Why downtime during insurance data migration matters in 2026
The COO I worked with at a US P&C carrier last spring - $1.2B GWP, 6 states, mostly personal auto and homeowners - opened our review meeting with one sentence: "My CEO does not care about the migration. He cares about whether the agents can quote on Monday morning. Tell me how we get to Monday morning without an 800-number flooded with calls." That carrier was 11 weeks from cutover. The migration plan was technically solid. The downtime plan - what stays running, for whom, on which system, for how long - was three pages of bullets that would not survive contact with a real Monday.
Minimizing downtime during insurance data migration is, in my experience, the dimension that most directly translates into customer outcome. Security failures produce findings 18 months later. Regulatory failures produce findings at the next examination. Downtime failures produce angry agents and lost claims on Monday morning. The CEO who tolerated a slipped budget will not tolerate an agent base that cannot log in.
A decade ago, "downtime" meant a clean handover over a weekend. That still works for small carriers, but for mid-tier P&C carriers between $500M and $5B GWP acceptable downtime has tightened to hours per domain, not days per system.
Three forces have raised the stakes:
Customer expectations are set by digital self-service. Deloitte's 2026 Global Insurance Outlook, citing J.D. Power's 2024 US claims digital experience study, ties rising auto and home claims satisfaction to the services carriers now offer through mobile apps and websites. Customers who got used to that do not distinguish between "we are migrating systems" and "the system is broken."
Distributed agent and adjuster workforces. Agents and adjusters spread across several states and time zones, working from offices, home, and the field, expect the platform up when they need it. A maintenance window that aligned to one workforce no longer aligns to a distributed one.
CAT season operational pressure. The Atlantic hurricane season officially runs from June 1 to November 30, according to NOAA's National Hurricane Center, and property writers carry wildfire exposure on top of it. Deloitte's outlook notes that the frequency and severity of natural catastrophes, including US wildfires, continue to escalate. Claim volumes can jump sharply within days of a landfall, and any residual fragility from the migration window magnifies under that load.
The margin for absorbing a bad cutover is also thinner: the same Deloitte outlook expects the US P&C combined ratio to worsen from 97.2% in 2024 to about 99% in 2026.
This guide is the playbook I would walk through with the COO in the opening anecdote. It is written for CIOs, COOs, and lead architects at US P&C carriers between $500M and $5B GWP. For the broader picture, the insurance data migration challenges guide is the pillar read; this article focuses on the downtime minimization workstream specifically.
Minimizing downtime during insurance data migration - direct answer
Minimizing downtime during insurance data migration means designing the cutover so each business domain stays within an agreed recovery time (RTO) and data-loss tolerance (RPO). For mid-tier P&C carriers, typical targets are under 2 hours for claims FNOL, 4 hours for policy quote-to-bind, 8 hours for billing and 24 hours for regulatory reporting, met through parallel run, change data capture (CDC) and a rehearsed war room.
The cutover weekend itself is one piece of a broader downtime minimization architecture; the architecture starts at readiness, not at H-72. The short version of the downtime minimization gate:
- Define RTO and RPO targets per insurance domain (not per system).
- Plan the parallel run window before deciding the cutover window.
- Use real-time Change Data Capture during parallel run to keep both systems in sync.
- Pre-write the war room playbook with named staffing for the cutover weekend.
- Communicate downtime windows to agents and customers 90 days in advance.
- Honest disclosure: some downtime cannot be eliminated; the practice is to minimize, sequence, and communicate.
The next sections expand each into a working playbook. The insurance data migration best practices article covers the wider migration framework around this work; this article goes deeper on the downtime dimension.
RTO and RPO targets per insurance domain
The two terms come from contingency planning. NIST Special Publication 800-34 Rev. 1 defines the recovery time objective as the time a system can stay in recovery before the business mission is harmed, and the recovery point objective as the point in time to which data must be restored after an outage. A migration cutover is a planned outage, so the same discipline applies.
