The Sprint Method: the six-step diagnostic behind every AI Opportunity Roadmap
Diagnose before you prescribe. Quantify before you build.
The Sprint Method is the six-step diagnostic that produces every AI Opportunity Roadmap. It runs over three to four weeks, scoped to one firm at a time. Discovery & Outcome Definition. Process Mapping. Bottleneck Identification. ROI Quantification. The Opportunity Matrix. Prescription. Each step has a job; each step earns the next step's time.
This page walks through each step in detail, with the worked examples behind it. If you want the offer summary, start with the AI Opportunity Roadmap landing page. If you want to see the method applied end to end, the worked case study sits at the bottom of this page.
Premise_01 // Why the Sprint Method exists
The Sprint Method exists to answer the one question most AI consulting skips: which workflows are worth pointing AI at, in what order, and for how much recovered value.
Most AI consulting falls into two camps: selling tools, or selling open-ended strategy with no clear finish line. Both skip that question.
You do not become an AI-mature business by buying more AI. You get there by being deliberate about where to deploy it, on which problems, for which teams. That discipline comes from years of systems design, experimentation, and implementation. It is the move that makes AI compound.
The Sprint Method brings that deliberation into a methodical diagnostic. It is the diagnose-before-prescribe spine that runs through every engagement.
Step 01 // Discovery & Outcome Definition
Turn vague wishes into definable outcomes
We start by turning vague wishes into definable outcomes. 'Help with marketing' becomes 'move the first response to a new lead from four hours to 30 seconds.' 'Save time on reports' becomes 'cut the monthly board pack from 14 hours to under three.'
Every outcome is scoped to the workflow and the firm. That becomes the standard for the rest of the diagnostic. If a workflow does not have a clear outcome by the end of step one, it does not earn process mapping time in step two.
This is also where we agree on the prioritisation lens: cash recovery, capacity creation, scaling readiness, or risk reduction. That lens shapes how we read the matrix in step five.
Step 02 // Process Mapping
Map the work as it is actually done
This is the AI workflow audit step. We interview leadership to understand where they think the firm should be going, and operators to see what the work actually looks like. Those two views rarely match. That gap is usually the first finding.
We then map every mouse click in the workflows we have scoped. Not the org-chart version, but the work as it is actually done. Click by click, system switch by system switch, handover by handover. Where the operator opens a tab, copies a value, switches systems, waits on a colleague, or retypes a number the CRM already has.
In every process map I have run, the documented workflow is missing 30–60% of the real work: the system switches, the manual retyping, the wait time. Surfacing that gap is the foundation for everything that follows.
Step 03 // Bottleneck Identification — the "Then What?" method
Ask "then what?" until the loop is exposed
A bottleneck is a specific point in the process where time gets lost. We find them by relentlessly asking 'then what?' at every step until the manual loop is exposed.
That last loop is the bottleneck. It will not show up on an org chart, and it rarely appears in process documentation. It only surfaces under 'then what?' pressure.
Step 04 // ROI Quantification
Tie every bottleneck to a loaded hourly cost
This is the AI ROI calculation step. I tie every bottleneck to a loaded hourly cost: the full cost of an hour of someone's time, salary plus overheads. The arithmetic is the same in every engagement I run:
Recovery rates are honest. AI typically closes 60–85% of a bottleneck. The rest is review time, edge cases, and the work that still needs human judgement. I state the recovery rate and the assumptions behind it for every bottleneck on the Roadmap.
Nothing makes it onto the Roadmap unless the arithmetic clears the threshold I set with you in step one: typically a minimum of £5,000 recovered annually against the build effort, scaled to the workflow's frequency. If a bottleneck only recovers £400 a year against five days of build effort, it does not survive the matrix.
Step 05 // The Opportunity Matrix
Quick wins fund bigger bets
This is the AI prioritisation framework. The matrix has two columns. The financing logic is explicit. Quick wins fund bigger bets. Three or four moves that recover £30,000 in the first month can finance a four-week build that recovers another £80,000 over the year.
Quick wins
High recoverable value, low effort, no foundation gap.
Bigger bets
Also high value, but need more effort or foundation work first.
