How CallMath normalizes AI receptionist prices

Step 1: turn your calls into each vendor's unit

• Answered calls = calls per month × the share the service answers. Every call: 100%. After hours only: your after-hours share. After hours plus missed daytime calls: after-hours share + (1 − after-hours share) × the share of daytime calls you miss.
• Minutes = answered calls × average call length, rounded up to a whole minute.
• Unique callers = answered calls × (1 − repeat-caller share), rounded up. Only Goodcall bills this way.
• Conversations = answered calls. Jobber also counts text threads; CallMath counts calls only, so add your texts if you use Jobber's text answering.

Counts derived from percentages are rounded up, so estimates never under-count usage.

Step 2: price one plan

• Inside the plan: you pay the plan price.
• Past the plan: plan price + (your usage − included usage) × the vendor's published overage rate.
• Required subscriptions are added. Quo's Sona only answers Quo phone numbers, so every Quo result includes Quo's cheapest plan (Starter, $19 per user per month billed monthly).
• Overage not published: we never guess. The result says "contact vendor" and shows the lowest possible cost (the plan price).
• Flat plans: one price at any volume. If the vendor states a fair-use limit (SkipCalls: 200 calls a month), volumes above it show "contact vendor".
• Starting prices such as "from $299/mo" are treated as a lower bound, never as a computed cost.
• Free plans cost $0 until the included usage runs out.
• Per-unit list prices: where a vendor lists a per-call rate for a volume option instead of a monthly total (two Smith.ai options), the monthly total is rate × calls, and the result says so.
• Month-to-month prices only. Annual prices are shown on vendor pages where published but are not used for ranking, because they lock you in for a year.

Worked examples

These lines come from the same code that runs the calculator.

Step 3: choose each vendor's plan, then rank

For every vendor we price all of its verified plans at your volume and keep one: the plan that best meets what you asked for; among those, one whose cost can be computed; among those, the cheapest. When a cheaper plan was passed over, the result says why (for example, it lacks live transfers, or its overage rate isn't published).

Vendors are then grouped by fit and ranked by that monthly cost:

1. Meets everything you asked for: the vendor's own page confirms every need you ticked.
2. Might fit: a need is not stated on the vendor's page, or is only partly met (for example, the AI texts a booking link instead of booking the appointment).
3. Missing something: the vendor's page shows the plan doesn't include it, or the product only works inside software you don't use (Jobber Receptionist needs Jobber).

Inside each group, computed costs come first from lowest to highest, then vendors whose cost can't be computed at your volume, ordered by their lowest possible cost. Commissions play no part: none exist today, and if affiliate links are added later they will not change the order (see the disclosure).

An integration counts only when the vendor's own page names it. Many products connect through Zapier; we don't assume that covers your software.

Starting values

The calculator opens with 100 calls a month at 2.5 minutes each, 30% after hours, 20% of daytime calls missed and 0% repeat callers. These are examples, not industry averages. The 2.5 minutes mirrors PATLive's pricing page, which equates 50 minutes to "About 20 calls". A 0% repeat share is the most expensive case for unique-caller billing.

Where the numbers come from

Every price was read on the vendor's own website on the date shown. A vendor is verified when its plan prices, included usage and overage rule (or the fact that no overage rate is published) were readable there. A vendor is not verified when the page was gated, ambiguous or did not publish a price; those vendors are listed with the reason and never ranked.

Vendor links pass through our click-counting redirect and are marked sponsored. The raw data is free to download as JSON or CSV.

How often we update

At least once a month a script re-reads every vendor pricing page, compares the dollar amounts and plan sizes with our data, and prints a change report. A person then confirms each flagged change on the vendor's page before the table is edited and the "checked" date moves. Vendors can also report changes through the contact form.

What we don't know

• Taxes, regulatory and telecom fees, and any cost to forward or port your phone number.
• Whether minute-based vendors round each call up to a whole minute. None of the pricing pages we read say, so real minute counts can run higher than calls × average length.
• How hang-ups, very short calls and robocalls count, beyond what each vendor states about spam.
• Overage rates that aren't published: Zoom Virtual Agent Receptionist, Rosie, Ruby.
• Whether Quo's 1,000 included automation credits stack with a paid credit pack. We assume they don't, which can overstate Quo's cost by up to 10 calls' worth.
• Whether the "1,000 monthly overage credits" listed on My AI Front Desk's plan are included free. We assume they aren't, which can overstate its cost by up to $10 a month.
• Promotions, negotiated discounts, custom plans and contract terms that aren't on the pricing pages.
• Quality. CallMath compares published prices and features only. It does not test how well any receptionist answers, books or transfers calls.
• Health-privacy compliance. Medical, dental and other health-care offices should ask each vendor directly about HIPAA; CallMath does not evaluate it, and nothing here is medical or legal advice.

Recording rules vary by state: Illinois and about a dozen other states generally require every party's consent to record a call (Justia's 50-state survey). This is general information, not legal advice.

Corrections

Found a wrong number? Tell us with a link to the vendor's page. We check it and update the table and the date.