Repairs and Renovation Should Not Compete Blindly for the Same Capacity

Jiri Binko

Microsoft

Nov 07, 2026

Reactive repairs and planned renovation often reach the same front door without knowing about each other. This article proposes a six-check coordination decision card, a bounded architecture, and an honest pilot design for deciding when combining work helps tenants and when it must not.

Editorial illustration of a three-story Dutch brick apartment building. On the left, two workers install a new window from a small scaffold; in the center, a technician repairs a front door while the resident looks on; on the right, a coordinator and an older resident study a plan at an entrance. Thin crimson lines connect the three activities, while most homes are untouched and lived in.

Consider an illustrative ground-floor apartment in a 1960s portiekflat, a Dutch walk-up apartment building. In February, the tenant reports a back door that sticks and lets in drafts. A repair contractor adjusts the frame and orders a part; a second visit in March installs it. In April, another contractor surveys the building for an insulation program starting in June, which includes new doors and frames. In June, the renovation crew replaces the door repaired in March.

Nobody acted unreasonably. The repair desk met its response time, the program followed its plan, and the tenant was simply asked to be at home more often than necessary. Reactive repairs and planned renovation need shared visibility and explicit coordination rules. Combining compatible work can reduce avoidable disruption and use scarce capacity better, but only within boundaries set by safety, urgency, skills, tenant preferences, and financing rules.

Two illustrative journeys for the same apartment. Fragmented: in February, a repair visit adjusts the sticking door and orders a part; in March, a second visit installs the part; in April, a separate contractor surveys the building for insulation, new doors, and frames; in June, the renovation replaces the repaired door; throughout, each team contacts the tenant separately. Coordinated: in February, the repair visit makes the door safe and the technician sees the planned scope; a coordination check finds the door in June’s scope, all six checks pass, and the named owner approves waiting; the tenant agrees to the timing, with a fallback repair date held; between April and June, the new door is installed with the planned work, and the repair is closed against that work order; throughout, there is one contact plan for the tenant. — **Figure 1.** Fragmented versus coordinated: two illustrative journeys for one home. Not a measured result; fewer visits are not guaranteed and depend on eligibility.

Why this matters now

Dutch housing associations (woningcorporaties) are renovating at scale while keeping up with daily repairs. Under the Nationale Prestatieafspraken 2025–2035, a sector agreement with the housing minister and municipalities rather than a statute, associations make every effort to remove all E, F, and G energy labels by 2028 at the latest, except monuments and homes designated for demolition, with complex-wide projects subject to agreement by 70% of tenants and any homeowners association. They also committed that from 2026 no homes should be in poor maintenance condition (category 5 or 6) unless concrete demolition or renovation plans exist.

Money and capacity are tight. The sector supervisor, the Autoriteit woningcorporaties, reports in its Staat van de corporatiesector 2026 that associations spent €11.7 billion on maintenance and improvement in 2025, about €5,000 per rental unit and almost 65% more than in 2020, for reasons not yet fully understood, and that the balance between tasks and means has been disrupted again. When visits, scaffolding, and tenant conversations draw on the same budgets and trades, uncoordinated work is a cost that is easy to see and rarely measured.

Not all separate work is waste

Coordination does not mean bundling everything. Some separation is avoidable: two visits where one would do, a repair to a component due for replacement within months, or several teams contacting the same tenant in the same week. Other separation is legitimate: urgent repairs that cannot wait; specialized work that needs particular qualifications; renovation that needs a written, reasonable proposal, presumed reasonable for ten or more homes forming a structural unit when at least 70% of tenants agree (Dutch Civil Code, article 7:220); and repairs and investment funded and approved through different budgets. What is needed is a decision method: can this work be coordinated safely, legitimately, and with the tenant’s agreement?

The coordination eligibility card

The decision card in Figure 2 is this article’s central contribution. All six checks must pass before coordination is proposed:

Compatible scope. The component is inside the planned scope, or the same trade can do both jobs in one visit without changing the design.
Safe timing. Waiting, or bringing work forward, creates no safety risk and keeps the repair within its urgency category. A temporary make-safe repair can still be paired with a planned permanent fix.
Qualified skills and parts. The crew holds the required qualifications and parts; otherwise, the combination merely moves the second visit.
Tenant agreement and access. The tenant agrees to the timing, and access, accessibility, and care needs are known through a flag, not through sensitive case notes.
Authorized work and budget. The work is authorized under the right budget and contract, and no cost is shifted between operating and investment budgets to flatter the combination.
Accountable decision owner. A named role approves the combination, and the decision is recorded with its reasons.

Decision card with six checks: 1 compatible scope; 2 safe timing; 3 qualified skills and parts; 4 tenant agreement and access; 5 authorized work and budget; 6 accountable decision owner. Outcomes: if all six are yes, propose coordination, the named owner approves, the decision is recorded, and one work order is issued; if any is no, do not combine and handle the work separately, with urgent or safety work going ahead now; if any is unsure, human review by the named owner, with separate handling as the default until resolved. — **Figure 2.** Can this work be coordinated? The coordination eligibility decision card, with refusal and human-review paths. Author guidance to adapt locally.

Any “no” means the work is handled separately, and urgent or safety work goes ahead at once. Any “unsure” goes to human review by the named owner, with separate handling as the default. The card is deliberately conservative: it turns coordination into a visible, owned decision instead of an informal favor between planners.

