The whole platformTwo recorders · One account · Three jobs
Two recorders, one account, and one record of the work.
Qualitify is not a recorder with a dashboard bolted on. It is one account per agent, collecting from the computer and the phone they actually work on, feeding one queue that a reviewer scores and one set of figures an owner can act on.
One account in the middle, and everything hanging off it.
Most teams end up with the opposite: a recorder on some machines, a spreadsheet of scores, a folder of audio files somebody renames, and a monthly meeting where the numbers are argued about rather than read. The reason is that each piece was bought separately and none of them knows who an agent is. Here, the account is the thing everything else is attached to.
Desktop app
Windows. Records the WhatsApp calls an agent makes at their computer.
Android app
Every call the handset carries: the SIM line, WhatsApp, Telegram, Viber.
One agent account
One computer and one phone at a time, both signed in as the same person. Every call is attributed to that account, labelled with the app that carried it and the recorder that captured it — so a day of work is one list, whichever device it came from.
The agent
Their own calls, their scores, the feedback written for them.
The QA reviewer
A queue of calls to score, a scorecard to score them against, notes pinned to the second.
The owner
Coverage, trends, targets and the bill. Nothing installed.
An account, not a licence on a machine.
Recording software is usually sold per seat and installed per machine, which means the machine is what gets measured. Swap an agent onto a different computer and their history splits in two; put two people on one machine and it merges. Attaching everything to the person instead is a small decision with a long tail.
One computer, one phone
Two device slots per account, and the second attempt at either is refused with a message naming the machine already holding it. That is what keeps "whose call is this" from ever being a question.
Unlink from the dashboard
A lost handset or a departing agent is a session an owner ends in the browser. The device stops collecting at its next request, and anything already recorded and not yet uploaded still arrives.
The recorders are for agents
An owner or reviewer signing into a recorder is refused rather than quietly given one: a recorder collects the calls of whoever is signed in, and an owner’s machine is not where the work happens.
Labelled at the source
Every call arrives knowing which recorder sent it and which app carried it. A reviewer about to spend four minutes listening knows what they are listening to before they press play.
Three jobs, and only one of them installs anything.
The roles describe responsibility rather than rank. A small company often runs with the owner doing the reviewing until there is somebody to hand it to, and nothing in the product insists otherwise.
The agent
- Installs
- The app on their computer or phone
- Opens
- Their own page
They do one thing differently: nothing. They make and take calls as before. What they gain is a page of their own — today’s calls, their scores over time, the feedback a reviewer wrote for them, and the coaching their supervisor asked for. Framed as their work rather than as a file about them, because an agent who reads it as surveillance stops reading it.
The QA reviewer
- Installs
- Nothing
- Opens
- A queue
They open a queue of calls chosen for them, listen, and score against the scorecard the company wrote. They pin notes to the second a thing happened, so feedback is about a moment rather than an impression, and they keep private notes a supervisor can read and the agent cannot. Their coverage is reported, so a reviewer who only ever scores short calls is visible.
The company owner
- Installs
- Nothing
- Opens
- The whole company
They set the criteria, the sampling rules and the targets, then read what came back: how much of each person’s work was actually reviewed, whether scores are moving, where coaching was given and whether it changed anything. They manage accounts, devices, retention and the subscription from the same place.
Who can see what.
Worth being specific about, because a quality programme that nobody trusts produces careful calls rather than better ones. Two rows are a company setting and the rest are not; the last row is the one that is not negotiable.
| Can they | Agent | Reviewer | Owner |
|---|---|---|---|
| Their own calls and scores | |||
| Another agent’s callsA company setting decides whether reviewers see the whole list or only their queue. | setting | ||
| Play a recordingAgents can replay their own calls when the company switches it on. Every playback is logged. | setting | ||
| Feedback written for the agent | |||
| A reviewer’s private notes | |||
| Score a call | |||
| Set the scorecard and sampling | |||
| Accounts, devices and retention | |||
| Another company’s anythingNot a setting. No role in any company can reach another company’s data. |
The life of one call.
Eight stations. What matters at each is not the activity but what exists afterwards, because that is where a quality programme usually breaks: the listening happens, and then nothing durable comes out of it.
Agent
The call happens
A conversation, and nothing else yet
Recorder
It is captured
Audio on the device, visibly
Recorder
It is uploaded
A call on the server; the local copy is deleted
Platform
It joins the sample
A queue entry, or a counted call nobody will review
Reviewer
It is scored
A score, criteria, notes pinned to seconds
Agent
The feedback arrives
Something specific to read, on their own page
Supervisor
Coaching is agreed
An improvement area with a follow-up date
Owner
The figures move, or do not
Coverage and trend, with sample sizes
What a QA manager actually runs.