The single most underestimated step in minimizing downtime during insurance data migration is, in my experience, setting RTO and RPO targets per domain - not per system. A single carrier system supports multiple business domains with different tolerance levels. The same policy admin system serves quote-to-bind, endorsements, and renewals, and renewals have almost no tolerance during the renewal window. Setting one RTO for "the policy admin system" produces an average that satisfies nobody.
Target RTO/RPO by insurance domain
These are the starting targets I use in workshops, not industry benchmarks. A specific carrier's targets should be set in the stakeholder alignment workshop covered in how to prepare for insurance data migration. The point of the table: targets vary by domain, by 12x or more in some cases, and the architecture has to accommodate the differences. The claims rows deserve the most scrutiny, because a missed FNOL or payment window is visible to the policyholder at the worst possible moment of the claims lifecycle.
How RTO/RPO targets shape the architecture
The most important consequence of per-domain targets is that they force the architecture to support different cutover patterns for different domains. Claims FNOL with a 2-hour RTO cannot tolerate the same cutover pattern as billing with an 8-hour RTO. The migration program has to either:
- Cut over all domains simultaneously and accept the tightest RTO as the constraint, or
- Cut over domains in waves, with parallel run windows that match each domain's tolerance, or
- Use real-time CDC to keep both systems in sync, allowing per-domain cutover with a near-zero gap.
In my experience, mid-tier P&C carriers between $500M and $5B GWP are increasingly choosing option 2 or option 3, because option 1 forces over-engineering on low-tolerance domains where it adds cost without value.
The 4 parallel run patterns - 30, 60, 90, 180 day windows
Parallel run - operating both legacy and target systems concurrently after technical cutover - is the most consistent way of minimizing downtime during insurance data migration at the domain level. The question is not whether to run in parallel; for mid-tier P&C carriers, the answer is almost always yes. The question is for how long, with what data flow, and what the exit criterion is.
Pattern 1: 30-day "validation parallel"
Both systems run for 30 days. The target system is the source of truth. Legacy is read-only for validation and exception scenarios. Data flows one way (target to legacy, for record only). Best for: small carriers with a simple data shape and high confidence in cutover.
Tradeoffs: lower cost (one month of dual operations), but little time to surface latent issues.
Pattern 2: 60-day "operational parallel"
Both systems run for 60 days. The target is the source of truth from day 1. Legacy continues to receive new transactions for the first 30 days (write-back), then goes read-only for the second 30 days. Best for: mid-tier carriers with moderate data complexity and standard cutover risk.
Tradeoffs: 60 days is enough to surface most latent issues. Cost is moderate. The write-back to legacy in the first 30 days adds engineering complexity but allows a fast rollback if needed.
Pattern 3: 90-day "rollback-capable parallel"
Both systems run for 90 days. The target is the source of truth, but full bidirectional sync (real-time CDC) is maintained for the first 60 days, allowing a complete rollback to legacy with no data loss up to day 60. From day 60 to day 90, sync becomes one-way, target to legacy. Best for: regulated carriers, large mid-tier carriers, carriers with low operational risk tolerance.
Tradeoffs: 90 days clears most regulatory and operational concerns. Cost is high (three months of dual operations plus engineering for bidirectional CDC). The rollback option through day 60 is real; I have worked with carriers that exercised it.
Pattern 4: 180-day "regulatory parallel"
Both systems run for 180 days. The architecture is similar to Pattern 3 but extended to cover a full quarter-end cycle and at least one regulatory filing window. Best for: carriers crossing major regulatory reporting cycles, carriers in heavily examined environments, carriers with multi-jurisdictional exposure.
Tradeoffs: 180 days is expensive. The benefit is that the parallel run can absorb a full NAIC Schedule P filing cycle, a state DOI examination window, and, for international groups, a Solvency II reporting window. The tooling questions this pattern raises (audit trail, access control, evidence for examiners) are covered in what to look for in a migration tool for regulated insurance environments.
