Technology Leadership Seat · USD 7,000 / month
90-day CTO mandate: Merrivale Retail Systems
Victor Chen owns the technical roadmap. The rewrite question is explicitly not his to decide alone.
Merrivale has nineteen engineers, a platform built in 2016 and an argument about rewriting it that has run for two years without resolution. A fractional CTO could take that decision in week three. This mandate says he will not, and sets a process to reach it by week ten instead.
From 6 April, Victor owns engineering structure, the technical roadmap, tooling and architecture decisions below a stated threshold. The platform rewrite sits above that threshold. It gets a decision paper in week ten and a decision by the CEO and board in week eleven, which is nine weeks sooner than the current trajectory and with one person accountable for the paper.
Document control
- Document number
- MAN-MRV-01
- Version
- 1.0
- Effective
- 6 April 2027
- Owner
- Victor Chen, Fractional CTO
- Approved by
- Helen Ashworth, CEO
- Next review
- 6 July 2027
- Classification
- Confidential. Executive and engineering leads
Why
What the diagnostic found
| Observed | Evidence |
|---|---|
| No technical leader since March 2025 | Two team leads splitting the role alongside delivery |
| The rewrite question is unresolved and live | Raised in 11 of the last 24 engineering meetings |
| Engineers are split roughly evenly on it | 9 for, 7 against, 3 no strong view, from the interviews |
| No decision record exists for any past architecture choice | Confirmed. Nothing written since 2019 |
| Deployment takes 4 hours and is manual | Measured, not reported |
| Delivery is not actually slow | 14 releases last quarter. This is a healthy team |
The last two rows sit together deliberately. The team ships well despite a four hour manual deploy, which means the deploy is a cost they are absorbing rather than a crisis. That distinction matters when deciding what to fix first.
Authority
Decision rights
| Victor decides | CEO decides | Board decides | |
|---|---|---|---|
| Engineering team structure and leads | Yes | ||
| Technical roadmap and sequencing | Yes | ||
| Tooling, CI, deployment and developer experience | Yes | ||
| Architecture decisions under 6 engineer-months | Yes, with a written record | ||
| Hiring within the approved plan | Yes | ||
| Vendor and infrastructure spend to USD 20,000 | Yes | ||
| Architecture decisions over 6 engineer-months | Yes, on a decision paper | ||
| The platform rewrite | Yes | Informed | |
| Anything that changes the product roadmap | Yes, with the CPO | ||
| Dismissing an engineer | Yes, with People | ||
| Security risk acceptance | Board |
Six engineer-months is the threshold and it is deliberately low. It means roughly one engineer for half a year, which is the point at which an architecture choice stops being a technical decision and becomes a commitment of company capacity.
The rewrite
Why this is not delegated, and how it gets decided
A platform rewrite at this company is between 18 and 30 engineer-months on any honest estimate. That is not a technical decision, it is most of a year of the company's capacity, and the person who holds the seat for 90 days should not be the person who commits it.
-
No position taken
Victor forms no public view on the rewrite in the first month. Saying so on day one is what makes the rest of the process credible.
-
Evidence gathered
Actual cost of the current platform: incident time, onboarding time, change failure rate, the features genuinely blocked by it. Measured, not asserted.
-
Both cases written by their advocates
The strongest engineer for and the strongest against each write their case. Victor edits neither.
-
Decision paper to the CEO
Options, costs, risks, what is reversible, and Victor's recommendation stated plainly as a recommendation.
-
CEO decides, board informed
Whatever the answer, it gets recorded as a decision with a date and a reason.
-
The decision is communicated to all 19
Including to the side that lost, by the CEO, with the reasoning.
Plan
What else the 90 days does
| Week | Change | Why now |
|---|---|---|
| 1 | Architecture decision records start | Every decision from here is written. Cheap, and it is why nothing from 2019 is explainable |
| 2 | Both team leads get written authority | They have been doing the job without permission for a year |
| 3 to 6 | Deployment automated | Four hours to under 20 minutes. Independent of the rewrite question, which is the point |
| 4 | On-call rota formalised | It exists informally and rests on two people |
| 6 | Change failure rate measured | Needed as evidence for clause 3, useful permanently |
| 8 | Hiring plan agreed with the CEO | Two engineers, roles shaped by the rewrite decision, hired after it |
Automating the deploy is done regardless of the rewrite outcome. Work that is valuable under every option is the work to do while a big decision is still open.
Measures
How this gets judged on 6 July
| Measure | Today | What good looks like |
|---|---|---|
| The rewrite question | Open for 2 years | Decided and recorded |
| Architecture decision records | None since 2019 | Every decision since week 1 |
| Deployment time | 4 hours, manual | Under 20 minutes, automated |
| Team leads with written authority | 0 of 2 | 2 of 2 |
| On-call | Informal, two people | Formal rota |
| Release frequency | 14 last quarter | Not a target. Do not let it fall |
Document control
Revision history
| Version | Date | Author | Change |
|---|---|---|---|
| 1.0 | 26 March 2027 | Victor Chen | Agreed and signed. Effective 6 April 2027. |
| 0.3 | 21 March 2027 | Victor Chen | Draft. Architecture threshold lowered from 12 to 6 engineer-months after the CEO and I compared it against last year's decisions. |
| 0.2 | 18 March 2027 | Victor Chen | Draft. Clause 3 process added. The first draft delegated the rewrite decision and that was wrong. |
| 0.1 | 14 March 2027 | Victor Chen | First draft from the engineering diagnostic of 9 March. |