When we sit down with an Australian SMB owner who has decided to "do something about automation", they usually have a tool, a budget, and a problem in mind — but no sequence. Three months later, the budget is half-spent, the tool is half-implemented, and the problem is still there. That is the gap a business process automation roadmap closes.
According to Deloitte's Automation with Intelligence survey, 73 per cent of organisations have started intelligent automation but only about 13 per cent have scaled it. The bottleneck is rarely the software — it is the absence of a sequenced plan from discovery to a running, measured workflow. Below is the 90-day roadmap we use with our own clients: six phases, one deliverable per phase, and the KPIs that tell you whether each phase actually worked.
Key Takeaways
- A business process automation roadmap is a sequenced 60–90 day plan that moves an SMB from process discovery to a measured, governed automation in production.
- According to Deloitte, 73 per cent of organisations have started intelligent automation but only ~13 per cent have scaled it — almost all of the gap is sequencing, not tooling.
- The 90-day roadmap below has six phases: Discovery, Prioritisation, Tool selection, Build and pilot, Measurement, and Handover and scaling.
- Each phase should produce one written artefact (a process inventory, an ROI scorecard, an architecture sketch, a pilot KPI report, a runbook) so progress is visible and auditable.
- The most common reason 90-day roadmaps fail is starting the build before the prioritisation phase has produced a scored, approved shortlist.
What is a business process automation roadmap?
A business process automation roadmap is a phased, time-bound plan that takes an organisation from "we want to automate" to "an automation is running, measured, and owned by a named person". It is not a tool shortlist and it is not a one-off audit. It defines the sequence of discovery, prioritisation, build, measurement, and scaling, with deliverables and a KPI at every step.
Automation programmes fail far more often from disorder than from bad tools. McKinsey Global Institute estimates that up to 45 per cent of work activities could be automated with current technology, yet typical SMBs have automated only a single-digit share of their eligible processes. The roadmap is what closes that gap — a living document where the phases, gates and KPIs stay fixed so leadership always knows where the programme stands.
Why does a roadmap matter more than picking the right tool?
A roadmap matters more than tool selection because most failed automation programmes had perfectly capable tools — they just deployed them into the wrong process, at the wrong time, without baselines or owners. According to Gartner, the hyperautomation-enabling software market is heading toward roughly US$600 billion in annual spend, yet adoption studies show only a small minority of buyers scale beyond the first workflow.
The industry has plenty of tools. What it lacks is the operating discipline to use them. Buying Zapier, Make, n8n, Power Automate, UiPath or any other platform without a roadmap usually delivers one of three outcomes:
- A single automation built by one enthusiastic person, with no documentation, that breaks the day they leave.
- A pile of half-built workflows that no one trusts enough to run in production.
- A working pilot that never scales because no one defined what "scaled" means.
A roadmap forces three decisions tool-led projects defer: which processes are in scope, what success looks like in numbers, and who owns each automation once it is live. Those three decisions are the difference between an experiment and an operating capability.
What are the six phases of a 90-day BPA roadmap?
The 90-day roadmap is six phases: Discovery (days 1–14), Prioritisation (days 15–30), Tool selection (days 31–45), Build and pilot (days 46–65), Measurement and tuning (days 66–80), and Handover, governance and scaling (days 81–90). Each phase has one written deliverable and a KPI that gates the move to the next phase.
| Phase | Days | Primary deliverable | Gate KPI |
|---|---|---|---|
| 1. Discovery | 1–14 | Process inventory with volumes and owners | All in-scope teams interviewed |
| 2. Prioritisation | 15–30 | Scored shortlist (volume × value × readiness) | Top 3 processes approved by leadership |
| 3. Tool selection | 31–45 | Tool decision memo + architecture sketch | One platform chosen and contracted |
| 4. Build and pilot | 46–65 | One automation live in production for a pilot team | Pilot ran 14 days without manual rework |
| 5. Measurement and tuning | 66–80 | KPI report vs baseline | Target KPI met within ±10 per cent |
| 6. Handover and scaling | 81–90 | Runbook, owner and scaling plan | Named owner accepts the runbook |
The phases overlap by a few days at the edges. That is realistic — discovery interviews surface tool questions early, and pilots produce measurement insights before they are formally "done".
Phase 1: Discovery and process inventory (days 1–14)
The discovery phase produces a written inventory of every candidate process: what it is, how often it runs, who owns it, which systems it touches, and how much time it consumes. The goal is not to decide what to automate yet — it is to make sure that when you do decide, you are choosing from a complete list, not the three processes that happen to be top of mind.
The inventory typically covers eight to fifteen processes across finance, sales ops, marketing ops, customer service, HR, and project delivery. According to the Australian Bureau of Statistics, most of Australia's roughly 2.5 million actively trading businesses are small or medium-sized — and most run daily operations with significant manual overhead. That overhead is what you are surfacing here.
For each process, capture five fields:
- Process name and one-sentence description
- Trigger (what makes it start)
- Frequency (runs per week or month)
- Systems touched (CRM, accounting, email, ticketing)
- Current owner and hours per week
Do the interviews live, not by survey — the second question is always more useful than the first — and ask people to walk you through the exact steps they did this morning, not the SOP version.
