RankDude

How I test white label AI receptionist platforms

Nine tests, the same for every platform, each graded from 1 to 5 on a fixed scale. The overall grade is their average plus my verdict, which can move it by up to one point. Every test is run by hand on an agency account I use, building the same receptionist for the same test client, and the recordings, screenshots, invoices, timings and exports behind each grade go in the full test results published with the review.

Written by Miles Foster, Owner and tester, RankDude.

I play a small agency with three clients. On each platform I build the same receptionist for the same test client, a two-chair dental practice with 20 facts, a calendar with gaps and clashes in it and two people to transfer to, then set up two more clients from it. I put the client dashboard on my own domain, send the client a login, phone the receptionist 40 times from different handsets, run a month of billing through the platform's own payment tools, and at the end try to take everything with me. Every call is recorded with the caller's knowledge, and nothing is graded from a feature list, a partner deck or a demo.

Each review states how the account was obtained: bought at the public price, used on a free tier, or provided by the vendor. A vendor-provided account never changes a grade, and a vendor never sees a review before it is published.

How I rate

Every tool gets one score out of 5.

First, I run the same 9 tests on every tool and give each one a grade from 1 to 5. The average of the graded tests is the starting point; a test that cannot be graded is left out of it, and the review says how many tests the score is based on.

Then I add my verdict: what the tests cannot measure, such as how the tool holds up after the test weeks, how billing and cancelling work in practice, and how it compares with the others in the category. My verdict can move the score up or down by one point at most.

Every review shows how much my verdict changed the score, and why, in its scores table.

What a grade means

5Excellent
4Good
3Average
2Below average
1Poor

The final score is the average of the graded tests (nine when every test can be graded), to one decimal place, plus my verdict, which can move it by up to one point. If a test couldn't be finished properly it gets no grade, and the review says how many tests the score is based on. Platforms that wrap another company's voice engine are graded on what the agency and its client actually get, and the review says which engine is underneath and who holds that account.

The 9 tests

1.Builder & Time to Live

The clock starts at sign-up and stops when a forwarded call to the test client's number is answered correctly with the client's name, my own fixes included. I note whether the build is a form, a flow chart, a prompt box or code, and how much I had to read. Then each point is tried, not read from a feature list: a new number, forwarding from an existing number, Google Calendar, Outlook calendar, a CRM, a webhook or Zapier, a text to the caller, call recordings, transcripts, and after-hours rules.

  • 5: live within 2 hours, 9 or 10 points working
  • 4: live within a day, 7 or 8 points
  • 3: live within three days, 5 or 6 points
  • 2: live within two weeks, 3 or 4 points, or setup needed a sales call
  • 1: not live within two weeks, or 2 or fewer points

2.Call Quality

All 40 calls are timed from the first ring to the greeting and checked for missed or dropped pick-ups. On 20 of them I measure the gap between the end of my sentence and the start of the reply at five points, from the recording. Three fixed 30-second samples are rated on naturalness, clarity, pacing and intonation by me and two listeners who do not know which platform they are hearing. On ten calls I interrupt mid-sentence and count whether the receptionist stops and answers the new question.

  • 5: every call answered within two rings, median gap under 1 second, listener rating 4.5 or more, all ten interruptions handled
  • 4: every call answered, median gap 1 to 1.5 seconds, rating 4 to 4.4
  • 3: one or two calls missed, or median gap 1.5 to 2.5 seconds, or rating 3 to 3.9
  • 2: three to five calls missed, or median gap 2.5 to 4 seconds, or rating 2 to 2.9
  • 1: six or more calls missed, or median gap over 4 seconds, or rating under 2

3.Task Accuracy

Ten calls ask for facts from the client profile and are marked right, wrong or invented. Eight book, move or cancel an appointment and are checked on the connected calendar for date, time, name, number and reason, with one planted clash. Six leave a message with a name and number, two of them mumbled, and are checked in what reaches the client. Six ask for a person and are checked for whether the right phone rings and what it is told first. An invented fact or a wrong phone number counts double.

  • 5: 28 to 30 correct, nothing invented
  • 4: 25 to 27 correct, at most one invented answer
  • 3: 20 to 24 correct
  • 2: 14 to 19 correct, or a double booking, or a wrong phone number in a message
  • 1: 13 or fewer correct

4.White-label Depth

After branding is set up, each point is checked as the client would meet it: the dashboard on the agency's own domain with HTTPS, the login page, the logo and colours, the password-reset and notification emails and their sending address, the invoice, the call summary email or text, the recording and transcript pages, the mobile app or its absence, the help documentation, the phone number's caller ID name, what the receptionist says if asked what it is, and the page source and cookies. Any point where the platform's name, domain or logo appears counts as a leak.

