I want to discuss a task or project
Done! We will get back to you within one business day
Something went wrong. Please try again
Alexander Start:Duck 🦢
@olga_startduck
Write to Telegram
Назад
1k+ участников
Вступить в Telegram
Case
17.08.2025

How the BigDuck.ai chatbot significantly increased Instagram conversions

Proven: On Instagram, a customer wants numbers, not courtesy

Here's what we've noticed:
The person who is polite doesn't win in Instagram sales — the one who wins hits the nerve customer right away. And he does it — without introductions, without “good afternoon” and without “my name is”.

Common opinion

Almost every second client who comes to us to implement an AI bot starts a conversation with the same phrase:

“We have already developed a script. He's polite, he works. Just automate it as it is.”

This was also the case with this client, a small business that sells electric air dispensers with pleasant smells for cars. Their product is non-standard. Not shampoo, not food delivery. The product is emotional and impulsive. Which means, logically, the dialogue must be accurate: soft, friendly, with presentation and a kind tone.

That's what the owner thought.

The bot's original script looked like this:

“Good afternoon, my name is Irina. You have submitted a request on our website. How can I help you?”

It seems perfect — culturally, without pressure. But Instagram is not a website. This isn't an email. And it's not even WhatsApp. There, a person reacts quickly and superficially. While he was leafing through the stories, something caught his eye, he clicked on the link — and... now he's in dialogue.

But is he ready to “meet” and “greet”?

An event that opened our eyes

A couple of weeks after launch, we connected the module analysis of logs and communication patterns within our platform BigDuck.ai.

The system has gone through 400+ dialogues, dividing them into categories:

  • Was there a response from a lead?
  • How many messages are in the dialog before proceeding to the order?
  • What language was used?
  • What dialogues did you get interested in?

And here's what came up:

More than half of the people did not answer, when the first message was long and “polite”.
But the same people started chatting, when the bot first reported logistics: price and delivery times.

This message turned out to be the most effective:

“Delivery within the city is 1 hour. Out of town — 1 day. All the smells are there”

The numbers spoke for themselves:

  • Old script: answer in 3 out of 10 cases.
  • New approach: answer in 6 out of 10.

Why this observation is important

The owner insisted on the classic “courteous consultant” model. And I was sure that this “should be the right thing”. But an idea of how people want communicating is not the same as them virtually react.

It is important to understand the context of the environment. Instagram isn't a place for ceremonies. This is a place for instant decisions. There is no “submit a request” button. There is only “opened a dialogue → received an answer → drew a conclusion”.

If a person doesn't get specifics in the first 3 seconds, they leave. And if he was immediately told the delivery dates, this is instead of dialogue gives him the information he needs. He can specify the part — or buy it right away.

This is our main difference from most chatbot integrators. They make a product — and they exit the process. We're staying in it: we are not just automating communication, but developing the client's product through analytics, testing and adaptation to the actual behavior of the audience.

For us, the bot is not the end point. This is a growth tool that we regularly improve together with our business.

What we've changed

  1. First, we added FAQ based on real questions from the correspondence archive.
    Most popular: “What is the delivery?” — implemented as the first replica of the bot.
  2. Removed “Thank you”, “Thank you”, “Have a great day!” in initial posts.
    Few people reacted to this politeness — it seemed as if they were being “sold” something right away. After removing these phrases, engagement increased by 8%.
  3. Neutral and harsh phrases were tested.
    The most effective tone is neutral and concise:
    “It is available. Shipment within a day. What kind of smell are you interested in?”
  4. We built an individual recommendation model within BigDuck.ai.
    The bot himself chose the wording depending on the incoming request: more polite if the user wrote politely, and drier if he just left a number.

Example: it was — it became

It was:

  • Script: “Hello, I'm Erina, you were interested in our product.”
  • The answer is 30% of the time.
  • The average response time is 8 minutes.

It became:

  • Script: “Delivery to the city is 1 hour. Out of town — 1 day. All the smells are there.”
  • The answer is 63% of the time.
  • The average time is 1 minute and 20 seconds.

An unexpected conclusion for business

In sales via messengers politeness is not a valueif it's worth the time. A person doesn't want to be a bot's friend. He wants what he needs.

The most polite line is a clear, helpful answer.

And if it weren't for AI analysis of correspondence, customers would continue to lose half of their leads, convincing themselves that they are “doing everything right”.

If you want such discoveries — subscribe! We don't just give you a bot — we we measure customer behavior in numbersso that you earn money instead of hoping.

Читайте также

No items found.

write to us and we will show you the way to efficiency