Seats
Published · automatic
Applies to three instruments and no others. The band is read off a printed ladder and applied without asking, because a discount you have to ask for is a discount for whoever asks loudest.
Price is a function of three things: how many seats, how many entities, and how long a term. One of those is published and automatic. Two of them are negotiated, and this page says exactly why.
It also publishes the doors we have closed, and what would open each one. Most of that list costs us money to hold — which is the only reason it is worth reading.
Pre-launch. Nothing on this site is purchasable and every figure below is anticipated — indicative terms awaiting my ratification, never final, confirmed or current.
Software has no marginal cost worth discussing. Support does. One person can onboard a fixed number of desks in a quarter, answer a fixed number of escalations in a week, and hold a fixed number of conventions in their head at once. Support capacity is the binding constraint on this business, and it is the thing the three axes are actually measuring.
Published · automatic
Applies to three instruments and no others. The band is read off a printed ladder and applied without asking, because a discount you have to ask for is a discount for whoever asks loudest.
Negotiated
The support surface scales with entities, not with seats. Two affiliated desks are two onboardings and two escalation paths whether they hold four seats or forty, so a group licence is scoped rather than looked up.
Negotiated within a published ladder
Commissioning cost is front-loaded. A longer term amortises a real cost across more months of service, and the ladder below is our share of that saving handed back — arithmetic, not a sale.
Being explicit about this is not modesty. A vendor who cannot tell you what limits their pricing is either not constrained by anything, in which case the number is arbitrary, or is constrained by something they would rather you did not know.
The band is a reduction on the per-seat rate, applied at the boundary automatically. The deepest published reduction is 20% — stated here so that nobody has to negotiate towards a number we were always going to stop at. Above the banded range the licence is negotiated rather than banded, which is a different thing from a deeper automatic step and is shown as such in the table.
| Band | Reduction on the per-seat rate | What the band is measuring |
|---|---|---|
| 1–4 seats | LIST RATE | Annual list rate per seat. |
| 5–9 seats | 10% | Ten per cent off the annual list, per seat. |
| 10–19 seats | 20% | Twenty per cent off the annual list, per seat. |
| 20+ seats | NEGOTIATED | Group licence — negotiated, not banded. |
| More than one entity | NEGOTIATED | A group licence, not a bigger seat band. See group licences. |
A seat band applies to these and to nothing else. Everything else in the catalogue is flat-rate — one subscription for the account, however many people read it. Applying a seat band to a flat-rate instrument is a category error, not a favour: there is no per-seat rate for it to reduce.
One month less per year, per step down the ladder. That is the whole mechanism, and it is stated in months rather than percentages because months are what you actually pay.
| Term | Months paid per 12 served | Below rolling monthly | What you are committing to |
|---|---|---|---|
| Monthly, rolling | 12 / 12 | BASE | Cancel at the end of any month. You carry the whole commissioning cost yourself. |
| Annual term | 10 / 12 | 16.67% | The commissioning cost is spread across twelve months instead of one. |
| Two-year term | 9 / 12 | 25% | One commissioning, two years of service. Rate fixed for the term. |
| Three-year term | 8 / 12 | 33.33% | One commissioning, three years of service. Rate fixed for the term. |
Rolling monthly billing costs 20% more than the annual rate — 12 payments against 10. The annual rate is therefore a 16.67% reduction, not a 20% one. The two numbers describe the same pair of prices from opposite ends, and the larger of them is the one a sales page would print. We print both, and we prefer the months.
At the bottom rung you pay 8 months for every 12 months of service. Over three years that is:
8 months paid multiplied by 3 years equals 24 months paid
12 months served multiplied by 3 years equals 36 months served
24 months paid for 36 months of service — which is exactly what 2 years of rolling monthly billing costs. Three years of service for two years of monthly money.
Commissioning cost is front-loaded. Support concentrates in the first days of a licence — onboarding, conventions, the first real measurement against your own data, the first argument about what a number means. That is a real cost and it is paid once, not monthly. A longer term spreads it across more months of service, and the ladder is the customer's share of that saving.
The other half of the reasoning is preference, stated with a number rather than a sentiment: a client who commits for three years is worth more to me than a higher monthly rate, because a committed book is what lets one person publish the methodology and maintain the instruments instead of spending the year chasing renewals. The reduction is what that preference costs me, and it is published rather than negotiated case by case, so nobody has to ask for it in order to receive it. The same preference holds on engagements: a retained relationship is worth more to me than a higher single-engagement fee, multi-engagement scopes are agreed in writing, and what is agreed is the scope — never a gag on the findings.
It is also why the Terminal has no rolling-monthly rung at all: its support cost concentrates so hard in the first days that a short tenure burns the commissioning without ever amortising it. Its ladder starts at the annual step, and its evaluation route is a paid pilot credited in full on conversion rather than a free trial.
