Answered from these docs only, by a model that cannot see your account. Check the pages it cites.

Running AI value measurement: the first 90 days

Who owns what, what to agree before results, the monthly close and the quarterly review: the process around Roiva, and what it leaves to you.

View as Markdown

Answered from these docs only, by a model that cannot see your account. Check the pages it cites.

Checked against the product on October 19, 2026

Roiva holds the figures, but the figures hold up only if a few people agree how they're counted, approve them each month and act on them each quarter. This page sets out that process: who does what, in what order, and how often. The articles it links to say how to do each step in the app.

It assumes the AI is already being rolled out, or is about to be. Choosing use cases, training people and changing how the work is done are yours to run. Roiva measures whether that work is paying back.

Who does what

Name these people before anything else. One person can hold more than one role in a small team, but the person who records a figure shouldn't be the one who approves it.

  • Program owner: whoever answers for the AI portfolio as a whole, often a Chief AI Officer, head of AI or head of transformation. Owns the account, sets the review cadence and runs the quarterly review.
  • Finance sign-off: someone in finance (FP&A, the controller's team) who signs the account's measurement charter, approves value and locks each month once it's closed. This is the person whose acceptance makes the figures stand up to a CFO.
  • Initiative owner: for each initiative, the business lead whose team does the work. Drafts the initiative's charter, enters the figures no system reports and answers for the initiative at review.
  • Cost owner: whoever sees the bills, usually in finance or procurement. Gets the books or card platform connected and says which spend belongs to which initiative.
  • System admins: whoever administers the help desk, CRM, cloud billing or data warehouse. They're needed to connect each system, not to run the program.

Roles in the app decide who can do which of these. Connecting a system takes an Owner or Admin, so either invite whoever connects it as an Admin or have an Admin connect it with them. Signing the account charter takes an Owner or Admin, and approving value above the senior approval threshold takes an Owner, or someone an Owner has delegated approvals to, so make the finance sign-off an Owner if they approve the large figures. An initiative's approval chain names who approves its value below the threshold. See Inviting your team and Approving value and initiatives.

Before kickoff

  • Pick a few initiatives, not the whole portfolio. One to three, ideally including one that's already running, so there are figures from the start. Add the rest once the first ones are measured. Recording when a running initiative went into use calculates each past month that has all its readings and a baseline that isn't Roiva's estimate, so sign its charter first (see below). Months your systems don't hold are typed in under Past months. See The initiative lifecycle for an initiative that was in use before it was tracked here.
  • Start access early. Security review and admin time are what usually hold up the first month, not the figures. Ask for them before kickoff.
  • Find out where the baselines are. Roiva measures a baseline from the three complete months before go-live, where a connected system or warehouse query holds them, whether it was connected before go-live or after. Where nothing holds that history, time the work before go-live if you still can. Once the AI is in use with no history, the baseline has to be reconstructed or estimated, and every report says so. See Baselines for an initiative that's already live.

Weeks 1 to 4: agree the rules before the results

  1. Set up each initiative. Its Health tab lists what it needs: a value formula, the systems that feed it, the figures only your team has, and its owner. Work down it until the count is complete. See What an initiative needs before it has a number.
  2. Set each baseline. History first, then a time study, then a declared estimate, in that order of preference. A time study needs a window of working days and at least three people timed in each arm, so start it in week one. See Running a time study.
  3. Sign the charters. Finance signs the account charter, the rules every figure is counted by. Each initiative's charter says what the AI gets credit for and what it doesn't, and is signed by someone other than its drafter. Signing before any value is approved, and for an initiative already running before you record when it went into use, is what lets you say the rules were agreed before anyone saw the number. See Measurement charters.
  4. Set the plan. An initiative's planned CapEx, OpEx and value are its business case. Approving it sets its budget from the planned CapEx and OpEx; planned value is what its results are compared against, not part of the budget. Its time-to-value estimate says when it should first deliver, and without one there's no schedule to measure it against. See Budgets and budget vs. actual and The initiative lifecycle.

Every month

A value formula's entries for a month can be approved once the month has ended. A short routine after each month end keeps the figures current:

  • Bring the figures in. Connected systems sync on their own. Figures typed by hand, such as a month's count or a re-measure that's due, show in the Inbox.
  • Approve the value. Entries wait in their approvers' Inbox. Read what each one rests on before approving it: a Roiva default, an estimated baseline, or a figure counted outside its charter.
  • Check the costs. Make sure each initiative's spend has reached it, and that nothing is counted twice. See How costs reach an initiative.
  • Lock the month. Once the month's costs have all arrived (costs count as soon as they're recorded, with no approval) and its value is approved, finance locks it so the figures can't change without a recorded reason. See Allocation rules and accounting periods.

Every quarter

  • Decide on each live initiative. Continue, scale, fix, pause or retire, each with a reason and kept with the figures it was made on. On Initiatives, filter Gap to Decision due to list what's due, and choose Record a decision on a row to walk through them. Roiva records the decision; it doesn't make it. See Portfolio reviews and decisions.
  • Report it. The Board Brief puts the quarter beside the one before, and says what the figures rest on and what could still move. Management commentary prints once an Owner or Admin approves it. See Board Brief and reports.
  • Revisit the rules. If a charter needs to change now that results are in, revise it. The revision is recorded, with the value approved under the old version, and the quarter's brief says so.

What you should have at day 90

  • A signed account charter, and a signed charter for each measured initiative.
  • At least one month locked, with its value approved by someone other than whoever recorded it.
  • A portfolio decision on each live initiative, with a reason and a next review date.
  • One Board Brief that a CFO can read without the spreadsheet behind it.

What Roiva leaves to you

  • The rollout itself: choosing use cases, training, adoption and changing how the work is done. Record what's holding one back as a blocker on its Activity tab, so it's beside the figures at review. See Activity, notes and blockers.
  • The judgment of what caused a change. A charter states what the AI gets credit for; it doesn't prove the AI caused it. See the methodology page.
  • The decisions. Roiva keeps every decision with its reason and figures, and schedules the next review. Which initiative to scale, fix or cut is still yours to decide.