Which pattern fits which carrier
In my experience advising mid-tier P&C carriers in the $500M-$5B GWP range, the rough split across the programs I have seen is: 10% Pattern 1, 30% Pattern 2, 45% Pattern 3, 15% Pattern 4. That is my own sample, not a market statistic. Pattern 3 is the default for the band; deviation toward Pattern 1 happens when the carrier is small and operationally simple, deviation toward Pattern 4 when regulatory exposure is heavy.
Real-time Change Data Capture for near-zero downtime
Change Data Capture (CDC) is the architectural pattern that makes near-zero downtime possible during parallel run. Every change to the source system (legacy pre-cutover, target post-cutover) is captured as it happens and replicated to the other system. AWS documentation for its Database Migration Service describes the mechanics plainly: CDC reads changes from the source database's transaction logs, and replication latency can stretch to several minutes or longer depending on load and configuration. So the honest promise is a gap of seconds to minutes, not zero. That is still enough to make the RPO targets in Section 3 achievable for most domains.
What CDC handles well for insurance
CDC works cleanly for:
- Policy header changes (new policies, endorsements, cancellations).
- Claim status updates (new claims, status changes, payments).
- Party / customer data updates (address changes, contact updates).
- Financial transactions (premium payments, claim payments, commissions).
- Reference data changes (rate tables, product configurations).
For these transaction types, a well-tuned CDC pipeline can keep both systems within the RPO targets above during parallel run.
Where CDC requires careful engineering for insurance
CDC requires extra engineering for:
- Schema changes during parallel run. If the target schema evolves during a 60-180 day parallel run, the CDC pipeline has to handle schema versioning. This is non-trivial.
- Conflict resolution. During bidirectional sync (Patterns 2 and 3), the same record can be updated on both systems in the same minute. The conflict resolution rule has to be pre-defined and signed off by the business.
- Regulatory reporting. Some regulatory reports (NAIC Schedule P, Schedule F) require point-in-time consistency. CDC provides eventual consistency, not point-in-time. The reporting workstream has to handle the gap.
- Long-running transactions. Claim adjudication can span hours or days. CDC has to capture intermediate states, not just final states, to support rollback.
CDC tooling for insurance migrations
CDC is available in several forms: database-native replication, CDC connectors in integration platforms, and change capture built into migration tooling. On our projects, Decerto's Data Migrator handles mapping, validation and migration reporting, and the CDC layer is designed per program around the carrier's databases. The category-level options are covered in the insurance data migration tools guide. I recommend one selection question above all others for downtime work: does the CDC option support the conflict resolution semantics your business needs, and can it handle schema versioning during the parallel run window?
CDC pipelines also carry production data, so encryption in transit and a per-record audit trail apply to them in full. The data security during insurance data migration article covers those controls.
Blue-green deployment for insurance PAS
Blue-green deployment runs two production environments side by side and routes traffic to one at a time. Martin Fowler's description of the pattern makes two points that matter for insurers: the cutover becomes a router switch, so rollback is a switch back, and the hard part is usually the database, which is why schema changes should be deployed separately from the application change. Done well, blue-green produces minutes of downtime instead of hours.
How blue-green applies to insurance PAS
For mid-tier P&C carriers, blue-green deployment of the policy administration system follows this pattern:
- Both legacy and target PAS run with current data, kept in sync via real-time CDC (Section 5).
- Agent portal, customer portal, and API gateway are configured to route to one PAS at a time.
- The cutover is a configuration change at the gateway - traffic switches from legacy to target.
- For 5-15 minutes around the switch, the gateway buffers writes to handle in-flight transactions.
The customer-facing impact is a brief blip - acceptable for most insurance use cases.
Where blue-green is harder for insurance
Three specifics make blue-green harder in insurance than in some other industries:
- Long-running transactions. A claim being adjudicated when the switch happens has to be preserved on the new side. The CDC pipeline must handle the in-flight state.
- Regulatory reporting timing. A regulatory report being generated when the switch happens must complete deterministically.
- Reinsurance integration. Carriers with treaty-driven cessions have point-in-time reporting requirements that the switch has to respect.