A reduction whose arithmetic you can check is not a dark pattern; a clock ticking beside a price is one. There is no clock on this site and no struck-through price either, because a price that was never charged is not a price — it is a stage prop. Every figure here is a rate we would actually invoice.
Available on any annual term. It softens cash flow, not commitment: the term is unchanged, the ladder rung is unchanged, and the total is unchanged. It is a payment schedule rather than a fourth tier, and it is listed here so nobody has to ask whether it exists.
A research group licensing the Epistemic Harness for 12 named researchers, on a three-year term. Every step is shown, because the point of the example is the order of operations rather than the total.
| Step | Arithmetic | Result |
|---|---|---|
| List rate | Epistemic Harness, one seat, billed monthly | £69.00 / seat / month |
| Seat band | 12 seats falls in the 10–19 seats band | £55.20 / seat / month |
| Term | Three-year term — 8 months paid per 12 months of service | £441.60 / seat / year of service |
| Seats | £441.60 × 12 seats | £5,299.20 / year of service |
| Term total | £5,299.20 × 3 years | £15,897.60 for 36 months of service |
The seat band is a factor and the term rung is a factor, so they compose by multiplication. A 20% seat band and a 33.33% term step do not land at 53.33%. They land at 46.67%.
Effective rate across the whole term: £36.80 per seat per month, against a list of £69.00. We publish the shortfall against the additive reading rather than letting a buyer find it in their own spreadsheet after the conversation, because that is the version of this number that costs us something to print.
A group licence is one agreement spanning affiliated entities: brands under a holding company, desks under a fund, venues under an operator, subsidiaries under a group. It is negotiated rather than banded, and the reason is specific — the support surface scales with entities, not with seats. Four seats in one entity is one onboarding, one escalation path, one set of conventions. Four seats across four entities is four of each, and no seat ladder can price that difference because a seat ladder is not measuring it.
The last point is the argument, not the invoice. An internal disagreement about a number is expensive in a way that a licence line item is not, and it is the failure mode a group licence is actually built to prevent.
Multi-network operators — several venues, several brands, one measurement standard — are the case this was designed around. If that is you, the scoping conversation starts with entity count, not seat count.
Every commercial arrangement we will enter, and every one we will not. The second list is the useful one: a vendor who will not tell you what they refuse has not decided yet, and the answer you get later depends on the size of the cheque.
OPEN NOW is in force today.AT LAUNCH exists when instrument checkout opens and is not purchasable today.NEGOTIATED is scoped in conversation rather than looked up on a table.
Pay-per-test on your own artifact: a fixed menu of run-once measurements — overfit, execution cost, dataset forensics and the rest — on the instruments’ published methods, returned as a written, hash-pinned report. Fixed menu and fixed price is what keeps it a product rather than consulting; the menu and anticipated pricing are published now, and intake opens with a payment rail and a proper submission channel (OR-030) The full menu.
One instrument, one subscription, no conversation with anybody. This is the default and it is designed to stay the default — support capacity is the constraint on this business, so a buyer who never needs us is the buyer the architecture is built for.
cTrader Store first, MQL5 Market second. The marketplace build is fully functional standalone at its price — never paid-but-crippled — and the site tier is priced at or below the marketplace-equivalent, so the platform fee never inverts your margin.
One instrument family, complete, on a single agreement. Every instrument in the family, its premium tiers, and the family receipt page. the seven suites.
The Terminal shell plus the engines you choose, at a stated reduction on the sum of the parts rather than at the sum. The shell is the provenance seat; the engines are peers that plug into it and are licensed separately. what a Terminal seat includes.
Published, automatic, and applied without asking. Three instruments are licensed per seat; the band is read off the ladder on this page, not negotiated case by case.
One agreement across affiliated entities, brands, desks or venues. Negotiated because the support surface scales with entities rather than with seats — two entities are two onboardings, two escalation paths and two sets of conventions, whether they hold four seats or forty.
A two- or three-year term fixes the per-seat rate for the whole term. List prices move; a signed term does not move with them. Renewal is at the then-current list, stated in the agreement rather than discovered at renewal.
The Integrity Services instruments are commissioned rather than subscribed, against published floors. The floor is the point: an engagement below it stops being product discovery and becomes a job. Integrity Services.
A completed engagement can convert to an ongoing retainer. It never runs the other way round — a retainer is not the entry rung, because the engagement is what establishes whether there is anything worth retaining.
The measurement artifacts, the methodology and the receipts corpus carry their own reuse terms, and those terms are in force today. No account, no waitlist, no permission request. the artifact licence.
Gated on role and affiliation rather than on price, at 50% of the annual list price, on the 6 instruments whose subject matter is research methodology. A licence is held by a named academic — faculty, permanent research staff or a principal investigator — and students are covered under that holder's licence rather than individually. eligibility and terms.