  • 5: all twelve clean
  • 4: 10 or 11 clean
  • 3: 8 or 9 clean
  • 2: 5 to 7 clean, or the platform's name appears in client emails
  • 1: 4 or fewer clean, or the login page carries the platform's brand

5.Client Management

The second and third clients are created by copying the first; I time each. Then I check: a separate phone number per client, a separate calendar, per-client usage and call logs visible to the agency, a client login that shows that client only, a permission that lets the client edit opening hours but not the prompt, bulk changes across all three, and what the client login reveals about the agency's costs or other clients. Any client login that exposes another client, the agency's margin or the platform's credentials fails the test.

  • 5: all set up in under 15 minutes each, every check passed
  • 4: under 30 minutes each, one check failed
  • 3: under an hour each, or two checks failed
  • 2: over an hour each, or three checks failed
  • 1: no client accounts, or a client can see another client or the agency's costs

6.Billing & Margin

I set a retail price and a minute allowance for the test client, connect the platform's payment tool (Stripe or its own), raise the first invoice and pay it with a test card. After the 40 calls I compare the minutes the platform billed the client with the minutes on my recordings, and check what happens when the allowance is exceeded. I record whether the agency can see its own cost beside the client's price, and whether the client's invoice shows the agency's name only.

  • 5: price and markup set by the agency, automatic invoice and payment, usage correct to the minute, overage handled as configured
  • 4: automatic billing, usage out by under 5%
  • 3: invoice raised but paid outside the platform, or usage out by 5 to 15%
  • 2: billing by spreadsheet, or the client's invoice shows the platform's name
  • 1: no client billing, or usage out by more than 15%

7.Agency Support

The same three questions go to every platform: why a transfer failed on a named call, how to move a client's number to a different account, and how usage is counted when a call goes to voicemail. Each is sent through the AI assistant and every human channel offered. I record the time to the first useful answer and whether the answer was correct when followed.

  • 5: all three correct within 1 hour, with a human channel
  • 4: all three correct within 4 hours
  • 3: within 24 hours, or one answer wrong
  • 2: over 24 hours, or two answers wrong
  • 1: no reply or no channel

8.Portability & Terms

I export the receptionist's prompt and knowledge, the call logs, the recordings and the transcripts, and check whether the files are usable without the platform. I ask support to port the test client's number out and record the answer and the time. I read the terms on who owns the client relationship, what happens to clients and numbers when the agency stops paying, whether the platform may contact the agency's clients, and which company actually runs the voice engine and holds the account.

  • 5: full export, number portable, terms give the agency its data and its clients
  • 4: full export, number portable, terms silent on the client relationship
  • 3: partial export, or porting needs a sales conversation
  • 2: logs only, or the terms let the platform contact the agency's clients
  • 1: no export, or the number cannot leave, or the terms claim the clients

9.Pricing & Value

Every plan is read from the order page, not the partner deck. The figure graded is the monthly cost of ten client accounts each using 300 minutes of inbound calls, including the platform fee, per-minute engine charges, telephony, phone numbers, any white-label add-on and any per-client fee. Where the platform wraps another engine, that engine's published rates are added. I record the contract length, the trial, whether a card is needed, and the renewal price.

  • 5: under $400 a month all-in, monthly contract
  • 4: $400 to $699
  • 3: $700 to $1,199, or a yearly contract
  • 2: $1,200 or more, or the price is only available from a sales call
  • 1: poor value or hidden costs

Documented but not graded

Outbound calling, chat and WhatsApp agents, languages beyond English, agency training, marketplace listings, affiliate schemes and compliance certificates are checked and described in every review. When they work well they count in the written assessment, but they never add points to the overall grade.

Rules that apply to every test

  • Real builds on a real agency account with real calls to a real number, never promotional descriptions, partner decks or demo recordings.
  • Identical test client, scripts, thresholds and sample sizes for every platform in the category.
  • Authorised accounts only; no password sharing, no bypassing of plan limits, test payments only in test mode or refunded, and every test call made by me or by someone who knows it is recorded.
  • Every grade has evidence: a recording, a screenshot, an invoice, an export or a table, in the full test results.
  • A grade changes only when a retest produces new evidence; the review records the date and the reason.

Questions about a specific grade go to the contact page. Makers can submit a white label AI receptionist platform for testing. No white label AI receptionist platform has been through these tests yet; every one I follow is listed on the White label AI receptionists page, and reviews appear there as tests finish.