The job is not listening to calls. It is deciding what a good call is, making sure a fair share of each person’s work gets heard, and keeping the scores comparable between the people giving them. Listening is what the reviewers do once those three things are true.
Write the scorecard
Criteria in your own words, weighted, with the mistakes that fail a call outright marked as such. A starting set arrives with the company; the first month of reviewing is what tells you which half of it was wrong.
Set the sample
How many calls per agent per period, spread across the team rather than taken from whoever is convenient. Unanswered attempts are counted in the figures and kept out of the queue, because there is nothing to listen to.
Keep reviewers honest
Calibration: several reviewers score the same call, and the spread between them is the thing being measured. A scorecard that two reasonable people read differently is a scorecard that needs rewriting, not an agent who needs coaching.
Work the queue
Assign, reassign, skip with a reason, and label calls for grouping. Everything is attributable later: who scored it, when, and what they changed if they changed it.
A weighted criterion counts double; a critical one fails the call on its own. Both are decisions a manager makes, not settings the product has opinions about.
What an owner gets out of it.
Not a dashboard with more charts on it. The useful output of a quality programme is narrow: you learn which agents are losing calls at a particular moment, whether the coaching you paid for changed anything, and what was actually said when somebody complains that it was not.
Coverage you can defend
How much of each person’s work was actually reviewed, per platform and per period. A quality score from two calls and a quality score from forty are different claims, and every average here carries the sample size behind it.
Targets against reality
Daily call and review targets per person, measured against what the recorders collected rather than what anybody reported. The gap between the two is usually the most interesting number on the page.
An answer when it is asked for
A customer disputes what they were told. The call is on the record, the score is on the record, and so is who reviewed it — which turns an argument about memory into a recording somebody can listen to.
Retention that happens by itself
Audio is deleted at the end of the period you set; the scores and written feedback are kept. Storage is reported per company so the bill does not surprise anybody, and every playback of a recording is logged.
What you will not find here is a claim that this raises conversion by some percentage. There is no study behind a number like that, and a figure invented for a marketing page is the fastest way to lose the reader who would have checked it. Record a fortnight, review it, and the first coverage report will tell you more than any promise could.
Common questions.
Does an agent need both the desktop app and the phone app?
Only what matches how they work. An agent who sells from a desk installs the desktop app; one who works from a handset installs the Android app; somebody who does both can be signed in on one computer and one phone at the same time. Both feed the same account, so their day is one list rather than two.
What stops two people sharing one account?
Each account holds one computer session and one phone session. Signing in on a second computer is refused and the message says which machine is holding the slot, so the usual outcome is somebody realising they are still signed in at home. An owner can end either session from the dashboard when a laptop is lost or a person leaves.
Can a reviewer see calls that are not assigned to them?
That is a company setting. By default a reviewer works from their queue; an owner can open the whole call list to reviewers when the team is small enough that queues get in the way. Neither setting lets anyone see another company’s calls — that boundary is not configurable.
Do agents see the private notes a reviewer writes?
No. A reviewer has two places to write: feedback addressed to the agent, which the agent reads, and private notes for the supervisor, which they never do. Mixing those would make the second one unusable, and a reviewer with nowhere candid to write stops writing candidly.
Who decides which calls get reviewed?
The sampling rules, which a QA manager sets and the platform applies: how many calls per agent per period, spread across the team rather than taken from whoever is easiest to review. Anybody with access can still open a specific call and score it — the sample is a floor, not a cage.
What happens to the recordings over time?
Audio is deleted at the end of your retention period, automatically. Everything written about the call — the score, the criteria, the feedback, the coaching that followed — is kept, so a review from a year ago is still readable long after the recording is gone.
Can an owner review calls themselves?
Yes, and most do at the start. The roles describe responsibility rather than permission tiers: an owner can score a call, a reviewer can be given the whole list, and a small company often runs with the owner doing both jobs until there is a second person to hand one to.
What does a new company have to configure before it is useful?
A scorecard and a sampling rule, both of which arrive with sensible defaults you edit. Everything else — accounts, devices, retention, the dashboards — works out of the box. The first useful day is the one where somebody reviews a real call and tells an agent what they heard.
There is more detail on the two recorders themselves: Android call recording covers every call a phone carries, and WhatsApp call quality assurance covers the desktop side.