Fixed-scope vs retainer: How to think about AI implementation spend

The conversation about AI implementation budget usually goes one of two ways. Either a business owner wants a retainer — ongoing access to someone who will keep iterating and improving — or they want a fixed project with a defined deliverable. Both are legitimate models. They solve different problems and carry different risks.

The retainer model: where it works and where it does not

A retainer makes sense when the work is genuinely ongoing — when your business needs continuous development, regular optimisation, or ongoing strategic input. Marketing agencies, for example, work on retainer because the work never stops: campaigns end, new ones begin, results need analysis, strategies need adjusting.

In AI implementation, retainers are often sold before the problem is properly defined. A consultant proposes a monthly fee for access to their expertise. What gets done each month is vaguely defined. The client pays for availability rather than outcomes. Six months in, it is hard to say what the retainer delivered in measurable terms.

The retainer trap

Retainers work when the scope of work is clear and recurring. They fail when they are sold as 'ongoing AI support' before the specific integrations and their costs are known. You end up paying for effort, not results.

The fixed-scope model: why we use it

Fixed-scope engagements work when the problem is well-defined, the deliverable is specific, and the success criteria are measurable. This is exactly what a diagnostic enables. Once we know what the integration gaps are, what they cost, and which ones to fix first, we can scope the build with a fixed price and a defined timeline.

The advantage is accountability. You know what you are getting before you commit. The price does not change unless the scope changes — and scope changes require your explicit approval. There are no surprise invoices at the end of the month.

How to measure ROI before you commit

The ROI calculation for a workflow integration is straightforward. You need three numbers: hours saved per week by the automation, loaded cost per hour of the team member doing the manual work, and the cost of the integration.

The formula

Hours saved per week × weeks per year × loaded hourly cost = annual saving. Divide the integration cost by the annual saving to get payback period. A workflow that saves a paralegal 4 hours a week at R250/hour saves R52,000 per year. An integration costing R18,000 pays back in roughly 4.5 months.

This calculation is only possible if you have a baseline. That is why the diagnostic comes first — it gives you the numbers needed to calculate ROI before any build begins. If a consultant is proposing a retainer before running a diagnostic, they cannot tell you the return. They are asking you to buy their time, not their results.

The Maru engagement model

We use fixed-scope engagements for all implementation work. The diagnostic is fixed at R4,500 to R5,000 and offsets against the build cost if you proceed. The core engagement — Phases 2 through 4, covering design, build, and the first measurement cycle — is priced at R35,000 fixed. If a site remediation is required, that is scoped separately after the diagnostic, typically R8,000 to R15,000.

After the build is complete and measured, we offer an optional Phase 5 Optimise sprint — a fixed 60-day engagement to improve the workflows based on 30 days of live data. This is the right time for ongoing engagement, not before the integration is built and proven.

Diagnostic: R4,500–R5,000 (offsets against the core engagement)
Core engagement (Phases 2–4): R35,000 fixed
Site remediation: R8,000–R15,000 (scoped after diagnostic)
Optimise sprint (Phase 5): R15,000–R25,000, fixed 60-day

Every engagement includes Phase 4 measurement built in. You see exactly what changed — hours saved, error rate reduction, lead-to-client conversion impact. If we cannot measure it, we have not delivered it.