value. The results were sobering. Of users who started the conversation, 31 percent completed the full qualification process in the question-first flow. Nearly seven out of ten people who entered the bot never saw what it could actually do for them.

Flipping it: value first

Then I flipped it. I rebuilt the same flow to start with value first. The bot greeted the user, immediately showed them three recent properties matching very broad criteria, gave them a quick tip about the local market, and offered to help them narrow their search. The first interaction was not a question. It was a gift of useful information.

The qualification questions came after. And here is the surprising part. They did not harm conversion. Users who had already seen value were far more willing to provide their budget and contact details. In the value-first flow, 78 percent of users who started the conversation completed the same qualification questions. The same questions, just asked in a different order.

But this raises a problem of its own. Giving value before qualifying means you will invest time and bot resources in people who will never convert. That real estate bot spent message credits and API calls showing properties to users who had no intention of buying. That stings if you are paying per conversation. This is the problem the hybrid approach solves. By making the free taste nearly cost-free to serve, you cap your exposure while still giving value first.

The hybrid: a taste, then a choice

I tested a hybrid approach next. The bot would give a small taste of value that cost almost nothing. A two-line market insight. A single example listing. Just enough for the user to feel like they had gained something. Then the first question appeared. It was not a qualification question. It was a choice. "Would you like me to help you find properties that match your specific needs?" The user had to opt in before the qualification began.

This changed the dynamic completely. Users who said yes were self-selected as interested. After opting in, they answered the qualification questions at a 91 percent completion rate. Users who said no could still browse a few more general listings without ever being pushed. They were not lost. They were just not ready yet. And some of them came back the next day.

The hybrid approach won in my tests. Not because it is the most efficient on paper. Because it respects the psychology of the first-time visitor.

Why this works

A brand new user in your bot is not a lead yet. They are a curious person who has taken a tiny step toward you. Your onboarding flow is not a qualification funnel. It is a hospitality experience. The best restaurants do not hand you a bill when you walk through the door. They seat you, pour you water, tell you about the specials. Then when you have decided you want to be there, they ask what you would like to order.

Your bot onboarding should work the same way. Give something small and valuable first. Let the user opt into the deeper conversation. Then qualify.

Test it in your own bot

Try this split in UChat. Create two onboarding paths using the Split step with a 50/50 split. In path A, use a Question step with a button set that asks for budget or timeline as the very first interaction after the greeting. Tag everyone who goes through this branch as onboarding_a.

In path B, start with a Send Message step that delivers one genuinely useful piece of information first. Then add a Question step with a single button that reads "Yes, help me find the right option." If they click it, route them into the same qualification flow you are using in path A and tag them as onboarding_b. If they skip it, let them browse freely without pushing.

Run both paths on equal traffic for two weeks. Then compare completion rates between the onboarding_a and onboarding_b tags. Look at how many users from each group finished the qualification process. Not just lead volume. Completion rate.

I have run this test on half a dozen bots now and the pattern holds every time. But your audience is different from mine. Your bot serves a different purpose. Your users have different expectations.

What does your first-time onboarding look like? Do you lead with value or qualification? And have you actually tested the other approach to know which one works better?

Try it together with the community. Run the test and share your results in the UChat community. Post which approach won in your bot and what your completion rates looked like. Real data from real bots is how we all get better at this.