In my experience, blue-green works cleanly for carriers that have already standardized on a stateless API gateway and that have CDC working in production. For carriers that have not - which is most carriers running a legacy on-premise PAS - blue-green is aspirational rather than achievable for the current migration. The architectural reason is covered in comparing monolithic and modular architectures in insurance software.
When to choose blue-green vs traditional cutover
Choose blue-green when the target PAS sits behind an API gateway, CDC already runs in production, customer-facing tolerance is under 30 minutes, and the team has done blue-green at scale. Choose a traditional cutover (a 4-40 hour window) when either system is monolithic, CDC is not production-ready, and the COO accepts a 4-40 hour window. For mid-tier P&C carriers, traditional cutover is still the more common pattern; blue-green is the realistic target for the next migration cycle.
Customer-facing impact mitigation - agents, adjusters, customers
Customer-facing impact mitigation covers what happens to the people who depend on the data - agents, claim adjusters, and customers. In my experience, this is the workstream of minimizing downtime during insurance data migration that is most often under-resourced relative to its actual importance.
What stops, what continues, what degrades
A clean cutover requires the carrier to publish, in writing, what is stopping, what is continuing, and what is degrading during the cutover window. Business continuity cuts across IT, operations, customer service, and compliance, which is why it is one of the harder items in the pillar guide to migration challenges. The table below is the format I recommend for capturing those decisions in one place.
The table should be carrier-specific, with named owners, named workarounds, and named recovery times. Without this written artefact, the carrier discovers what is broken at 9 a.m. Monday by phone calls coming in.
Communication to agents 90 days out
Agents are the front line, and the agent portal is usually the first thing they notice is missing. The communication runway should begin 90 days before cutover:
- T-90 days. Initial announcement, planned cutover window, expected impact.
- T-60 days. Detailed change document with screenshots of the new system.
- T-30 days. Training begins. 4-8 hours per agent depending on role.
- T-14 days. Final cutover confirmation, helpdesk hours, escalation contact.
- T-7 days. Reminder, war room phone number for issues.
I've worked with carriers on both ends of this. The ones that communicate well see far fewer "where do I quote?" calls on the Monday after cutover. Agents who know what to expect tolerate brief downtime; agents who do not, escalate.
Communication to customers
Customer communication is more nuanced. Most P&C customers do not need to know there is a migration. They need to know:
- If they cannot log in for 8 hours, when service resumes.
- If their claim is in process, that it remains in process.
- If they have an emergency, which phone number to call.
The communication should be visible (banner on the customer portal, IVR message, email to customers with claims in process) but not alarming. "We are upgrading our platform" framing is safer than "the system is down."
War room operations during cutover weekend
The war room is the operational nerve center during the cutover weekend. In my experience, the carriers that handle cutover well are not the ones with the best technical plan. They are the ones whose war room operations are pre-written, pre-staffed, and rehearsed before the actual weekend. That rehearsal is where minimizing downtime during insurance data migration stops being a slide and becomes a staffing plan.
War room staffing model
For a mid-tier P&C carrier cutover weekend (32-40 hours), the staffing model I recommend:
- Migration program director. Full-time presence for the entire window.
- Lead enterprise architect. On-call for the full window, on-site for the first 12 and the last 8 hours.
- Data architect. Full-time during the data load and reconciliation phases (H+0 to H+28).
- CIO. Available within 30 minutes by phone for the full window; on-site for go/no-go decisions.
- COO. Available within 30 minutes for the full window; on-site for operational decisions.
- CFO. Available within 30 minutes for reconciliation review (H+22 to H+28).
- Compliance director. Available within 30 minutes for the full window.
- CISO or delegate. Available within 30 minutes; on-call for any security incident.
- Vendor lead architect (Decerto or equivalent). On-site for the full window.
- Helpdesk lead. Full-time from H+24 to monitor post-cutover incoming calls.
- Agent operations lead. Available from H+24, on-site Monday morning.
That is 11 named roles. For smaller carriers, some roles consolidate. For larger carriers, each role has a backup.