Pro tip
Pro tip: Time-box every discovery interview to 30 minutes and write the inventory entry inside the meeting. If you save it for "later", later never arrives — and discovery quietly stretches from two weeks to two months.
Phase 2: Prioritisation and ROI scoring (days 15–30)
The prioritisation phase turns the inventory into a scored shortlist of three to five processes. For each candidate, score volume (how often it runs), value (hours saved or revenue protected) and readiness (how documented, consistent and rules-based it is). Multiply the three for a single opportunity score, then let leadership approve the top three. Anything below the cutoff goes into a parking lot.
This is where most SMB programmes get into trouble. Without a scoring step, teams pick the loudest process, not the highest-value one. According to MIT Sloan Management Review, organisations that tie automation to specific KPIs see roughly three times more measurable value than those that automate based on enthusiasm alone.
Use this scoring grid:
| Dimension | 1 (low) | 3 (medium) | 5 (high) |
|---|---|---|---|
| Volume | Under 20 runs per month | 20–100 runs per month | Over 100 runs per month |
| Value | Under 2 hours saved per week | 2–8 hours saved per week | Over 8 hours saved per week |
| Readiness | Undocumented, inconsistent | Partly documented, mostly consistent | Fully documented, fully rules-based |
A process scoring 75 or higher (e.g. 5 × 5 × 3) is almost always a strong first candidate. Anything under 27 (3 × 3 × 3) is usually not worth the build cost yet — fix documentation or consistency first. Our deeper dive on which business processes to automate first walks through worked examples.
Phase 3: Tool selection and architecture (days 31–45)
The tool selection phase produces a one-page decision memo naming the platform you will build on, the integrations it needs, the high-level data flow, and the 12-month cost. The point is not to pick the best tool in the abstract — it is to pick the one your team will still be using a year from now without specialist help.
Most Australian SMBs choose between three platform tiers, depending on technical comfort and process complexity:
| Tier | Examples | Best for | Indicative cost |
|---|---|---|---|
| No-code orchestration | Zapier, Make, n8n cloud | Marketing and sales ops, simple multi-app workflows | A$30–A$200 per month |
| Process automation suites | Power Automate, Workato, Tray.io | Finance, ops, multi-system workflows with approvals | A$1,000–A$5,000 per month |
| RPA + intelligent automation | UiPath, Automation Anywhere, ABBYY | Document-heavy back-office processes | A$5,000+ per month |
The memo should answer four questions in a page: which platform, which integrations the top three processes need, who will administer it, and what the 12-month total cost is. If you can't answer those four, you are not ready to start building. Our walkthrough of business process automation tools compares the leading platforms by use case.
Phase 4: Build and pilot (days 46–65)
The build phase delivers exactly one production automation for one pilot team — not a portfolio. Pick the highest-scoring process from Phase 2, build it on the platform from Phase 3, and run it live for fourteen days alongside the manual process. The goal is not "look how much we built" — it is "look how reliably this one workflow runs in real conditions".
Three rules we apply to every pilot:
- Shadow run for week one. The automation runs, but a human checks every output. This catches edge cases discovery missed.
- Logging from day one. Every step writes to a log so you can answer "what happened on Tuesday at 3pm" without guessing.
- One named owner. A real operator owns the pilot — not "the team" and not the consultant. If it breaks at 5pm Friday, that person picks up the phone.
According to the Australian Financial Review, Australian SMBs accelerated automation spend through 2026 but consistently under-invest in measurement and ownership. The pilot is where you fix that discipline. For a deeper view of pilot-stage decisions, see our piece on running an AI pilot programme for small business.
Phase 5: Measurement and tuning (days 66–80)
The measurement phase compares the pilot's KPIs against the baseline captured in Discovery. The four metrics that matter for almost every BPA pilot are cycle time, hours saved per week, error or rework rate, and cost per completed task. If the automation hits within ten per cent of its target on all four, move to Phase 6. If not, fix it or kill it.
"The thing that separates a successful automation programme from a failed one is rarely the technology — it is whether somebody is willing to look at the numbers honestly after the pilot and act on what they see." — Industry practitioner quoted in Deloitte's Automation with Intelligence survey
A measurement report at the end of Phase 5 should be one page: baseline number, post-pilot number, per cent change, and a one-sentence interpretation per KPI. If the report runs longer, you are hiding something behind detail. Our framework on business process automation ROI and the broader piece on AI implementation metrics walk through the calculations.
Pro tip
Common mistake: Skipping the baseline. Teams measure the automation in isolation ("it ran 400 times this month, great") without a baseline cycle time or cost per task — so they can never prove it actually improved anything. Capture the baseline in Phase 1.
Phase 6: Handover, governance and scaling (days 81–90)
The handover phase converts the pilot into an operating capability. Three artefacts must exist before Day 90: a runbook (what to do when the workflow breaks), a named owner who has formally accepted the runbook, and a scaling plan that lists the next three processes from your Phase 2 shortlist with their target start dates. If any of the three is missing, you have a pilot, not a programme.
