Real-Time KPIs for Small Construction Firms: Peru 2026

Real-time KPIs for small construction firms in Peru 2026 cover 8 critical dimensions: physical progress vs. schedule, actual costs vs. budget (Cost Performance Index), schedule performance (Schedule Performance Index), crew productivity, warehouse materials vs. next-2-week requirements, open supervision observations, monthly billing issued and collected, and projected cash flow for the next 30 days. Typical stack: mobile app for on-site data capture, Supabase as a Data Lake, integration with SUNAT via OSE for progress billings, and an executive dashboard in Metabase or custom React. Operational cost between $200 and $500/month. Custom development between $12,000 and $25,000 as a one-time fee. For a firm running more than 3 simultaneous projects, a proprietary system pays for itself in under 18 months compared to recurring SaaS subscriptions.

If you run a small construction firm in Lima, Arequipa, Trujillo, or Cusco and your team still assembles reports manually every Monday, this guide gives you the winning pattern.

The 8 Essential KPIs

Canonical Stack

Step 1: On-Site Capture with a Mobile App

The rule: if the data doesn't get entered on-site, on-site, it never reaches the dashboard on time. A mobile app for the site superintendent or supervisor:

• Geotagged photo of progress at the start and end of each day
• Man-hour and machine-hour capture by work item
• Materials receiving with QR scan of the delivery slip
• Observations by work item with photo and severity level
• Electronic signature from site superintendent and supervisor

Typical stack: Capacitor + Lovable + Supabase. Every entry goes directly to Bronze with a timestamp and geolocation.

Step 2: Calculate KPIs in Gold

Postgres or Supabase materializes views:

create materialized view gold.constructora_dashboard as
select
  obra_id,
  partida_id,
  sum(metros_avanzados) as avance_fisico,
  sum(horas_hombre) as horas_hombre_totales,
  sum(metros_avanzados) / nullif(sum(horas_hombre), 0) as productividad,
  sum(costo_real) as costo_real,
  sum(valor_ganado) as valor_ganado,
  sum(valor_ganado) / nullif(sum(costo_real), 0) as cpi
from silver.partidas_avance
group by 1, 2;

refresh materialized view gold.constructora_dashboard;

Refresh every hour. For large projects, use a change-triggered refresh.

Step 3: Integrate with SUNAT via OSE

When you issue a monthly progress billing to your client, the electronic invoice is stamped via a SUNAT-authorized OSE (Nubefact, Sumico, Efact, or others). The flow:

1. System builds the invoice with the billing amount and work item descriptions
2. API call to the OSE
3. OSE stamps it and returns the CDR (Comprobante de Recepción)
4. Invoice is valid with SUNAT
5. System stores the CDR in the Data Lake
6. Dashboard cross-references billing issued vs. collected vs. physical progress

You immediately identify:

• Clients with overdue billings unpaid for more than 30 days
• Progress billings approved on-site but not yet issued to SUNAT
• Discrepancies between amounts collected and amounts stamped

Step 4: Executive Dashboard with Drill-Down

Metabase or custom React. Main view with 8 KPIs as cards. Drill down to:

• List of work items with CPI below 0.95 (over budget)
• List of work items with SPI below 0.95 (behind schedule)
• Crews with productivity below threshold
• Materials that will run out within the next 2 weeks
• Critical open observations

The Real Case: 28 KPIs with Zero Hallucinations

A multi-country distributor with 100 franchises needed advanced reporting. Catalizadora delivered:

• 28 KPIs hardcoded in JavaScript with browser-side calculation (zero server CPU)
• AI narrative only for text, not for KPI calculation
• Two-level pattern: KPI headline + AI paragraph
• Immutable audit trail with SHA-256 hash chain
• Verifiable audit.verify_chain_integrity function
• 5 reporting sections (Financials, Sales, Services, Complaints, System Usage)
• Investment included in COT-0003 at $26,000

This is what we call guardrails: KPIs in code, no hallucinations. Narrative generated on top of verified data. Auditable and defensible.

For a small construction firm in Peru, the pattern is identical: KPIs in SQL or TypeScript, narrative with Claude or GPT on top of already-calculated data. The LLM never calculates CPI or SPI — it only describes what the code already calculated.

Comparison: SaaS vs. Custom System

What NOT to Do

1. Daily Excel that arrives on Monday: you lose 6 days to react
2. No on-site capture: Thursday's data gets reported on Monday (already late)
3. No SUNAT integration: invoicing and project operations live in parallel systems
4. International SaaS without SUNAT: you pay the subscription and still do invoicing by hand
5. Dashboard without drill-down: the CEO sees a red number but can't investigate

Next Steps

If your small construction firm in Peru manages 2 to 10 simultaneous projects and still assembles reports manually every Monday, a real-time dashboard with a custom stack pays off fast.

Catalizadora runs the diagnostic in a 30-minute call — no pitch deck, a real conversation about your operation.

• MAGIA / Core builds custom systems with a mobile app, Data Lake, SUNAT integration, executive dashboards, and per-project modules in 12 weeks for $15,000. Code and data in your name, no retainers.
• For enterprise cases with an AI engine and day-one observability, MAGIA / Forge delivers in 12 weeks for $20,000.