How to scope an executive reporting automation project

The baseline to measure before anyone writes a scope, what belongs in it, what gets added by accident, and the seven acceptance criteria to put in the contract as numbers rather than qualities. Five of the seven are checkable on the last day by a sponsor who never attended a working session.

A proposal to automate the management pack is in front of you. The useful question is not whether the firm can build it. It is whether the scope covers the hours you actually have, and you cannot answer that until you have measured them.

The baseline, before anyone writes a scope

The baseline to measure before anyone writes a scope

One cycle of measurement. Without it you cannot tell afterwards whether the project worked, and the business case is an assertion.

• Elapsed days from period close to the pack landing with its audience. Calendar days, weekends included, so it matches how the benchmark is defined and how the audience experiences it.
• Person-hours by step, timed separately. Pull, reconcile, chase exceptions, assemble, write commentary, rework. The split is the whole scope decision.
• Named people and their loaded cost. 480 person-hours a quarter at a known rate is a number a finance committee can act on. A percentage is not.
• Count of distinct numbers in the pack, and how many have one named owner. The ratio tells you how much of the pack is automatable today and how much would just be published faster and wrong.
• Rework rate: how often a reviewer sends the pack back, and for what. Rework is invisible in a process map and often the largest single block of time.
• The decisions that were late or not made because the pack arrived when it did. The company-level cost. Harder to measure and the only one an executive sponsor cares about.
• A software shortlist. Not a baseline. Shortlisting before the step timings is how a tool gets bought for a problem sitting two steps upstream.

Nine of the engagements in our case library published a before and after cycle time. All nine could because the before was measured first.

Source: Thinklytics case library, published delivery outcomes per engagement.

One reporting cycle of measurement. Six things.

Elapsed calendar days from period close to the pack landing with its audience, weekends included, because that is how the APQC benchmark is defined and how the audience experiences it. APQC's June 2026 publication puts top-quartile performers at six calendar days, median 10, bottom quartile 15, with no fieldwork date disclosed for the quartile split, so use it as a rough frame rather than a target.

Person-hours by step, timed separately: pull, reconcile, chase exceptions, assemble, write commentary, rework. The split is the whole scope decision. Most teams find it is not where they assumed.

The named people doing them, and their loaded cost. 480 person-hours a quarter at a known rate is a number a finance committee can act on. A percentage is not.

The count of distinct numbers in the pack, and how many have one named owner who can state the definition and the source without opening a spreadsheet. That ratio tells you how much of the pack is automatable today.

The rework rate: how often a reviewer sends the pack back, and for what. Rework is invisible in a process map and frequently the largest single block of time.

And the decisions that were late or not made because the pack arrived when it did. This is the company-level cost, it is the hardest to measure, and it is the only one an executive sponsor cares about.

Nine engagements in our case library published a before and after cycle time. All nine could only do that because the before was measured first.

What belongs in the scope

What belongs in the scope, and what gets added by accident

The out-of-scope column is where these projects overrun. Each item is defensible on its own and none of them shorten the cycle.

Ask for the out-of-scope list in writing. A proposal that has no out-of-scope section has not been scoped.

Source: Thinklytics engagement pattern across the 14 automated executive reporting engagements in the case library.

Six items, each mapping to a step in the cycle so you can check the scope covers the hours you measured rather than the hours that are easiest to automate.

The recurring pack, named by audience and frequency. Monthly management, quarterly board, statutory filing. These are different packs with different rules and different audiences, and treating them as one thing is how a scope becomes unbounded.

One written definition per contested number in that pack. The step that collapses the cycle. Contested ones only.

A scheduled extract, plus an exception report that runs before the pack. The exception report is the item most often missing from a proposal and it carries a lot of the benefit.

A template bound to the certified numbers. Removes the reformatting and most of the rework.

A review step a person signs, with a change log. Needed the first time an audience asks why a figure moved.

Human commentary. It is the analysis, and it is what the recovered hours are for.

What gets added by accident

Six things, each defensible on its own, none of which shorten the cycle.

Every report the company produces, rather than the recurring pack. Defining metrics nobody argues about, which looks like completeness and is where metric registry projects go to die. Real-time streaming for a pack that is published monthly. A redesign of the pack's layout and branding, which is a different project with a different sponsor. An approval workflow tool. And generated commentary, which changes the writing rather than the cycle and adds a verification obligation on top.

