For ops and leadership teams of 10–150
Make the decision.
Keep the learning.
Decize is one durable record of what your organization decided, who owns it, what actually happened, and what the next person should know. Decisions come in by email or from the web, and the record follows them all the way to the outcome.
$99/month for your whole team — never per seat. First 10 decisions a month free. See pricing.
Move our customer data region from US to EU
Renewing in the current region keeps EU customer records outside the EU for another two years. Migrate before the 30 June renewal.
Original capture
“Re: data residency — the Helsinki team can’t sign until records sit in the EU. Renewal is 30 June, so if we’re moving we decide this month.”
via Email · acme@in.decize.app
Illustrative record. Not a customer’s data.
What it’s for
Your organization records the discussion.
It almost never records the decision.
The thread scrolls away. The owner drifts. The forecast nobody wrote down turns out to have been wrong, and six months later the same argument starts again from zero.
Where the decision lives now
Where it lives in Decize
Move our customer data region from US to EU
Owner Priya R. · decided 4 Mar · shipped 22 Jun · forecast resolved yes
Lesson: dry-run the export before you commit to a window — the estimate was two hours optimistic.
Move our customer data region from US to EU
Lesson: dry-run the export before you commit to a window — the estimate was two hours optimistic.
Resurfaced on the next decision like it
New decision
we should move the analytics warehouse to the EU too
What we already learned about this
Your team has made one close call before: the customer-data region move, which landed successfully but ran long. The lesson on that record was about the maintenance window, not the migration itself.
Watch-outs
A similar migration last year ran over its maintenance window.
The team’s track record on calls like this: 2 of 3 turned out well.
Illustrative record. Not a customer’s data.
Six months later, the ledger pays you back.
Someone starts typing a new decision. Before they finish, Decize pulls the closest calls your team has already made, with how each one turned out, and writes a short brief from them. Watch-outs come only from the priors that ended in regret. If there is nothing close, it says so — “no close past decisions yet, this looks new for your team” — rather than inventing a precedent to look clever.
The meeting ends.
The record doesn’t.
What it costs you to use
You write a sentence. It writes the record.
Nobody fills in a form. A decision arrives as the message somebody already sent, Decize proposes every field from it, and a person corrects what is wrong. The proposal is labelled a proposal until they do.
What a person types
“Helsinki can’t sign until our customer records sit in the EU. Renewal is 30 June, so if we’re moving we need to decide this month.”
One sentence. Twenty-eight words.
Forwarded from the thread, or pasted into the box. No form, no fields, no template.
What comes back, drafted
Title
Move our customer data region from US to EU
Owner
Priya R. — resolved from the sending address
Deadline
30 Jun
Reversibility
One-way door · high importance
Success criterion
All customer records served from the EU by 30 June
Decize’s read, not a source of truth — and it says so on the record.
Every field is a proposal, labelled as one. Nothing is scored or counted while it is still a draft.
What you actually do
Reversibility
One-way door · high importance
Success criterion — corrected
…by 30 June, with under five minutes of customer-visible downtime
Confirmed — now it counts
A confidence the model inferred never reaches your calibration until a human owns it.
Correct what is wrong, confirm the rest. On this record that was one field and one tap.
The record
And this is what it is worth having.
Every field exists because it changes what happens next — who chases it, what would have to be true, what you promised yourself you would check, and what you will want to know when it comes back around.
Original capture
The original message, kept verbatim.
Decize drafts the title and the summary, and keeps the exact text it came from underneath. Months later you can still see what was actually said, not just what somebody summarised.
Reversibility
Most decisions never see this frame.
A title, an owner and a date is the whole record for a two-way door, and that is most of them. Only a one-way or high-importance call earns the full frame below — and there, Decize holds the decision open until it is complete rather than letting you mark a big call as made on a sentence.
Options
Options, with the reason each one lost.
At least two labelled alternatives on a high-stakes call, each with what it had going for it and why it was rejected. The losing option is usually the one you need in twelve months.
Assumptions
Assumptions that can be proven wrong.
Each carries a confidence, the thing that would falsify it, and a date to check it. This is the part a retrospective can actually use.
Forecast
A forecast you can be scored on.
A yes/no question, a probability, a resolution date. It cannot be resolved before that date, and once you record what happened it cannot be edited.
Frame lock
The frame locks the moment the call is made.
When the decision moves to decided, the reasoning becomes read-only — enforced in the database, not just in the interface. Nobody gets to rewrite the story after the result is in.
Values
Read against what you said you stand for.
Declare your organization’s values once and every decision is read against them, with the tension named where there is one. It stays an estimate captioned as Decize’s read until a person confirms it — an unconfirmed score never counts toward anything.
Move our customer data region from US to EU
Renewing in the current region keeps EU customer records outside the EU for another two years. Migrate before the 30 June renewal.
Original capture
“Re: data residency — the Helsinki team can’t sign until records sit in the EU. Renewal is 30 June, so if we’re moving we decide this month.”
via Email · acme@in.decize.app
Reversibility
One-way door · High importance — the full frame below is required before this can be marked as decided.
Decision question
Do we migrate the customer data region to the EU before the 30 June renewal?
Success criterion
All customer records served from the EU region by 30 June, with no customer-visible downtime over five minutes.
Options
- Migrate now — unblocks the EU contracts, costs one sprint. Chosen.
- Renew in place — cheapest this quarter. Rejected: locks the constraint in for two more years.