Communication channels
Recovery communication is not an afterthought in recognized frameworks: the Recover function of the NIST Cybersecurity Framework 2.0 covers both restoring operations and coordinating communication with internal and external parties during restoration. For the cutover weekend, pre-establish:
- Primary war room channel. Slack, Teams, or equivalent. All war room participants present.
- Out-of-band channel. Signal or a phone tree, in case the primary fails.
- Vendor incident channel. Direct line to the vendor's senior engineer.
- Regulator notification template. Pre-written, ready to send if a cybersecurity event is confirmed during cutover. The NAIC Insurance Data Security Model Law (#668) sets a 72-hour window to notify the commissioner, and some states that adopted it use different deadlines. For New York-licensed carriers, the amended 23 NYCRR Part 500 also requires 72-hour notice to DFS (section 500.17) and written business continuity and disaster recovery plans (section 500.16), which the cutover plan should reference.
The hourly checkpoint
I recommend a 30-minute checkpoint cadence during the active cutover (H+0 to H+40), narrowing to 15 minutes during reconciliation (H+16 to H+28). Each checkpoint has a written disposition: green (proceed), yellow (proceed with watch), red (pause or roll back). Decision authority is documented in advance - typically the named CIO or delegate. The decision is not made by committee at 3 a.m.
The war room executes a cutover timeline that is planned earlier in the program; the best practices guide for insurance data migration covers the planning and testing steps that feed it.
6 honest patterns when downtime cannot be eliminated
I will state this directly: some downtime cannot be eliminated. Anyone who tells you that minimizing downtime during insurance data migration can always reach zero is selling something. The honest practice is to acknowledge this, minimize what can be minimized, and communicate what cannot. Below are six patterns where downtime is structural, not optional.
Pattern 1: Schema migration that fundamentally changes data shape
When the target system has a meaningfully different data model from legacy (different primary keys, relationship structures, normalization), there is a window during the schema transformation where neither system is consistent. The window is short - typically 30-90 minutes in the programs I have run - but it is real.
Pattern 2: Regulatory reporting that requires point-in-time consistency
Regulatory reports run on point-in-time data. During the cutover window, point-in-time consistency is broken. The reporting workstream has to wait for the cutover to complete - typically 8-24 hours of reporting downtime even with otherwise near-zero downtime architecture.
Pattern 3: Reconciliation that requires read freezes
CFO reconciliation requires source and target to be read-frozen simultaneously for the reconciliation snapshot. The freeze window is typically 30-60 minutes. It is unavoidable for any reconciliation method that produces a defensible variance number.
Pattern 4: Long-running batch processes
Nightly batches (premium, commission, regulatory report generation) can take hours. During cutover the batch either runs on legacy and is replayed on target, or runs on target with legacy input. Either way, there is a gap.
Pattern 5: Integration with third-party systems
Carriers integrate with reinsurance brokers, payment processors, regulatory portals, and other third parties, and each integration has its own cutover. Reinsurance flows are often built on the ACORD Global Reinsurance & Large Commercial (GRLC) standards for placing, accounting and claims messages, which helps, because the message format stays stable while the endpoint moves. Coordinating all partners to the same window is still rarely possible; the practical answer is to phase the integration cutovers, with brief gaps for each. The integration side is covered in connecting legacy systems through insurance software integration.
Pattern 6: Month-end and quarter-end overlap
If cutover falls near a month-end or quarter-end, the financial close cycle adds operational downtime that no migration architecture can eliminate. The right answer is to plan cutover to avoid month-end, quarter-end and year-end - covered in the readiness phase. It is also one of the calendar items teams most often forget, as described in the hidden traps of migration planning.
The lesson: the goal is not zero downtime in absolute terms. The goal is per-domain downtime that meets the RTO/RPO targets, with the residual downtime communicated honestly and contained to non-critical windows.
