MLiQ · machine-learning pricing

Every price,
explained
never guessed.

MLiQ fits demand curves from your venue's own booking record, not your industry's, then prices each date off that fit and three explicit floors — cost recovery, displacement, and any minimum you set yourself. It shows the confidence band, and it shows its work.

56.4%head-to-head backtest vs. rate card
Everyprice ships with a band
8bookings before a cell is fitted
3CRMs connected natively
The product

A price is a
decision, not a
list entry.

§ 01 / product

A price set from last year's sheet and a feeling about the season is a guess made twice. MLiQ treats every date as its own market — lead time, day of week, tentative holds already on the date, comparable won bookings — and returns one number you can hold a conversation about.

01 · dates

Per-date recommendations

Each open date gets a recommended price with an explicit confidence band. Every evaluation returns a band with its point estimate — a recommendation never travels alone.

02 · leads

Lead scoring on your own demand

Inbound is scored on the venue's own weekday and season curves for the date requested, and every score ships with the sentence that explains it — so the team works the top of the list, not the top of the inbox.

03 · curves

Your curve, not the category's

Weekday, season and event-type curves are fitted from the venue's own booking record, on medians so one outsized booking cannot reshape a weekday. A cell with fewer than eight bookings keeps the shipped default and is reported as unfitted, never fitted on noise.

04 · defensibility

Every number ships with why

Each recommendation carries the inputs behind it — predicted value and band, arrival probability, breakeven and carrying cost — in one sentence. Nobody has to defend a price they can't explain in a room.

05 · verdicts

Four answers, not one price

Every evaluation lands on TAKE, HOLD, NEGOTIATE or PASS, with the price band and the sentence behind it. The decision is the product; the price is how it is expressed.

06 · multi-day

Per-day quoting for multi-day events

A three-day booking prices each day on its own weekday and season multipliers. A day you have blocked prices at zero with the reason named, so a quote never averages over a date you cannot sell.

07 · CRM

It lives inside your CRM

Salesforce imports closed-deal history — won deals become bookings, lost ones become demand observations — and every evaluation writes back onto the record. HubSpot and Pipedrive connect for lead import.

08 · controls

The operator sets the floors

Date classes, blackouts, a per-date minimum and your venue's own variable and fixed costs feed the floors directly. A tentative hold carries its own expected-value window and conflict check; a what-if is one call.

mliq.org/evaluate

Sample data Drawn procedurally in your browser — no screenshots, no stock imagery. The figures on it are illustrative, not a customer's.

Interactive model · runs in this page

The pricing lab.

§ 02 / interactive

A constant-elasticity demand model, solved in your browser as you move the controls — an illustration of the tradeoff, not the shipped engine. This is the shape of the question MLiQ answers for every open date on a venue's calendar: what does this Saturday cost?

Illustrative revenue curve · constant-elasticity demo

expected revenue = price × win probability · demo constants: 62% win at the $14,000 list price, 98.5% ceiling
modeled win rate
modeled expected rev
model optimum
gap to model optimum

Illustrative model Elasticity above 1 means demand falls faster than price rises — the regime where discipline pays. The lab holds cost constant so the tradeoff stays legible. Every number it prints comes from the demo curve above, not from MLiQ.

Under the hood

Honest about
its own limits.

§ 03 / model

A pricing model that only ever sounds certain is a liability. MLiQ reports the band, names every cell it could not fit, and downgrades a confident TAKE to NEGOTIATE when there is no comparable history behind the date.

Objective
Weigh holding the slot against taking the offer on expected value — P(another lead arrives in the hold window) × its value, less the carrying cost of waiting. A full calendar at the wrong price is a busy way to lose money.
Form
Multiplicative: base price per event type × weekday × month, fitted on medians. The lead-time term is a Poisson arrival rate — today an assumed baseline of one lead per 60 days, scaled by your weekday and season curves — over a hold window capped at 30 days.
Fit
Per venue, over event type × weekday × month, from its own booking record. A cell the history cannot support keeps the shipped default and is named as unfitted — nothing decays out silently.
Confidence
Every evaluation returns a band — a recommendation never travels alone. Today that band is a fixed ±20% around the point estimate, not yet a function of how much history sits under it.
Thin history
MLiQ does not abstain, it flags. A cell under eight bookings is reported unfitted, and a TAKE with no comparable bookings behind it is downgraded to NEGOTIATE rather than shipped as confidence it has not earned.
Provenance
Every evaluation is written to an audit row with its model version and tier — so “what did MLiQ say the day we quoted this, and what produced it?” has an answer.
Placement
Scored inline on the evaluate call, so pricing sits inside the quote flow instead of arriving as a nightly report nobody opens.
0head-to-head backtest win rate
0backtest MAE in $ · rate card $32.4k
0bookings before a cell is fitted
0CRMs connected natively

Basis for the backtest: MLiQ's fitted curves replayed against 666 billed reservation-days of one venue's own booking history and scored against the rate card that venue actually used — 56.4% head-to-head, MAE $29,308 vs $32,354, bias −$1,903 vs −$13,445. It is a retrospective, in-sample benchmark, not production traffic, and it measures agreement with the price the customer accepted rather than counterfactual profit. What MLiQ refuses to do: no fabricated metric, no invented delta, and a dash instead of a number when history doesn't exist yet.

Start a conversation.

No account and no sales sequence. Leave an email and a person replies from info@mzeiq.com.

Or email info@mzeiq.com directly.