Below the matrix sits the foundation layer map: the centralised knowledge base, documented processes, and clean data structure that have to exist for higher-leverage opportunities to land. Some firms can run quick wins immediately and spend a quarter on the foundation. Others need to invert the order. The matrix makes it clear which path fits your firm.
Step 06 // Prescription
Connect specific problems to specific solutions
I connect specific business problems to specific AI solutions. Sometimes it is an existing tool, sometimes a custom build, a process redesign, or an agentic workflow (an AI-run sequence that completes a task end to end without human prompting at each step). Whether the answer is an off-the-shelf product, a scheduled AI assistant, a custom connector between AI and your existing systems, or simply 'don't buy anything yet, fix the foundation first,' the prescription only comes after the diagnosis is complete.
The Roadmap names the type of solution, not the vendor, unless you ask. Vendor neutrality is part of why clients trust this process. There is no commercial conflict in any line.
The result is the AI implementation roadmap your team can actually act on: type of solution per opportunity, sequenced, and tied to the arithmetic.
The Sprint Method ends here. If you want the systems it recommends to be maintainable and extendable by your own team, there is an optional capability layer.
Prioritised opportunities & recovered value
Quick wins
Bigger bets
Worked example // UK retail bank
The method applied: £16,575 of contractor cost recovered per workflow
Here is the Sprint Method run end to end on one workstream at a UK retail bank.
Workflow: structured component review against four quality standards, with every finding turned into a user story on the development backlog. Manual baseline: five hours of three roles per item. Loaded cost: £343.75 per review.
The bottleneck was not the depth of the review. It was the coordination cost between calendar and backlog. The prescription replaced the three-role loop with a five-stage AI-driven workflow, run end to end by one engineer in 20 minutes.
Annual recoverable value at one review per week: 243 engineer hours and £16,575 of contractor cost. A 93% reduction in time and cost per item.
The full case study walks through how each of the six steps applied to this workflow, what surfaced, and what the final prescription looked like.
Reference // Frequently asked questions
Frequently asked questions about the Sprint Method
The Sprint Method is the six-step diagnostic that produces every AI Opportunity Roadmap: (1) Discovery & Outcome Definition, (2) Process Mapping, (3) Bottleneck Identification using the 'Then What?' technique, (4) ROI Quantification, (5) The Opportunity Matrix, and (6) Prescription. It runs in 3–4 weeks, scoped to one firm at a time.
The Sprint Method identifies AI opportunities by mapping each workflow click by click (Process Mapping), exposing the manual loops where time is lost (Bottleneck Identification via the 'Then What?' technique), then costing each one (ROI Quantification). Only bottlenecks that clear an agreed recovery threshold make it onto the Roadmap: typically a minimum of £5,000 recovered annually against the build effort. Identification is evidence-led, not opinion: every opportunity is tied to hours lost, annual volume, and a stated recovery rate.
The Opportunity Matrix is the prioritisation step of the Sprint Method. It sorts every quantified bottleneck into two columns: quick wins (high recoverable value, low effort, no foundation gap) and bigger bets (high value, but dependent on more effort or foundation work first). The financing logic is explicit: quick wins fund bigger bets. The matrix is what makes the 90-day plan in the Roadmap sequenced rather than wishlist-ordered.
The foundation layer map sits beneath the Opportunity Matrix. It names the centralised knowledge base, documented processes, and clean data structure that have to exist for the higher-leverage opportunities to land. Some firms run quick wins first and spend a quarter on the foundation; others invert the order. The map makes the dependency explicit so you do not try to build a bigger bet on a foundation that is not there.
The Sprint Method's Prescription step names the type of solution: existing tool, custom build, process redesign, or agentic workflow. It only names a specific vendor if you ask for it. Vendor neutrality is part of why the Roadmap is trusted. There is no commercial conflict in any line.
The full Sprint Method runs three to four weeks end to end. Discovery typically takes the first week. Process Mapping and Bottleneck Identification run together across weeks two and three. ROI Quantification, the Opportunity Matrix, and Prescription land in the final week, ending in the delivery workshop. The pace is deliberate: long enough to do the work properly, short enough to land while the trigger is still felt.
Book a scoping call
See the Sprint Method applied to your firm
The Sprint Method is how the work runs. If you want to see what it would look like applied to your firm, tell me about the workflows you would diagnose and the question you need the Roadmap to support. You will have a scoped quote inside five working days.