# A baseline before the build. Then use at 30 days, value at 90.

Every AI system we lead starts with a figure your team already trusts, recorded before the build. We report against it every month, and finance reviews it with us every quarter.

## Four moments. One figure followed from start to finish.

The same two or three KPIs run through every step, from the diagnostic to the review with finance every quarter.

### 1. Before the build: The baseline

In the 4-week diagnostic we pick two or three KPIs for each use case, taken from reports finance or operations already use. Before the build starts, we record the baseline from the system of record or from a timed walk-through of real cases, and the design gate signs it off.

**What we measure**

- The figure: hours or cost per case, cycle time, error or rework rate, or beneficiary satisfaction
- How it is measured, and from which source
- Who reports it, by name

**What you get:** One row per use case in the AI Value Ledger: the baseline, the target and the person accountable.

### 2. 30 days after go-live: Use

Value follows use, so the first check is use. We measure the share of eligible cases that go through the new system, and the cases the system handed to a person.

**What we measure**

- The share of eligible cases handled by the new system
- Cases handed to a person, and why
- For training: the share of participants using the approved tools in their work

**What you get:** The first monthly value report, with the fixes that raise use.

### 3. 90 days after go-live: Value against the baseline

At 90 days we compare each KPI with its baseline, next to the value the business case first claimed. The business case's first figure stays unchanged, so the gap is visible.

**What we measure**

- Each KPI against its baseline
- Realised value against the business case's first figure
- The full cost so far: build, run, change and compliance

**What you get:** A gate decision: continue, change course or stop, with the reason in writing.

### 4. Every month, and every quarter: The finance review

Value is reported monthly, reviewed with finance every quarter. Finance checks the method, the baseline and the realised value, and the next operation is chosen on that evidence.

**What we measure**

- The monthly value report from the AI Value Ledger
- The review with finance every quarter
- The ranking of the next use cases

**What you get:** Value signed off by finance, and the next operation chosen.

## Four things finance agrees before it approves the value.

Finance stays accountable for the numbers. We prepare the evidence, and finance checks it.

1. **The baseline and its method:** Agreed with the team that runs the process, recorded before the build, from a source finance accepts.
2. **Realised value against the baseline:** Measured on the same KPIs, the same way, every month after go-live.
3. **The full cost:** Build, run, change and compliance, counted over at least three years.
4. **The gap to the business case:** The business case's first figure next to the realised value, so a shortfall leads to a decision, not a new estimate.

- [Get the AI Value Ledger](https://pioctopus.ai/en/toolkit/ai-value-ledger)
- [Download the Go-Live Gate Checklist](https://pioctopus.ai/en/toolkit/go-live-gate-checklist)
- [Read the guide to measuring AI ROI in Saudi Arabia](https://pioctopus.ai/en/insights/measure-ai-roi-saudi-arabia)

## We publish results in numbers. We keep our clients' names private.

Each published case gives the sector and scale, the challenge, what went live and the metric, with its before and after. The cases sit on our work page.

### Operations automation

Public sector · government services entity

An integrated system that automates operations end to end, on the entity's existing systems with real data. The before and after of manual work are on the work page.

### Digital transformation roadmap

Public sector · government entity

A full digital transformation roadmap, built with the entity from a starting point with no earlier roadmap.

We add a figure to the site only when the client agrees and the measurement behind it is documented.

[See the cases on our work page](https://pioctopus.ai/en/work)

## Questions finance and boards ask.

### When do you record the baseline?

Before the build starts, and the design gate signs it off. We pick the KPIs in the 4-week diagnostic, then record the baseline from the system of record or from a timed walk-through of real cases. A baseline recorded after go-live is hard to check, because the old process has already changed.

### What do you measure at 30 and 90 days?

Use at 30 days, and value against the baseline at 90 days. At 30 days we measure the share of eligible cases that go through the new system and the cases handed to a person. At 90 days we compare each KPI with its baseline and with the business case's first figure, and the gate decides whether to continue, change course or stop.

### What does finance sign off?

The baseline and its method, the realised value, the full cost, and the gap to the business case. Value is reported monthly, reviewed with finance every quarter. Finance stays accountable for the numbers; we prepare the evidence.

### Do you guarantee a result before the work starts?

We commit to the method and the targets agreed in the business case, and report against them. Each gate has a kill criterion set before the build, so a use case that misses its target leads to a planned decision: change it, or stop and move the budget to the next one.

### Why do your cases leave out client names?

Our clients' work is confidential, so we publish the sector, the scale and the result. Nothing about an engagement is published without the client's agreement. Procurement teams can ask for references from past clients, where they agree.

## Want to see how we would measure your first use case?

30 minutes with our team. We tell you plainly where we would start.

- [Book a session](https://pioctopus.ai/en/contact)
- [Request the diagnostic](https://pioctopus.ai/en/contact?interest=diagnostic)

---
Web page: https://pioctopus.ai/en/how-we-measure
Arabic version: https://pioctopus.ai/how-we-measure.md
Index for AI agents: https://pioctopus.ai/llms.txt
