Most pages on your website get browsed. The pricing page gets studied. Someone who reaches it has usually read enough to be interested and now wants to know whether the numbers work for them. They are close to a decision, and they are close to leaving.
What stops them is rarely the price itself. It is the question sitting next to the price: does the middle option include the thing I need, what happens if I go over the limit, is there a setup fee nobody mentions. If the page does not answer it, they open another tab and ask your competitor instead.
That is why the pricing page is the best single place to put an AI sales assistant, and also the place where a badly set up one does the most damage.
What people actually ask on a pricing page
Read enough transcripts from pricing pages and the questions fall into a small number of groups. Very few of them are "can I have it cheaper".
| What they type | What they are really asking | A good answer does |
|---|---|---|
| "Does the Standard package include X?" | Is this the right tier for me? | States clearly whether X is included, and which tier includes it if not |
| "What happens if I need more?" | Will I get stung later? | Explains the limit and the upgrade path in plain terms |
| "Is there a contract?" | Can I get out if it goes wrong? | Gives the actual terms, notice period included |
| "Are there any other costs?" | Is this the real number? | Lists setup, travel, materials or add-ons honestly |
| "Which one would you recommend?" | Help me choose | Asks one or two questions about their situation, then suggests a tier |
| "Can you do a better price?" | Is there room to negotiate? | Does not negotiate; offers to pass the request to a person |
Notice how many of those are about exclusions and edge cases. That is the content most businesses leave off the page because it looks negative. The assistant needs it anyway.
The discount question, handled properly
Sooner or later someone asks for a deal. A poorly grounded assistant will sometimes try to be helpful and hint at flexibility that does not exist. That leaves you with a customer who expects a price you never offered.
The right behaviour is simple: no improvising on money. The assistant explains what the published options are, mentions anything that genuinely reduces cost (annual billing, a smaller package), and offers to pass the request on.
Visitor: This is a bit more than I was hoping to spend. Any chance of a discount if I sign up this week?
Assistant: I can't offer discounts myself, but I can tell you two things that might help. Paying annually works out cheaper per month than monthly billing, and the Essentials package covers most of what you described if you don't need the quarterly review. If you'd still like to talk about price, I can pass your details to the team and someone will reply within a working day. Would that be useful?
That answer does not lose the visitor, does not commit you to anything, and puts a named lead in front of a person who can judge whether a concession is worth making. A grounded assistant like SpideyChat answers only from the content you give it, so if your pricing content never mentions discounts, it has nothing to invent from. Still, say it explicitly in your instructions.
House rules for a pricing page assistant
Before you switch it on, agree a short set of rules. These are the ones that tend to matter most.
- Quote only published figures. If a number is not in your content, the assistant should say it does not have that figure and offer to get an answer.
- Name the exclusions. A visitor who discovers a missing feature after buying costs you more than one who learns it before.
- Recommend with a reason. "Pro suits you because you mentioned three locations" is useful. "Pro is our most popular plan" is noise.
- Custom work goes to a person. Anything that sounds like a quote rather than a package should end in captured details, not a guessed price.
- Do not chase. One offer to take contact details is enough. If they decline, keep answering questions.
The last one matters more than it looks. Pricing page visitors are wary. An assistant that keeps asking for an email address reads as a sales rep hovering by the till. Chatbot lead capture without killing the conversation goes into the timing.
Feeding it the right material
Your pricing page alone is not enough content. It was written to look clean, not to answer follow-up questions. Give the assistant the material a good salesperson would have in their head:
- The full breakdown of what each option includes, down to the small print.
- What each option excludes, written plainly.
- Billing details: when payment is taken, renewal, refunds, cancellation terms.
- The questions your team gets asked on sales calls, with the answers they give.
- The situations where you always quote individually, so the assistant knows when to stop and hand over.
Q&A pairs work well for items four and five, because they capture your exact wording. If you are starting from nothing, the AI FAQ generator will give you a first draft of likely questions to edit.
Greeting or no greeting
A proactive greeting that fires the second the page loads is the fastest way to make a pricing page feel pushy. Visitors want a moment to read the table first.
A better pattern is to wait. Let the widget sit quietly, or use a short greeting that appears once someone has been on the page long enough to have read the options. Exit-intent greetings can also work here, because someone moving to close the tab on your pricing page has usually hit a question they could not answer. Something like "Anything about the packages I can clear up before you go?" is enough. When to nudge with a proactive greeting covers the trade-offs in more detail.
What to do this week
Start by collecting the pricing questions you already get by email and on calls. Twenty real examples will teach you more than any template. Write the answers down, including the awkward ones about exclusions and extra costs.
Then add the assistant to the pricing page only, with no proactive greeting for the first week. Read every transcript. You are looking for two things: questions it could not answer, which tell you what content to add, and questions that keep coming up, which tell you what your pricing page itself should say more clearly.
After two weeks, compare enquiries from that page against the fortnight before. If you want a wider view of turning that traffic into conversations, pricing page visitors into enquiries picks up where this leaves off.