Shared visibility, bounded decisions

Figure 3 shows the architecture in six steps. Repair events, from tenant reports, inspections, and contractor findings, are matched with asset and planned-work context for the same home. Deterministic checks apply the card’s hard rules. Where they pass, analytics compares feasible options. A named planner or program manager approves or rejects each one, with the budget holder signing off on budget changes. Approved work flows to contractors as ordinary work orders, and outcome evidence is recorded against each decision.

Six-step flow: 1 repair events and planned-work context, by integration; 2 eligibility and policy checks, by deterministic rules; 3 feasible coordination options, by analytics with AI assisting; 4 accountable approval by a named planner or program manager, with the budget holder for budget changes; 5 contractor and work-order execution; 6 outcome evidence recorded per decision. Below, authoritative records, each with one owner: repair request and work-order record, asset register, maintenance and renovation program, tenancy record with an access flag, and the contractor’s own planning system. Caseworker notes are kept separate; only a “contact the caseworker before scheduling” flag crosses into coordination. — **Figure 3.** Shared visibility, bounded decisions: the coordination flow, its authoritative records, and the approval step. Conceptual author guidance, not a validated deployment.

Authoritative records keep the flow honest. The repair and work-order record owns each job and its urgency category; the asset register owns components, condition, and labels; the maintenance and renovation program owns scope, budget, and dates; the tenancy record holds contact details and an access flag; and contractors keep their own planning systems. Caseworker notes on personal circumstances stay in case management: coordination sees only a flag such as “contact the caseworker before scheduling.”

Share only what each step needs. Under the GDPR, every use of tenant data needs a specified purpose and a lawful basis, which is not always consent, and data must be minimized. The Dutch data protection authority’s guidance that an association may request income data only when genuinely needed shows the regulator’s expectation of necessity. Coordination must never justify giving contractors more personal data than the job requires.

What technology should, and should not, do

Integration provides visibility: repair events and planned work on one timeline per home.
Deterministic rules enforce boundaries: urgency categories, safety checks, and budget authorization are not left to a model.
Analytics compares options: visits avoided, travel, lead times, and capacity.
AI may extract component details from free-text reports, summarize a home’s history, or suggest candidate combinations with reasons, always for human review.
People keep the designated decisions: approving combinations, talking with tenants, and resolving conflicts.

This keeps the design proportionate under the AI Act. Classifying repair reports is not among the high-risk uses in Annex III; the assessment changes if a system profiles tenants or infers vulnerability: within an Annex III use, profiling is always high-risk, and any such use needs its own legal assessment. If tenants report repairs through an AI chatbot, the Act’s transparency rules, applicable since August 2, 2026, require that people are informed they are interacting with AI. None of this is legal advice.

When not to combine

Recommend against coordination when:

an urgent or safety-critical repair would be delayed;
the work is incompatible in sequence, trade, or method;
asset information is too uncertain to confirm compatible scope;
the right skills or parts are not available for the combined visit;
the tenant objects, or access and accessibility needs conflict with the planned date;
the work or budget is not authorized; or
integration has failed or data is stale. Fall back to separate handling rather than leaving a repair waiting on a plan nobody can see.

A bounded pilot and an honest baseline

Start small: one complex with planned work in the coming year, one or two contractors, and the card as the rulebook. Before switching anything on, measure a baseline over a comparable period: eligible jobs actually coordinated, visits per comparable repair or job, breaches of urgent-repair service levels (SLAs), repeat contacts, first-time fix where already defined, and tenant disruption and satisfaction.

Figure 4 separates three benefits that are often blurred. Service outcomes come first. Cash is only money no longer spent, such as invoices that never arise. Released capacity, meaning contractor and staff hours freed, is valuable but becomes cash only if contracts or budgets change. Count each benefit once. Agree in advance to stop if urgent-repair breaches rise, tenants report delayed repairs, or avoidable visits do not fall, and to adjust if few jobs are eligible or data gaps block the checks.

Three columns of pilot measures. Service outcomes, measured first: eligible jobs actually coordinated, visits per comparable repair or job, urgent-repair SLA breaches that must not rise, repeat contacts, first-time fix where already defined, and tenant disruption and satisfaction. Cash, only money not spent: invoices that no longer arise, duplicate service calls and parts avoided, and shared access or scaffolding costs actually saved, excluding savings counted elsewhere. Released capacity, valuable but not cash: contractor hours and staff and tenant-contact time freed, becoming cash only if contracts or budgets change. Decision gate: expand when service outcomes hold or improve with no urgent-repair harm; adjust when few jobs are eligible or data gaps block the checks; stop when urgent-repair breaches rise, tenants report delays, or avoidable visits do not fall. — **Figure 4.** What the pilot must prove: service outcomes, cash, and released capacity measured separately, with an expand, adjust, or stop gate. No improvement is assumed.

The recommendation

Make coordination an explicit, owned decision for a defined set of eligible jobs: not an aspiration, and not an algorithm’s choice. Start with one complex, one rulebook, and one accountable owner, and expand only when evidence shows fewer avoidable visits without harm to urgent repairs or tenant trust.

Recommend against it where asset data cannot confirm compatible scope, where contracts cannot accommodate changed work orders, where tenants cannot be offered a genuine choice, or where the pilot shows no drop in avoidable visits. In those cases, better communication between separate teams is still worth the effort.