How ScanMyJob reads your exchanges
No diagnosis, no verdict: a method that counts, dates and cites — and says plainly what it cannot conclude. Here is what sits underneath the report.
Before the models: one principle governs all of them, and everything else on this page follows from it.
Established facts, never a characterisation
ScanMyJob records what your conversations contain: dates, counts, sequences. It does not say whether this is harassment, puts no clinical label on anyone, and does not settle who is right.
What it does
- Count, date, place things in a sequence.
- Show what contradicts your reading as much as what confirms it.
- Say when the conversations provided do not document the question.
What it does not do
- Say who is right and who is wrong — that is not its role.
- Diagnose a person who has not been heard.
- Prove you right.
This is not editorial caution, it is in your interest: an automatic verdict teaches you nothing you did not already think. A dated chronology does.
How the models are chosen
No conversation gets all 49 models: applying every grid to the same exchange would produce that many partial readings and drown out what is actually useful. A router keeps a handful, based on three things you provide.
Who is exchanging with whom
The report distinguishes an exchange with your manager, with a colleague, or within a group: each relationship calls for different frameworks.
How many people are involved
- 2With two people, the interaction and its sequence are observable — nothing collective is.
- 3–5From 3 to 5, a difference in how people are treated becomes noticeable.
- 6–15At 6 or more, it is the group's structure — who answers whom, who is left out of a thread — that becomes observable. But the more participants there are, the less cleanly a behaviour attaches to one person, and the report is deliberately more cautious on that point.
What you are trying to find out
Before the analysis, you pick up to 3 objectives out of 8 — understanding the situation, documenting dated facts, preparing a meeting or a negotiation… The first one drives how the report opens and closes. This choice changes neither the facts established in your exchanges nor the level of evidence — only what gets emphasised.
When your exchanges do not document the question
An objective never manufactures a section: if the submitted exchanges do not contain enough to answer it, the report says so — the question is marked as not documented rather than answered without material.
A ceiling for readability
Even when several frameworks apply, the report keeps a capped number per situation — 4 on the Essential tier, 7 on the Premium tier. A report built on five well-supported models can be read; one that name-drops twenty cannot.
The evidence scale
Every statement in the report carries one of these three levels, and no other word: you will not find "presumably", "likely" or "proven" in it.
Observed
Directly verifiable in the exchanges you submitted.
Indicated
Several elements point the same way, without being conclusively established.
Hypothesis
A lead worth checking yourself — nothing here is taken as settled.
This level can be lowered depending on how many people are involved — never raised: a ceiling bounds, it does not promote. With many participants, a behaviour is harder to attribute cleanly to one person, and the report reflects that.
A separate section gathers findings backed by at least two different frameworks: no reading there rests on a single model.
49 models, ten families
Each model comes from an identifiable field of research — work psychology, conflict management, negotiation… What you see below are the families; the detail of each framework appears inside the report, at the moment it is actually used.
Clinical
5Thinking patterns, sense of self-efficacy, attachment style at work.
Cognitive
4Attribution bias, distorted readings, dissonance — how a situation is interpreted rather than simply experienced.
Commercial relationship
3Customer journey, objection handling, structured sales technique.
Conflict
3Styles of handling disagreement, degrees of escalation, what pits interests against power.
Interpersonal
7Assertiveness, trust, non-violent communication, recurring relational patterns.
Management
5Delegation, leadership styles, structured feedback, sources of authority.
Negotiation
6Fallback positions, possible margins, anchoring effects, concessions actually observed.
Organisation
4How decisions are distributed, how information flows, roles and responsibilities.
Team dynamics
4Stages of team formation, collective processes, each person's position in the exchange network.
Work psychology
8Job demands and workload, perceived fairness, psychological safety, the implicit contract with an employer.
What this page does not do: a page for every single model. With 49 frameworks, a dedicated page for each would be a project of its own — for now, each model is introduced inside the report where it is applied, with its limits stated alongside it.
Anonymisation, in your browser
Full names, phone numbers and addresses are replaced with initials and a role — "Manager J.D." — on your device, before anything is sent. The substitution happens there, and nowhere else: no table anywhere links an initial back to a name, so there is none to store or protect.
What this does not do: in a team you know, initials can still be recognisable. This is pseudonymisation, not full anonymity — which is why an acceptable-use commitment is asked for before the report is handed over. The content of the messages, though, is genuinely sent for analysis: better to say so than let you assume everything stays on your device.
What the upload accepts
Frequently asked questions
How long does it take to get my report?
Once payment goes through, the analysis takes about two minutes. The upload itself is immediate for a text export; for screenshots it usually takes a little under a minute longer, while the text is extracted from them.
Does the report replace advice from a professional (lawyer, occupational physician, HR)?
No. ScanMyJob establishes dated facts and sets them against theoretical models — it makes no legal qualification, no diagnosis, and does not tell you who is right. For a decision that matters, what the report documents is something to bring to that professional, not a substitute for them.
What data is kept, and for how long?
Names, phone numbers and addresses are replaced with initials on your device before anything is sent — only the anonymised content is transmitted for analysis. The link to your paid report stays active for 30 days after payment; an unfinished upload, on the other hand, is kept for only six hours.
What happens if my conversation has more than 15 participants?
The upload refuses the file and says so, rather than silently analysing part of the group. Beyond fifteen, the least active positions can no longer be told apart clearly enough to be described accurately.
What formats are accepted?
WhatsApp, Slack and Teams exports, email text, and screenshots — for those, the text is extracted before anonymisation.
Is there a subscription, or price tiers?
No subscription: two tiers to choose from, Essential and Premium, each a single payment, no account needed. They differ only in how many frameworks the report retains — same corpus, same anonymisation engine, same template for both. The tier is chosen before payment, never added afterwards.
Can the report establish that I am being harassed, or say that I am right?
No, and that is deliberate. The report counts, dates and places things in sequence — it does not rule on who is right or wrong, and puts no label on someone who has not been heard. A dated timeline holds up better than an automatic verdict, which can be dismissed in a sentence.
Can I share the report with someone?
Yes, it is yours. It names no one: participants appear by their initials and role. A shorter record of facts can also be exported separately for that purpose.
See what your report would contain
Upload your exchanges, choose your objectives: the report applies this same method to your own situation.
Upload my exchanges