How support works
Every request is a ticket, at every tier. There is no phone line and no live chat: a measurement question deserves an answer with the sample examined, and a written answer is a record where a call is not. Which queue a ticket lands in, and how quickly someone reads it, follows the value of the contract rather than the price of a single unit. Nothing here is a service-level agreement, and there is a reason for that further down the page.
The honest version of why the model looks like this: published prices across the catalogue span roughly a hundredfold, while the cost of supporting one customer is roughly flat. A single uniform support level would mean either the smallest contracts consume the capacity or the largest ones quietly pay for them. The binding constraint on this business is hours, not money — so the tiers say plainly what each one buys.
The tiers
The threshold figures below are ratified. The list prices they are measured against remain anticipated, like every figure on this site, so a contract's tier is settled by arithmetic that is fixed even while the numbers going into it are not.
| Tier | Who lands here | Channel | First response |
|---|---|---|---|
| Self-serve | Everyone, customer or not. The documentation and the changelog are open to anyone, and always will be. The demonstration environment and the receipts corpus join them when they publish. | Documentation · changelog | No queue — nothing to wait for |
| Standard | Contracts under £1,000 a year. | Ticket | Three business days |
| Priority | Contracts from £1,000 to £6,000 a year. | Ticket, prioritised | One business day |
| Named | Contracts above £6,000 a year, plus the HADES Terminal, the Hadal Desk and every multi-entity group licence regardless of value. | Ticket, prioritised · named onboarding | Same business day |
| Academic | Any licence at the academic rate. | Ticket, from the named licence holder only | Five business days |
| Engagement | Commissioned audits — the Dashboard Lineage Audit and the Gate Falsification Kit engagement. | Direct, scoped in the statement of work | As the statement of work sets it |
A first response is not a fix. A first-response target is the time to a human reading your ticket and replying, not the time to a fix. Blurring those two is the standard way support pages mislead, so: the first is what the target names, and the second depends entirely on what broke.
Self-serve
- The full documentation centre, including the limits and refusal behaviour of each instrument.
- The methodology pages and the changelog.
- A read-only demonstration environment on synthetic data, and the published receipts corpus: both NOT YET PUBLISHED. They belong to this tier when they arrive, and they are not open today.
- This tier is not a lesser version of support. It is the part I would rather you never have to leave.
Standard
- One ticket queue, answered in turn.
- Defect reports read with the same seriousness at every tier — a bug is a bug regardless of what you pay.
- No onboarding session. The documentation is written to make one unnecessary, and if it fails at that, that is a documentation defect worth a ticket.
Priority
- Tickets are taken ahead of the standard queue.
- Written commissioning guidance for your first measurement.
- Advance notice of methodology version changes that affect a metric you rely on.
Named
- A named person who knows your deployment, rather than whoever picks the ticket up.
- Scheduled commissioning, because this is where commissioning cost actually lives.
- Methodology-version consistency across every entity on a group licence, so two desks in one group are never reading different definitions of the same metric.
- Scheduled calls where a written answer genuinely will not do. The record of what was decided still goes in the ticket.
Academic
- Tickets from the named holder. Students reach support through their supervisor, never directly — that routing is what makes a two-hundred-student cohort one relationship instead of two hundred.
- No onboarding session, and no commissioning support.
- Answered after the paid queues, honestly and without apology: the rate is half, and the queue position reflects it.
- A defect report is escalated to the normal defect path immediately. Discounted support is slower; it is not less careful.
Engagement
- Support is part of the engagement, not a tier bolted onto it.
- Scope, availability and duration are written into the statement of work before the work starts.
- When an engagement becomes a template SKU, the SKU carries the tier its contract value earns — the engagement terms do not follow it.
What decides your tier
Annual contract value, never unit price. Support tier follows the annual contract value, not the price of a single unit — more seats, a suite licence or a group agreement raise the contract value and can raise the tier with it. A twenty-seat licence of a per-seat instrument is a five-figure contract, not a small one, and banding on the per-seat figure would mis-tier every multi-seat and group customer on the list.
Standard runs below £1,000 a year, Priority from there to £6,000, and Named above it. The thresholds themselves are ratified; the anticipated list figures measured against them are not, so which side of a threshold a given contract falls on can still move before launch.
Where the band is overridden
Contract value is the rule; it is not the whole rule. 3 instruments carry a fixed tier regardless of what they cost, and so does every arrangement below. The HADES Terminal is the case that proves banding alone would be wrong: its support concentrates in the first days of commissioning, and its annual-only licence exists specifically to fund that, so tiering it as though it had an ordinary support profile would contradict the reason it is priced the way it is.
- The Hadal Desk bundle
- Any multi-entity group licence
- Any contract with a named onboarding fee
What support does not cover
Published here rather than discovered later. A customer who finds the boundary after paying is right to be annoyed, and for a product this close to money a vague boundary is how a support channel turns into an advice channel.
- Trading advice, strategy review, or any opinion on whether your system is good
- Hadal measures instruments, never portfolios. An answer to "is this strategy any good" would be discretionary advice, which Hadal is not authorised to give and would not give if it were.
- Interpreting a measurement into a decision
- I will explain exactly what a number is, how it was computed, and what it does not establish. What you do about it is yours — that boundary is the same one every instrument page draws.
- Custom development, bespoke metrics, or one-off analysis
- That is a configured engagement with a floor, not a support ticket. Mixing the two turns support into unpaid consultancy and starves the queue everyone else is in.
- Modified or reverse-engineered deployments
- Hadal cannot reason about a build it did not produce, and pretending otherwise would mean guessing at your provenance — the exact failure this company exists to argue against.
- Recovering data Hadal never held
- On-platform instruments run inside your own environment and their data never reaches Hadal. There is nothing on this side to recover, and saying so plainly is better than a hopeful reply.
Why there is no service-level agreement
The published response times are targets, not a service-level agreement, and Hadal does not publish one. A target is what gets aimed at; what is not on offer is an SLA, meaning a contractual promise with remedies attached, because I will not write a promise I have never measured myself against. What publishes instead, on the status page, is the measured distribution of actual first responses — median, ninetieth percentile, and the count it was computed from. Until enough tickets exist to compute it, that reads NOT YET MEASURED, because a support page inventing its own performance figure would be the exact failure the whole portfolio was built to catch.
It is worth being blunt about the trade. An agreement with remedies attached sounds stronger than a target, and for a large vendor it often is. What it is not is checkable: almost nobody audits whether a vendor met the agreement they published. A distribution of actual first responses, with the count it was computed from, can be read by anyone — including when it is unflattering. I would rather publish the weaker promise and the real number than the stronger promise and no number.
Measured response performance publishes on the status page when there are enough tickets to compute it from. Until then it reads NOT YET MEASURED.
Standing policy
- No promise of profit, ever. Hadal makes no performance claims and carries no implied edge. Past measurements describe instrument behaviour — never future returns.
- You own risk management. Hadal cannot control it and does not insure it. Good tools do not fix bad discipline — and this site says so.
- Analytical tools, for discretionary use. Nothing here is investment advice or a recommendation to trade. Every decision, and every outcome, is yours.
The same statement stands in the footer of every page.Terms Privacy
Reaching support today
The ticket ingress exists:support@hadalinstruments.com. A ticket is an email — the queue discipline, the tiers and the targets above are about how it is handled, not about the software behind it, and a dedicated tracker arrives when volume justifies one rather than before. Nothing is for sale yet, so today the queue's honest contents are questions about the site and the published method — welcome, and answered under the same asynchronous rules. The documentation centre at /docs/ remains the fastest answer for anything it already covers.