4 of these 11 are closed permanently — there is no offer, no scale and no future ruling that opens them. The rest name their own condition.
Measuring venues while taking venue money is the exact conflict the Observatory exists to escape. A venue cannot buy a measurement, buy a position in a list, buy an earlier publication date, or buy its portrait down.
What would open itNothing. There is no figure, no structure and no future scale at which this opens. If it ever did, every measurement published before it would be retrospectively worthless, which is the whole reason it cannot.
A rating pipeline and a revenue pipeline that are the same pipeline is the structure of every review site nobody believes. The stream was designed, costed, and then deferred by ruling before a penny of it existed.
What would open itPost-scale, with professional counsel, and only if the money can be structurally decoupled from the measurement — published disclosure on every surface, links present wherever a relationship exists regardless of what the measurements say, and a recomputable proof that nothing crossed. Today it cannot be, so the door stays shut with its hinges intact.
A finding you can suppress is not a finding. We hold relationships only where the right to publish accurate measurements is contractually unrestricted, and where a partner later introduces a publication restriction the relationship ends and the profile stays.
What would open itNothing. This applies to every counterparty class without exception — venues, clients, academics, and the entity paying for a configured engagement.
A single editorial number is precisely the judgement surface that corruption needs. Every venue gets the identical battery and a multi-dimensional profile; there is no composite, no star rating and no ordering.
What would open itNothing. Profiles, never rankings. Alphabetical order is the only order the Observatory publishes in.
These are analytical instruments for discretionary use. They measure instrument and process behaviour; they do not forecast, recommend, allocate or manage.
What would open itNothing. This is the first leg of the standing disclaimer and it is a design boundary rather than a legal hedge.
Never hourly, never small. Configured engagements carry published floors so that the work stays product discovery — a configuration that becomes a template SKU — instead of quietly becoming a job with one client and no product at the end of it.
What would open itNothing on the hourly axis. Work below the floor is declined rather than discounted, because a floor that bends is a rate card with extra steps.
Two at a time is the cap. Engagements are delivered by people, and the number of them that can run at once is a capacity fact rather than a commercial preference.
What would open itMeasured support capacity above the current threshold — not demand. Demand is the reason a cap gets broken; capacity is the only reason it should move.
Custom once, product forever. Every configuration built during an engagement becomes a template SKU available to everyone, including your competitors. We state this before anyone commissions anything, because a buyer who discovers it afterwards is entitled to be angry.
What would open itNothing on offer. The engagement fee buys the configuration, the delivery and the priority; it does not buy a private branch, a buy-out or exclusivity over the shape of the work. If that is a hard requirement, this is the wrong supplier and we would rather say so at the first meeting than at the third.
A free tool anchors the work at zero and never recovers. A paid tool with a trial teaches the price while letting you taste the product — and a paid-at-the-door listing has to stand on its own, which is a useful discipline to be forced into.
What would open itNothing. Trials run on each platform’s native mechanics instead, and the Terminal offers a paid pilot credited in full on conversion rather than a free one.
Not refused on principle — simply not built. There is no deployment path we could hand you today that we would be willing to support, and shipping one we cannot support is worse than not shipping one.
What would open itA ruling from me and a documented, supportable deployment path. If this is a hard requirement for you, say so on the launch list — a requirement stated before the architecture sets is worth more than one raised after.
A reseller channel puts a layer between the measurement and the person reading it, and support arrives through that layer having already been through a game of telephone.
What would open itA ruling I would only take post-scale. Not before there is a support organisation that can stand behind a partner’s promises as well as Hadal’s own.
The register above carries a version and an issue date for the same reason the methodology and the kill ledger do: a commercial policy you can rewrite overnight is a marketing page. A closed door only means something if the act of opening it leaves a mark.
Honestly: there is not much to negotiate with yet. Nothing is purchasable, there is no sales team, and no enquiry address is published — see the contact page for every channel that exists and every one that does not.
The one channel that works is the launch list. Join it and say what you need — seat count, entity count, term. Those three numbers are the entire scoping conversation, and sending them now means the conversation starts at the second question instead of the first.
Group and multi-year enquiries are read first. Not as a courtesy — long-term committed clients are who this business is built around, and a multi-entity requirement stated before the commercial architecture sets is worth considerably more to us than one raised after.
One home for what every instrument page states in a sentence. None of it is live: this site has no checkout, no card capture, and no product account to create.
Join the launch list. Group and multi-year enquiries are read first — they are the ones that shape what gets built.
Standing policy
The same statement stands in the footer of every page.Terms Privacy
Related: the pricing model ·refunds and cancellation ·terms ·artifact and data licence ·changelog