Assumptions
The export tooling can move the dataset inside one maintenance window.
70% confident · check by 12 Mar · falsified if a dry run runs over four hours
Forecast
Will the migration finish before 30 June with under five minutes of downtime?
75% · resolves 30 Jun
Illustrative record. Not a customer’s data.
How it works
Three ways in. One loop out.
Nothing here needs a new meeting or a new ritual. The record only works if writing it down beats not writing it down, so the loop closes in the inbox the decision arrived in.
Forward the email
Every workspace gets its own capture address. Forward a thread and it becomes a record with a title, an owner and a due date — and people on your roster can contribute by email without ever creating an account.
Type it the way you’d say it
The capture box takes plain language. Decize drafts the summary, the owner, the deadline and whether it is a one-way door — and if it looks like something already in the ledger, it shows you that one first.
Bring what you already decided
Paste a retro, an architecture decision record or a project history, and Decize pulls out the decisions the document already records — with the outcomes it already knows — for a person to approve row by row.
Capture
Email, the web box, an import or a harvest.
Frame
Question, options, assumptions, a forecast.
Decide
The frame locks and the moment is stamped.
Do
Owner, deadline, sign-off, nudges.
Outcome
One tap, straight from the email.
Review
What changed, and why.
…and it comes back as a precedent on the next decision like it.
Open decisions near their deadline nudge their owner on the cadence you set, and escalate when they go past it.
When it is time to say how it went, the email carries the answer buttons — went well, mixed, regret, didn’t matter. One tap, no login.
A recorded outcome fires the retrospective automatically: what you predicted against what happened, plus a one-line lesson that stays on the record.
What leadership gets
A read on the system, not a read on your people.
One dashboard, computed from a view of the ledger that physically contains no title, no summary and no message text.
Decision Health
How we decide, follow through, and learn
Open
9
3 overdue
Overdue
3
open past deadline
Done
14
of 26 total
Total
26
all-time
Needs attention now
- Move our customer data region to the EUForecast ready to resolve
- Second support engineer before Q4Overdue (12d)
- Sunset the legacy importerNo owner
Loop health
Outcome closure
71%
5 of 7 mature
On-time execution
—
needs 5 · 4 elapsed
Owner coverage
85%
22 of 26 decisions
3 decisions are overdue.
This week
Capture is steady and owner coverage is holding. Two one-way doors are still waiting on sign-off, and the oldest overdue decision has now slipped twelve days.
Decize summary of your numbers — guidance, not a source of truth.
Illustrative dashboard. Not a customer’s data.
It answers whether decisions are closing, whether they close on time, and whether anyone owns them — then names the single most useful thing to fix this week. The same calculation writes the weekly email, so the dashboard and the digest can never tell you different things.
Note the dashes. Any rate that could be traced back to one individual stays blank below five data points, and forecast calibration needs five resolved forecasts — which takes months. The numbers arrive when they mean something, not before.
What to expect, honestly
Week one is useful. Month six is the point.
Some of this works on your first decision. Some of it cannot exist until you have a track record, and we would rather show you the shape of that than let you find out in week two.
From your first decision
The queue stops being a guess.
Open, overdue, unowned, awaiting an outcome, forecasts ready to resolve — one attention list, plus counts, median days from capture to decided, and the whole ledger exportable as CSV.
After a handful
The ledger starts answering back.
New decisions surface the closest calls you have already made and what came of them. Recurring themes are written from your own records. Loop-health rates appear once there are five data points behind them.
After a few months
You find out how good your judgement actually is.
Forecast calibration — a Brier score, your skill against the base rate, a reliability plot — needs five resolved forecasts, and the outcome-closure run chart wants about ten weeks. This is the number nobody else will ever show you.
If your ledger starts empty, it stays quiet for a while. That is the honest reason the import exists — paste the retros and architecture decision records you have already written and the ledger starts with a history instead of a blank page.
What Decize refuses to do
There is no surveillance layer in here.
A decision ledger is one bad product decision away from becoming an employee-monitoring tool. Three refusals keep it from becoming one.
No per-person scores, ever — no leaderboard, no individual quality score, no manager drill-down. The per-member analytics screen was deleted, not hidden.
Any rate that could identify one person stays blank below five data points.
The dashboard is computed from a view with no title, no summary and no message body, so a reporting bug cannot leak decision text into a chart.
Which decisions belong in here
Inside a workspace there are no private decisions — every active member reads the whole ledger, including the original message. So this is for the calls you would already explain at an all-hands: strategy, architecture, vendors, process, priorities, the one-way doors. It is not the place for compensation, individual performance, or anything else you would not put in a document the whole company can open. We would rather scope that for you now than have you discover it with the wrong record in the ledger.
And the part we would rather you heard from us: Decize is in beta and we have not been audited — no SOC 2, no penetration test. Decision text goes to Anthropic, with Google as a fallback, only for the feature you asked for; related-decision embeddings are generated inside our own Supabase infrastructure. Workspace isolation is enforced by Postgres row-level security rather than application filtering. What is fenced is the organization, not the individual.
Read our own decision ledger →All six refusals, and the gates →What your legal team will ask →Security →
Start the ledger with one decision.
Create a workspace, forward one email into it, and look at what the record becomes when it is actually written down. If it is not obviously better than the thread it came from, you have lost ten minutes.
Unlimited people · $99/month · nothing to cancel