Decerto Generali Group Poland - single-weekend cutover with no data loss
The reference deployment that most directly illustrates minimizing downtime during insurance data migration in practice is the Generali Group Poland data migration delivered by Decerto. It is a European life insurance program rather than a US P&C one, and I say that up front; the downtime mechanics transfer, the regulatory specifics do not.
The downtime constraint
After Generali Group Poland acquired another insurer, all products, processes, and customer data had to move to Generali's architecture. The case study describes the acquired company's data as poor quality, needing significant validation and correction, and the migration had to fit a short, predefined window with no room for data loss or downtime in customer services.
The downtime minimization practices
Three practices from this article are visible in how the program ran:
Single-weekend cutover. The migration itself took place over a single weekend, a timeline the case study describes as crucial to minimizing disruption to Generali's operations. Compressing the cutover into one weekend is the war room practice from Section 8 in production - it limits the customer-impact window and the sustained-effort window.
Custom migration tool with built-in validation. The tool Decerto built imported, validated, corrected, and transformed legacy data before exporting it to the target and generated migration reports for verification. Data that lands clean does not force the system to be paused later to fix it.
Dozens of trial migrations. Decerto and Generali ran multiple rounds of testing, including dozens of trial migrations and detailed technical and business testing, before go-live. Trial runs surface the downtime gaps in advance, so they can be closed before the real cutover.
The published outcomes
From the Generali case study: the balance of financial transactions remained intact, no data loss occurred, and data consistency and accuracy improved through validation and correction. A single-weekend cutover with intact financial balances is the practical proof point that the approach works when applied with discipline.
The same approach scales to US mid-tier P&C carriers between $500M and $5B GWP. The Migration Architecture Review (Section 12) is where we translate the downtime framework to your specific RTO/RPO targets and parallel run constraints.
FAQ
Thearchitecture combines RTO/RPO targets per domain (Section 3), parallel run for30-180 days (Section 4), real-time Change Data Capture (Section 5), optionallyblue-green deployment (Section 6), customer-facing impact mitigation (Section7), and disciplined war room operations (Section 8). The goal is per-domaindowntime that meets carrier-specific targets, with residual downtimecommunicated honestly.
RTO(recovery time objective) is how long a function can be unavailable; RPO(recovery point objective) is how much data, measured in time, can be lost. Both terms are defined in NIST SP 800-34. Typicaltargets for mid-tier P&C: claims FNOL under 2 hours RTO and 15 minutes RPO,policy quote-to-bind under 4 hours RTO and 1 hour RPO, billing under 8 hoursRTO and 2 hours RPO.
Parallelrun is the practice of running both legacy and target systems concurrentlyafter technical cutover, with the target as the source of truth. Four patterns:30-day validation parallel, 60-day operational parallel, 90-dayrollback-capable parallel, and 180-day regulatory parallel. The pattern ischosen based on carrier complexity, regulatory exposure, and risk tolerance.
Near-zerodowntime (minutes, not hours) is achievable for some domains using blue-greendeployment with real-time Change Data Capture. True zero downtime is rarelyachievable in insurance because of structural patterns: schema migration,regulatory reporting consistency, reconciliation freezes, batch processes, andthird-party integration cutovers. The honest practice is to minimize, sequence,and communicate residual downtime.
Change DataCapture (CDC) is the pattern where every change to the source system is readfrom its transaction log and replicated to the other system, typically withinseconds to minutes. For insurance migrations, CDC enables parallel run with aminimal data gap, supports per-domain cutover with controlled downtime, andprovides the rollback capability needed for high-risk programs.
The fourstandard patterns are 30, 60, 90, and 180 days. For mid-tier P&C carriersbetween $500M and $5B GWP, the most common choice in my experience is the90-day rollback-capable pattern. The 180-day regulatory parallel is used bycarriers crossing major regulatory cycles or multi-jurisdictional environments.The 30-day pattern is reserved for small, low-complexity migrations.
Set an RPOper domain, keep systems in sync with CDC during parallel run, and close thecutover with a read-frozen reconciliation snapshot that the CFO signs off.Rehearse it: on the Generali Group Poland program, dozens of trial migrationspreceded a single-weekend cutover in which no data was lost and financialbalances stayed intact.
