Migrating Mid-Market Systems Without Breaking Prod

A staged migration roadmap for mid-market teams replacing legacy systems: how to sequence the work so production stays up and cash flow stays sane.

Sanjay Mandal · · 5 min read

Xin

Most mid-market companies do not fail at digital transformation because they chose the wrong platform. They fail somewhere in month four, with two systems writing to the same customer table, an invoice format that no longer matches, and nobody willing to switch the old one off. The people who absorb that risk are usually the finance lead reconciling GST filings and the CTO running the whole thing with three developers and no staging environment.

What should a transformation roadmap sequence first?

A workable roadmap sequences by risk and reversibility, not by which system is oldest or which vendor demoed best. Take every process you want to move and rank it on one question: what happens if this breaks for a full working day? Invoicing and GST filing sit at the top, order capture just below, internal dashboards far lower. Then order the phases so the first ones are read-only and can be reverted in an afternoon. A roadmap that starts with your most dangerous process is not ambitious, it is just unordered. Write the rollback step next to every phase before you write the timeline.

Why do big-bang cutovers break production?

Big-bang cutovers fail because they put every process on the line in the same 48 hours, and mid-market teams rarely have a mirrored staging environment or the people to rehearse it twice. In a 100-400 person company there is typically one person who understands the legacy database, no automated test suite, and no clean way to reproduce production data outside production. So the first real test happens in front of customers. Timing makes it worse: month-end closing, the GST filing dates, and festival sale weeks all sit in the calendar as natural freeze periods. A go-live on the 28th of a month is not a scheduling decision, it is a bet.

How does the strangler pattern actually work?

Run the new system alongside the old one and move traffic over one route at a time, so the legacy application stays a working fallback until the last route has moved. Practically, that means a routing layer in front — a reverse proxy, feature flags, or simply a login that decides which interface serves a given path. Move read paths first: reports, dashboards, order history. Then move write paths individually, starting with the least consequential. Where both systems need the same data, dual-write for a fixed period and run a nightly reconciliation job that diffs totals and emails the difference to whoever owns that process. For anything customer-facing, use shadow mode: the new system generates the same invoice or confirmation but does not send it, and you compare output for two full cycles before anyone sees it.

Which slice do you migrate first?

The best first slice is the one that is read-only, quietly painful, and small enough to reverse in a day. That is almost always reporting. Use these filters when you are choosing it:

  • Read-only: it consumes data and produces no side effects on customers, stock, or tax filings.
  • Fails quietly: if it breaks overnight, someone complains in the morning rather than a buyer receiving a wrong invoice.
  • Already painful: your team is exporting to Excel every Monday, and that spreadsheet is effectively your requirements document.
  • Small surface: one or two screens, under ten users, one clear output.
  • Reversible in a day: you can point the link back at the old report without a change request.

What you are buying in those first two weeks is not the report. It is the deployment pipeline, the database access, the backup habit, and the argument about who owns the schema — all acquired at near-zero risk. If you want a starting baseline of what your current public-facing stack actually does, our site checker will record response times and basic uptime before you change anything.

What does a phased migration look like in numbers?

A typical mid-market sequence runs 12-16 weeks and costs materially less than a rewrite, because you never stop shipping. Take a composite example: a 180-person B2B distributor with order entry and invoicing in a 2011-era ASP.NET application, Tally for accounting, and a full-rewrite quote of roughly 60 lakh and nine months. The staged version looked like this:

  • Weeks 1-2: read-only reporting replica on the new stack, outstanding balances per customer.
  • Weeks 3-4: WhatsApp order confirmations sent from the new service, reading only.
  • Weeks 5-8: new order entry for one sales region, about 30 users, dual-writing to the old database.
  • Weeks 9-11: GST invoice generation in shadow mode for two filing cycles, compared line by line against the old output.
  • Weeks 12-14: invoicing switched for all regions, old application kept read-only for 90 days.

Budget landed in the 18-24 lakh range with no full weekend of downtime. Figures vary by stack and data quality, but the shape holds. You can see similar phased rebuilds in our case studies.

Where to start

Find the report your team still exports to Excel every Monday and make that the first thing the new stack serves. Two weeks of that will teach you more about your data than a forty-page roadmap, and it will not put a single invoice at risk. Maysan Labs builds these staged migrations — web platforms, dashboards, B2B portals and internal tools — and we usually argue for a smaller first phase than the client walks in asking for.

Buying software or a website?

Twelve questions to ask any vendor before you sign: price, your data, fit for an Indian business and support.

Get the free checklist

Related case study

Storefront we build and run ourselves

We needed a storefront we could sell and run, not only build: stock, orders, payments and customers in one place.

View case study

Written by Sanjay Mandal at Maysan Labs, a software studio in Gurgaon, India. Follow us on LinkedIn