Skip to content
ChatLens360
Revenue & customer experience

When your chatbot quotes the wrong price

A wrong price from your chatbot either loses the sale or sets up a dispute. Learn why bots get pricing wrong and how to test it properly.

Illustration of a falling bar chart with coins slipping away for the article "When your chatbot quotes the wrong price", tagged Pricing, Revenue, Chatbot accuracy.

Pricing is one of the most common questions any chatbot gets, and one of the easiest to get wrong. A price that is too high sends the buyer away. A price that is too low creates a dispute at checkout. Either way, the business pays.

This post looks at why chatbots misquote prices, what each type of mistake costs, and how to test pricing answers properly.

Two directions, two different losses

A wrong price can go either way, and each direction hurts differently.

Quoted too high. The customer thinks you are more expensive than you are. They leave to compare, and many do not come back. You never see this loss, because the customer never reached checkout.

Quoted too low. The customer gets excited, adds the item to their cart and sees a different number. Now they feel misled. Some abandon the purchase. Some complain. Some ask you to honor the price the bot gave them, and you have to decide whether to absorb the difference or risk losing them.

Neither outcome is a small thing. Pricing is where trust is most fragile.

How it looks in a real conversation

Here is an example from a software company with monthly and annual plans:

Customer: How much is the Team plan? Chatbot: The Team plan is $29 per user per month.

The real answer is $29 per user per month billed annually, or $35 per user on monthly billing. The bot picked the cheaper figure and dropped the condition.

The customer signs up for monthly billing, sees $35 on the invoice and opens a ticket asking why they were charged more than they were told. Now support has to explain, apologize and maybe offer a credit. A clear answer would have cost nothing.

Why chatbots get prices wrong

Pricing errors usually come from a handful of causes:

  • Stale pages. Old pricing pages, blog posts announcing past prices and archived PDFs still sit on your site.
  • Missing conditions. Annual vs monthly, per user vs flat, introductory vs renewal, excluding tax or shipping. The bot keeps the number and drops the condition.
  • Promotions that ended. A sale banner from last season is still in the content the bot reads.
  • Currency and region. The bot gives a US dollar price to someone in Europe, or the other way round.
  • Mixed-up tiers. The bot confuses the features or prices of two similar plans.
  • Made-up numbers. When the bot cannot find a price, it sometimes invents a plausible one.

An illustrative cost example

These numbers are invented for illustration only.

Say 300 people a month ask your chatbot about pricing. If 4% get a price that is too high, that is 12 people who may walk away. If a third of them would have bought at $400 each, that is $1,600 a month in lost revenue.

Now say another 4%, also 12 people, get a price that is too low. If half of them complain and you give each a $50 credit to keep them, that is $300 a month, plus the support time and the frustration.

None of this would show up as a "chatbot error" in your reports.

A pricing test checklist

Use this list to test your chatbot's pricing answers today:

  • Ask for every plan or product price directly. "How much is X?" for each one.
  • Ask with conditions. "How much is X per month if I pay monthly?" "How much for 10 users?"
  • Ask about extras. Tax, shipping, setup fees, add-ons and overage charges.
  • Ask about discounts. "Do you have a student discount?" "Is there a sale on?" Make sure the bot does not invent offers.
  • Ask from another region. "How much is it in euros?" "Do you charge VAT?"
  • Ask leading questions. "So it's $10 a month, right?" A weak bot just agrees.
  • Ask a comparison. "What's the difference in price between plan A and plan B?"
  • Check every number against your live pricing page, not against memory.

Fixing pricing answers for good

When you find pricing mistakes, these fixes make the biggest difference:

  1. Have one source of truth. One current pricing page or document, clearly dated. Remove or update old copies.
  2. Write conditions next to numbers. "$29 per user per month, billed annually" is much harder to misquote than a table with a footnote.
  3. Tell the bot how to handle prices. Instruct it to always include the billing period and conditions, and to link to the pricing page.
  4. Tell the bot not to invent offers. If a discount is not in the source, the bot should say it is not aware of one.
  5. Re-test whenever prices change. A price update is exactly when stale content causes the most damage.

Pricing mistakes are also a big driver of abandoned carts. See chatbots and abandoned carts: where buyers drop off for more on that.

Check your chatbot's pricing answers

ChatLens360 reads your pricing pages and documents, asks your chatbot price questions in many forms, including leading and tricky ones, and flags any answer that does not match your real prices. Each issue shows the conversation, the expected answer, the source page and a fix prompt. Run a free chatbot audit or see pricing.

Put these ideas into practice

Run an automated audit of your own chatbot for free.