Cover photo: Rodeo Project Management Software / Unsplash. Illustrative photograph, not a record of a client engagement.

A fractional CTO’s first 90 days should leave the founder able to make better technical decisions, with evidence that the team can act on them. Expect an agreed remit, an honest baseline, decisions on the largest risks, a small delivery demonstration and a funded next-quarter plan.

A new diagram is useful only if it helps with one of those jobs.

The sequence below is a suggested working method. It is not an official 90-day standard or a promise to fix a business in a quarter. A live incident, missing access or a two-day-a-month engagement will change the pace. Agree what can be proved with the time you are actually buying.

Before day one: agree authority as well as availability

Write down why the engagement exists. “Help with technology” leaves too much room for a successful adviser and a disappointed founder. “Decide whether the current platform can support the next customer commitment” is a question someone can answer.

Name the founder or sponsor who can approve spending and changes in priority. Specify which decisions the CTO can make, which need approval, and who owns implementation. Advisory time does not automatically include engineering capacity, supplier management or incident cover.

Record the working rhythm: available days, response expectations, the escalation route outside those days and the people needed at decision meetings. A UK–UAE team should agree actual overlap hours and decision deadlines. An unanswered question at the end of one working day can consume the beginning of another.

Give the CTO authorised access to the product, delivery backlog, architecture, incident history, commercial commitments and relevant costs. Prefer named, limited access. Put a missing-access owner and date beside each gap; do not quietly consume the first month waiting for invitations.

Days 1–30: establish what is true

Follow one customer journey through the product and one change through the delivery process. Speak to the people doing support as well as those writing code. Ask what fails, what needs manual intervention and which promises have already been made to customers.

Build a short baseline with four parts:

  • Product and commitments: critical journeys, upcoming obligations and known customer pain.
  • Delivery: recent changes, waiting points, release responsibilities and a representative failed change.
  • Operational control: service owners, recovery evidence, access gaps and supplier dependencies.
  • Economics: current recurring costs, near-term commitments and major unknowns. Separate currencies rather than adding GBP and AED together.

Label each finding observed, reported or unverified. “Backups are configured” and “we restored a usable copy” are different statements. The baseline should expose that difference without turning the month into an exhaustive audit.

The first review ends with a small decision list: what needs attention now, what can wait, and what evidence would change that judgement. The founder accepts the priorities, including what the team will stop doing.

Days 31–60: resolve a constraint and demonstrate the change

Choose a constraint that matters to the business and is small enough to test. If releases depend on one unavailable specialist, a useful proof may be a second engineer completing a release from maintained instructions. If customer onboarding repeatedly fails, trace and improve one failure path before proposing a replacement platform.

Document the alternatives, the decision owner, the reason for the choice and the remaining risk. Run the work through the normal review and release controls. Do not bypass them to produce a more attractive demonstration.

Illustrative example: a team cannot release while its lead developer is away. The month-one baseline records the missing instructions and access. Month two produces an updated runbook and an observed release by another authorised engineer. That supports a narrow conclusion: the tested release no longer depended on that one person. It does not prove that every incident or deployment can be handled independently.

The GOV.UK guidance on measuring and reporting progress connects reporting to decisions and uses existing delivery evidence. Applied here, that means showing the release, the test or the resolved decision rather than commissioning a separate reporting industry.

Days 61–90: make the next quarter executable

Compare the evidence with the initial remit. Keep unresolved risks visible, especially those blocked by a commercial decision or missing capability. A quarter can end with a justified decision not to rebuild.

Translate the findings into a short technology roadmap. Each committed item needs an outcome, delivery owner, capacity, dependency and next decision date. Distinguish funded work from options. If the delivery team has no capacity after support, say so before adding another initiative.

Decide what leadership the business needs next. Continue the fractional arrangement, narrow it, bring capability in-house or end it with a usable handover. The recommendation should follow the work, not assume the engagement renews itself.

Use a deliverables scorecard, not an activity total

The downloadable scorecard contains acceptance questions, suggested evidence, blank ownership fields and one clearly labelled example. Replace the suggested timing with the agreed engagement schedule. Use proposed, in progress, evidenced, accepted or blocked as statuses; “accepted” needs the named reviewer and date.

Do not average the rows into a reassuring percentage. An accepted roadmap does not compensate for an unresolved production-access dependency.

At each review, ask: what decision changed, what evidence supports it, and what still needs the founder? If a deliverable is blocked, record who can unblock it. If it no longer matters, record the decision to remove it.

The useful handover is a business that knows what happens next and who can make it happen.

Download the editable worksheet (CSV) ↓

Practical questions

What if the CTO only works a few days a month?

Reduce the number of commitments and agree dates against actual availability. Keep urgent operational cover separate. A three-month calendar does not create three months of full-time capacity.

Should the first month include a full technical audit?

Only if that is the agreed need and the engagement has enough time and expertise. Start with the evidence needed for the founder’s priority decisions, and commission deeper security, legal or specialist reviews where the findings justify them.

What should the founder bring to the first review?

The current customer commitments, spending constraints and decisions that cannot be delegated. Be ready to stop or defer work as well as approve it. Otherwise the CTO can produce recommendations without anyone able to act on them.

Put this to work

Sources checked 8 November 2026. The examples and worksheet are suggested working aids, not client results, legal advice or guaranteed outcomes.