Ask three providers what an AI receptionist costs and you will get three numbers that cannot be compared. One quotes per minute. Another quotes per resolved conversation. A third quotes a monthly fee with an allowance buried in the plan table. All three sound reasonable on the sales call, and all three produce very different bills once real enquiries start arriving.
The headline price is the least useful number. What decides your AI receptionist cost over a year is the unit you are billed for and how that unit behaves when you get busy.
This guide sets out the pricing models you will meet, shows how each one moves with volume using an example you can rerun with your own numbers, and states SpideyChat's pricing exactly so there is nothing to decode.
First, be clear which product you are pricing
"AI receptionist" covers two quite different categories. Many products in the search results answer phone calls with a synthetic voice. Others, including SpideyChat, work on your website: they answer visitors in a chat window, capture leads, open support tickets and hand over to a person. SpideyChat does not answer phones.
They are priced differently because the costs behind them differ. Voice products tend to bill by the minute or call. Website assistants tend to bill by conversation, resolution or a flat subscription. If your problem is missed phone calls, compare phone products and answering services with each other; our guide to phone answering service costs covers that. If your problem is website visitors who leave without getting an answer, compare website products.
The five pricing models you will meet
| Model | You pay for | When enquiries double | Watch for |
|---|---|---|---|
| Per minute | Talk time, usually on phone products | Bill roughly doubles, and long calls cost more | Minimum charges, rounding up to the next minute |
| Per call or per conversation | Each interaction, however short | Bill roughly doubles | What counts as a conversation, and whether spam or one-word chats count |
| Per resolution | Conversations the system marks as resolved | Bill rises with success | Who decides what resolved means |
| Per seat | Each team member with a login | Bill does not change unless you add staff | Automation features locked to higher tiers |
| Flat monthly | A plan with an allowance | Bill unchanged until you exceed the allowance | The allowance itself, and the price of the next plan up |
None of these is dishonest. They simply shift risk. Usage-based models are cheap when you are quiet and expensive when you are busy. Flat models cost the same in a quiet month, which can feel wasteful, and protect you in a busy one. The right one depends on how steady your volume is, not on which sounds cheapest.
A worked example across three months
Rates here are invented for illustration, not taken from any provider. Replace them with the quotes in front of you.
Say you compare a usage-based website assistant at $0.80 per conversation with a flat plan at $59 a month. Your business has a quiet month with 40 website conversations, a typical month with 90, and a busy month with 250.
| Month | Conversations | Usage-based at $0.80 | Flat at $59 |
|---|---|---|---|
| Quiet | 40 | $32 | $59 |
| Typical | 90 | $72 | $59 |
| Busy | 250 | $200 | $59 |
| Year, if 4 quiet, 6 typical, 2 busy | 1,200 | $960 | $708 |
On this example the flat plan is cheaper over the year, but only by a few hundred dollars, and a business with more quiet months could easily see the reverse. The bigger difference is in the shape: the usage-based bill swings from $32 to $200, and it swings upwards precisely when you are busiest and least likely to be watching it. If your busy months are busier than this, the gap widens quickly. Why per-conversation pricing punishes your busiest months follows that thread further.
One more check for flat plans: confirm the allowance covers your busy month. Allowances are often counted in messages, and a single conversation contains several messages, so look at a few real transcripts to see how many yours typically use.
SpideyChat's pricing, stated exactly
SpideyChat uses flat monthly pricing in USD. It is not charged per conversation or per resolution.
- Free, $0: 1 assistant, 100 messages a month, no credit card.
- Starter, $20 a month ($17 billed annually): 3 assistants, 2,000 messages a month, website crawler and lead capture.
- Pro, $59 a month ($49 billed annually): 10 assistants, 9,000 messages a month, support ticket inbox, removal of SpideyChat branding and up to 3 companies.
- Enterprise: custom pricing.
The full comparison is on the pricing page.
Price is only half the sum
A cheaper tool that misses enquiries is not cheaper. Before settling on a price comparison, estimate what one captured enquiry is worth to your business, using your conversion rate and average job value. If a single extra enquiry a month covers the subscription several times over, the gap between two plans matters far less than whether the assistant actually answers well and hands over cleanly.
That is also why a free plan is useful: it lets you check answer quality on your own content before any pricing model comes into it. Working out what one captured enquiry is worth takes you through the arithmetic.
Costs that do not appear on the plan table
The subscription is rarely the whole cost. Budget for these too:
- Your time teaching the assistant: gathering pages, documents and Q&A pairs, typically a few hours to start.
- Ongoing upkeep: updating content when prices, services or hours change.
- Someone reading captured leads and replying quickly, which is where most of the value is actually realised.
- Any connected tools, such as an automation platform the webhooks feed into, which have their own pricing.
- Overage charges or forced upgrades on usage-based and allowance-based plans.
Ask every provider about the last one directly, and get the answer in writing before you sign.
Working out your own number
Before you request another quote, pull three figures from your analytics or inbox: website enquiries in your quietest month, a typical month and your busiest month. Then ask every provider the same question, in writing: what would my bill be in each of those three months, including any overage?
Put the answers in a table like the one above. Then run the value side through the chatbot ROI calculator, so you are comparing cost against what the enquiries are worth rather than against each other.