Governance does not need to be heavy. For most Australian SMBs, it is a fortnightly 30-minute review covering: what ran, what broke, what the KPIs say, and which process is next. Two hours a month is the difference between a portfolio that compounds and one that quietly rots over twelve months.
This is where the roadmap connects to the broader programme — the AI Implementation Playbook, our AI Workflow Automation and AI Strategy & Implementation services, sales process automation patterns on Sales Mastery, the marketing ops roadmap on Marketing Edge, and the AI implementation frameworks library on AI Insights.
What does a 90-day BPA roadmap look like in practice?
A typical 90-day roadmap for a 35-person Australian professional services firm starts with seven candidate processes, narrows to three, picks Microsoft Power Automate plus one integration platform, automates client intake as the pilot, hits a 62 per cent cycle-time reduction in Measurement, and ends Day 90 with a named operations owner and the next two processes queued. Here is what each phase produced:
| Phase | Output | Concrete result |
|---|---|---|
| 1. Discovery | 7 processes catalogued | 11 interviews, 28 hours/week of manual work surfaced |
| 2. Prioritisation | Top 3 chosen | Client intake, invoice approvals, onboarding emails |
| 3. Tool selection | Power Automate + Make | A$420/month total platform spend approved |
| 4. Build and pilot | Client intake live | Intake form → CRM → Slack alert, shadowed for 7 days |
| 5. Measurement | Cycle time cut from 4.5 days to 1.7 days | 62 per cent reduction; 11 hours/week saved |
| 6. Handover | Runbook + owner + scaling queue | Ops manager owns it; next 2 processes scheduled |
That is what success looks like at Day 90 — not a vast portfolio, but one workflow running reliably, measured against a real baseline, owned by a real person, with the next two on the queue. This sequenced rollout sits naturally inside our work across professional services firms.
Where should you start this week?
Start this week by booking the Phase 1 interviews — not by demoing tools. Pick five to eight roles across your business, put 30-minute discovery slots in their calendars over the next two weeks, and create a one-tab spreadsheet with the five inventory fields (process, trigger, frequency, systems, owner, hours). That single artefact unlocks every later phase. If your processes are inconsistent, document first and automate second — automating an inconsistent process multiplies the inconsistency. If you would rather have an external team run discovery and prioritisation with you, that is exactly the kind of engagement we do at GrowthGear — sequenced, measured, and handed back as a runbook, not a black box.
Frequently Asked Questions
A business process automation roadmap is a sequenced, time-bound plan — typically 60 to 90 days — that takes an organisation through discovery, prioritisation, tool selection, build, measurement and handover. It defines who owns each automation and how success is measured, so the programme produces an operating capability rather than a one-off project.
For most Australian SMBs, an end-to-end BPA roadmap takes about 90 days from kickoff to a measured, owned automation in production. Single-process roadmaps can compress to 60 days. Anything shorter usually skips prioritisation or measurement and produces a pilot nobody trusts.
A BPA roadmap focuses on automating specific, scoped processes. A digital transformation roadmap is broader — data, customer experience, organisational change, and platforms. BPA is usually a workstream inside digital transformation, not a replacement.
A 90-day BPA roadmap for an Australian SMB typically costs A$15,000 to A$60,000 in external services plus A$400 to A$2,000 per month in platform fees once live, according to common pricing across Deloitte, mid-market consultancies and our own client engagements. ROI is usually visible within the first 90 days through hours saved.
According to Deloitte, the largest single reason BPA programmes stall is starting the build phase before prioritisation has produced an approved shortlist. Teams default to the loudest process rather than the highest-value one, and the pilot never produces a measurable result that justifies scaling.
You do not strictly need one, but most SMBs benefit from a consultant in Discovery and Prioritisation, because internal teams find it hard to interview colleagues neutrally. After Phase 2, an in-house owner can usually run the rest of the roadmap with a clear runbook and a fortnightly governance cadence.
A BPA roadmap is the operational base layer that AI implementation builds on. Once a process is automated, documented and measured, layering AI on top (for example, an LLM extracting fields from documents) becomes a small, low-risk increment rather than a green-field project.
Sources & References
- Deloitte Automation with Intelligence 2024 — "73 per cent of organisations have started intelligent automation, but only around 13 per cent have scaled it across the business" (2024).
- McKinsey Global Institute — Jobs Lost, Jobs Gained — "Up to 45 per cent of work activities could be automated using currently demonstrated technology" (2017, refreshed).
- Gartner — Hyperautomation Software Market Forecast — "Worldwide hyperautomation-enabling software market projected to reach nearly US$600 billion" (2022).
- Australian Bureau of Statistics — Business Use of Information Technology — Australian small and medium business digital adoption baselines (current release).
- MIT Sloan Management Review — Ten Ways to Make Data and Analytics Pay Off — "Organisations that tie analytics and automation initiatives to specific KPIs see roughly 3x more measurable value" (2023).
- Australian Financial Review — Technology coverage — Reporting on accelerating Australian SMB automation spend through 2026 and the persistent measurement gap.
