How I work
Understand the business. Find what’s actually costing money. Fix that, in the right order.
The method is the same whether the answer turns out to be a small process change or a three-month integration project. Here is how it works and why the order matters.
How I work
Diagnose. Prioritise. Design. Implement. Measure.
The same five steps every time, whether the answer turns out to be a two-day fix or a three-month project. Technology comes in at step three, not step one.
Step 1
Diagnose
I start with the business, not the technology. Where is staff time going? Where are leads and money leaking? What does management wish it could see? I talk to the people doing the work, look at the systems and the data, and establish a baseline.
Step 2
Prioritise
Not everything is worth fixing. Each problem gets sized: what it costs you now, what fixing it is likely to be worth, and what it would take. You end up with a ranked roadmap, not a wishlist.
Step 3
Design
The right answer might be automation, an integration, AI, replacing a tool, or fixing the process itself. I design the simplest solution that solves the problem and fits how your team actually works.
Step 4
Implement
I build it, or I manage the developers and vendors who do, and I take responsibility for it working inside your business. You get a working system, not a slide deck.
Step 5
Measure
We agreed a baseline at the start. Now we check whether the intervention did what it was supposed to. If it didn't, we adjust. If it did, you know what it was worth.
Technology enters at step three. Never step one.
The order matters
Business problem first. Technology last.
Most technology projects fail because they start with a tool and go looking for a problem. I work in the opposite direction, and the order is deliberate.
Business problem
What is actually costing you time, money or customers? Stated plainly, in the owner's words.
Process
How does the work flow today, really? Not the documented version. The version that happens on a busy Tuesday.
People
Who does what, what do they work around, and what would they need to actually adopt a change?
Data
Where does the information live, which system should own it, and can it be trusted?
Technology
Only now: what's the simplest tool or change that solves the problem given everything above?
Measurement
Did it work? Against the baseline we agreed, not against a feeling.
No manufactured ROI
I build a business case, then measure whether it held up.
Every recommendation in a roadmap follows the same structure. It keeps me honest, and it means you're never asked to spend money on something that can't be checked afterwards.
Problem
The specific operational issue, not a vague category.
Baseline
What's happening now, measured. If we can't measure it, we say so.
Impact
What the problem is costing, as a defensible range.
Intervention
The recommended fix, and the alternatives considered.
Cost
What the fix will cost to build, run and maintain.
Value
What has to be true for the fix to pay off, and how likely that is.
Measurement
How and when we'll check, and what we'll do if it's not working.
AI and automation, in plain English
Two different tools for two different kinds of work.
A lot of confusion comes from using "AI" to mean any software that does something automatically. They're not the same thing, and the distinction matters when you're deciding what to spend money on.
Conventional automation
For predictable, rule-based work. If the input is structured and the correct action is always the same, automation is cheaper, faster and more reliable than AI.
Good fit
- Payment succeeded, so update the CRM and send the receipt.
- New customer signed, so create the onboarding tasks and the welcome sequence.
- Quote sent, so schedule the follow-up and alert the rep if nothing happens in five days.
- Job completed, so raise the invoice in Xero and notify the customer.
AI
For work that needs interpretation. When the input is messy, human language or documents, and the right action depends on understanding what's in it, AI can do things rules can't.
Good fit
- Read sales call transcripts and pull out objections, next steps and risk signals.
- Classify incoming enquiries and route them to the right person.
- Summarise a customer's full history before a call.
- Extract data from invoices, forms and contracts.
- Help staff find the right answer in your internal documentation.
A person stays in the loop for anything that matters.
AI drafts, suggests, summarises and flags. For consequential decisions, a person reviews and approves. That's how you get the productivity gain without handing a customer relationship, a quote or a compliance decision to a model.
For completeness
Tools I work with regularly
Listed for completeness. The tool is chosen last, and it's chosen to fit your business, not the other way round.
- CRM & sales
HubSpot
Salesforce
Pipedrive and similar
- Automation & integration
n8n
Zapier
Make
APIs and webhooks
- Finance & payments
Xero
Stripe
Payment platform integrations
- Operations
Google Workspace
Microsoft 365
Learning management systems
Job and project platforms
- AI
Claude
ChatGPT
AI-assisted development
Custom AI workflows with human review
- Build
Websites and web applications
Internal tools
Mobile applications
Managing developers and vendors
The next step
A 30-minute conversation about your business. No pitch.
The discovery call is for working out whether an audit makes sense. I'll ask about your business, your systems and where things are getting stuck. You'll get a straight opinion on whether I can help, and what I'd look at first.
Step 1
You tell me what's frustrating you
Slow follow-up, double handling, a CRM nobody trusts, an AI question you can't get a straight answer to. Whatever it is.
Step 2
I ask the questions I'd ask on day one of an audit
Which systems, who does what, where the information lives, what management can and can't see.
Step 3
You get a straight opinion
Whether an audit is worth doing, what I'd focus on, and roughly what it would involve. If I'm not the right person, I'll say so.