Ask for the out-of-scope list in writing. A proposal with no out-of-scope section has not been scoped, it has been described.

Acceptance criteria, as numbers

Acceptance criteria, written before the work and tested on the last day

Each one is checkable by the sponsor without attending a working session, and each is stated as a number rather than a quality.

• Elapsed cycle time hits a stated figure, measured the same way as the baseline. Name the figure in the statement of work. 3 weeks to 2 days, 6 weeks to 4 days, 18 days to 3 days: these were contracted, not discovered.
• Person-hours per cycle hit a stated figure. 480 to 32 per quarter at one insurer. The hours are what converts to money.
• Every number in the pack traces to one certified source without opening a spreadsheet. Test it by asking a named owner in front of the sponsor.
• The exception report runs before the pack and fails loudly. A silent pipeline that publishes a stale number is worse than a manual one, because nobody is checking.
• Reporting kept running throughout, with no missed cycle. Ask for it explicitly. A missed board pack costs more than the project saves in its first year.
• A named owner for the pack after handover, and the residual run cost. A proposal with no ongoing cost is hiding one.
• Zero findings or restatements in the first audit or exam cycle after go-live. Where the pack is regulated this is the criterion that matters most. One insurer had zero restatements across four quarters post-deployment; one bank had zero MRAs at the next exam.

Five of these are numbers. Write the numbers in before the work starts and the final review takes an hour instead of a negotiation.

Source: Thinklytics case library, published delivery outcomes and acceptance measures per engagement.

Written before the work starts, tested on the last day, five of the seven stated as figures. The property that matters is that a sponsor who never attended a working session can check them.

Elapsed cycle time hits a stated figure, measured the same way as the baseline. Three weeks to two days at a regional bank, six weeks to four days at an insurer, 18 days to three at a telecom, three months to two days at a public university. Those were contracted, not discovered afterwards.

Person-hours per cycle hit a stated figure. 480 to 32 per quarter at the insurer. The hours are what convert to money.

Every number in the pack traces to one certified source without opening a spreadsheet. Test it by asking a named owner in front of the sponsor.

The exception report runs before the pack and fails loudly. A silent pipeline publishing a stale number is worse than a manual one, because nobody is checking it any more.

Reporting kept running throughout with no missed cycle. Ask for this explicitly. A missed board pack costs more than the project saves in its first year.

A named owner for the pack after handover, and the residual annual run cost.

And where the pack is regulated, zero findings or restatements in the first audit or exam cycle after go-live. An insurer had zero restatements across four quarters post-deployment. A regional bank had zero matters requiring attention at the next exam, having gone in with four data quality issues flagged at the previous one. A public university had zero compliance findings in review.

Timeline and what drives it

In our case library these ran 8 to 20 weeks where the platform stayed and 16 to 26 where it did not. The driver is the number of contested numbers in the pack rather than the size of the company.

Eleven weeks took a statutory NAIC filing from six weeks to four days. Fourteen weeks put 67 school districts on one platform and took an eight-week submission cycle to three days, with zero change orders. Eighteen weeks took a quarterly call report from three weeks to two days across 140 defined line items. Twenty weeks certified 42 metrics across eight administrative units and took federal reporting from three months to two days. Twenty-six weeks replaced nine reporting systems and cut a monthly close from 18 days to three.

If a proposal quotes a duration before counting the contested numbers by name, the number is a guess. Counting them is a conversation, not a discovery phase, and it belongs before signature.

The residual running cost

There is always one, and a proposal showing none is hiding it.

At Kaiser Permanente the reconciliation labour attached to 14 regional definitions fell from $2.1M a year to under $180K, so roughly 9% of the eliminated cost stayed behind as the ongoing cost of running the governance. Ask what that figure buys: the exception report and its alerting, the test suite, the change log, the named owner of the definitions, and the forum that settles the next dispute.

Five questions before signing

Show me a before and after cycle time from a previous engagement, measured the same way both times. Which pack is in scope by name and frequency, and what is explicitly out. Who on my side has to be in the room, and for how many hours a week. What is the residual annual cost after handover. And what happens if two of my teams will not agree a definition.