For amid-tier P&C cutover weekend (32-40 hours), the war room includes 11 namedroles: migration program director, lead architect, data architect, CIO, COO,CFO, compliance director, CISO, vendor lead, helpdesk lead, and agentoperations lead. Each role has a defined presence pattern (full-time on-site,available within 30 minutes, on-call). Pre-established communication channelsand a 30-minute checkpoint cadence are mandatory.
Agentcommunication starts at T-90 days with the planned cutover window, then adetailed change document at T-60, training at T-30, final confirmation at T-14,and a reminder at T-7. Customer communication is more nuanced: visible but notalarming, focused on when service resumes and which phone number to call in anemergency. "We are upgrading" framing is safer than "the systemis down."
Talk to Decerto about minimizing migration downtime
If you read one section of this article on minimizing downtime during insurance data migration, I would point you here.
Motivation. Downtime that lands on agents and customers on Monday morning compounds. The CEO who tolerated a slipped budget will not tolerate an agent base that cannot log in or a call volume that overwhelms the helpdesk. The architecture that minimizes downtime has to be designed at readiness, not at H-72. Carriers that designed it in are passing cutover weekends with hours of downtime per domain. Carriers that retrofitted it are managing days.
Value. A free 30-minute Migration Architecture Review with me (Janusz Januszkiewicz). Vendor-neutral. I bring 15+ years of insurance migration experience plus a senior Decerto architect. You bring your current state architecture and your operational tolerance per domain. We leave with preliminary RTO/RPO targets for your portfolio, a parallel run pattern recommendation, an honest assessment of whether blue-green is viable for your architecture, and a list of the highest-risk downtime gaps.
Skepticism, addressed honestly. Decerto's Data Migrator and our migration services are built for mid-tier P&C carriers between $500M and $5B GWP that want measurable downtime targets, parallel run architecture, and rollback options. For sub-$250M GWP single-line carriers, a simpler architecture is often the right call. For $5B+ enterprise carriers, an enterprise platform such as Guidewire with a large systems integrator is usually the better fit, and we say so on the first call.
Climax. The downtime practices in this article are the ones we applied on the Generali Group Poland single-weekend cutover. Our wider references include the Warta (HDI/Talanx Group) eAgent platform for 40,000+ agents and the BNP Paribas Cardif claims platform built on Higson. The RTO/RPO table and the four parallel run patterns are the artefacts we bring to the first call.
Sources
- National Institute of Standards and Technology. (2010). Special Publication 800-34 Rev. 1: Contingency Planning Guide for Federal Information Systems. Swanson, M., Bowen, P., Phillips, A. W., Gallup, D., & Lynes, D.
- National Institute of Standards and Technology. (2024). The NIST Cybersecurity Framework (CSF) 2.0. NIST Cybersecurity White Paper (CSWP) 29.
- National Association of Insurance Commissioners. (2017). Insurance Data Security Model Law (Model #668).
- New York State Department of Financial Services. (2023). Second Amendment to 23 NYCRR Part 500: Cybersecurity Requirements for Financial Services Companies (adopted November 1, 2023).
- NOAA National Hurricane Center. Tropical Cyclone Climatology: Atlantic and Eastern Pacific hurricane season normal activity. Accessed October 2026.
- Deloitte Center for Financial Services. (2025). 2026 Global Insurance Outlook: Changing customer expectations, broker consolidation, and the importance of modernization could reshape the insurance landscape.
- Fowler, M. (2010, updated 2015). Blue Green Deployment. martinfowler.com (bliki).
- Amazon Web Services. AWS Database Migration Service User Guide: Creating tasks for ongoing replication using AWS DMS. Accessed October 2026.
- ACORD. Global Reinsurance & Large Commercial (GRLC) Data Standards. Accessed October 2026.
- Decerto. (2024). Successful Data Migration for Generali Group Poland (case study).
.avif)


.avif)
.avif)