The last one matters most, because it is a governance decision rather than a technical one. A firm that has done this work before answers with escalation to a named decision maker. A firm that has not answers with a workshop.

What we would do first

Measure the six baseline items over the next reporting cycle. Then count the contested numbers. Then read the proposal against both.

If the measurement shows the hours sit in reconciliation rather than assembly, the route is in automate management reporting without new systems. The payback test on a specific workflow is in when AI reporting automation pays back, and choosing between firms is covered in how to choose a BI consulting firm.

Delivery sits in AI reporting automation and analytics and BI, with data visualisation services for the pack layer, Power BI consulting on Power BI estates and SAP reporting and analytics modernisation on SAP. To size the case before asking for budget, the reporting baseline and benefits worksheet is the worksheet we use. The full set of work in this area sits under reporting takes too much work.

Frequently asked questions

What has to be measured before the project can be scoped?

Six things, over one reporting cycle. Elapsed calendar days from period close to the pack landing with its audience. Person-hours by step, timed separately across pull, reconcile, chase exceptions, assemble, write commentary and rework. The named people doing them and their loaded cost. The count of distinct numbers in the pack and how many have one named owner. The rework rate, meaning how often the pack comes back and for what. And the decisions that were late because the pack was. A software shortlist is not a baseline.

Why does the baseline have to come before the shortlist?

Because the step timings decide the route, and the route decides what you are buying. If the hours sit in reconciliation, a reporting tool addresses the wrong step and the project will deliver a faster version of the same disagreement. Nine engagements in our case library published a before and after cycle time, and all nine could only do that because the before was measured first. Without it there is no way to tell afterwards whether the project worked.

What belongs in the scope?

The recurring pack named by audience and frequency, one written definition per contested number in it, a scheduled extract plus an exception report that runs before the pack, a template bound to the certified numbers, a review step a person signs with a change log, and human commentary. Six items. Each maps to a step in the cycle, which is how you check the scope covers the hours you measured rather than the hours that are easiest to automate.

What gets added by accident and should not be?

Every report the company produces rather than the recurring pack. Defining metrics nobody argues about. Real-time streaming for a monthly pack. A redesign of the pack's layout and branding. An approval workflow tool. And generated commentary. Each is defensible on its own and none of them shorten the cycle, so each one pushes the delivery date past the point where the sponsor's attention runs out. Ask for the out-of-scope list in writing: a proposal without one has not been scoped.

What acceptance criteria should go in the contract?

Seven, five of them stated as numbers. Elapsed cycle time hits a stated figure measured the same way as the baseline. Person-hours per cycle hit a stated figure. Every number traces to one certified source without opening a spreadsheet. The exception report runs before the pack and fails loudly. Reporting kept running with no missed cycle. A named owner after handover plus the residual run cost. And zero findings or restatements in the first audit or exam cycle where the pack is regulated.

How long should it take and what drives the duration?

In our case library these ran 8 to 20 weeks where the platform stayed in place and 16 to 26 where it did not. The driver is the number of contested numbers in the pack, not company size. Eleven weeks took a statutory filing from six weeks to four days. Fourteen weeks put 67 school districts on one platform and an eight-week submission cycle to three days. Twenty weeks certified 42 metrics across eight administrative units. If a proposal quotes a duration before counting the contested numbers, the number is a guess.

What should the residual running cost be?

There is always one and a proposal showing none is hiding it. At Kaiser Permanente the reconciliation labour attached to 14 regional definitions fell from $2.1M a year to under $180K, so roughly 9% of the eliminated cost stayed behind as the cost of running the governance. Ask what the figure buys: the exception report and its alerting, the test suite, the change log, the named owner of the definitions, and the forum that settles the next dispute.

What should I ask a firm before signing?

Five questions. Show me a before and after cycle time from a previous engagement, measured the same way both times. Which pack is in scope by name and frequency, and what is explicitly out. Who on my side has to be in the room and for how many hours a week. What is the residual annual cost after handover. And what happens if two of my teams will not agree a definition, because that is a governance decision rather than a technical one and the answer tells you whether the firm has done this before.

The work behind this

Fourteen engagements in the case library carry automated executive reporting, and nine published a before and after cycle time measured the same way at both ends. Each states the duration, what was in scope and what the cycle became.

Automated executive reporting, 14 engagements.