Playbook

How to modernize Lotus Notes applications

Learn a step by step playbook for modernizing Lotus Notes applications without a risky big bang rewrite, from inventory to cutover and rollback.

By the Cloudverg engineering team · · 6 min read

The problem

Lotus Notes and Domino applications tend to outlive everyone who built them. A workflow database gets created by one person in 2004, gets copied into six variants, and keeps running because the business depends on it. The original developer is gone. The design is locked in an NSF file. Nobody wants to touch it, so it keeps growing.

The usual reaction is to call for a full rewrite. That is where projects die. A big bang rewrite means you have to understand every rule in the old system before you ship anything, and you have to match years of accumulated quirks on day one. The team underestimates the hidden logic: computed fields, agent schedules, form validation, reader and author fields that control who sees what. The rewrite slips, costs climb, and the old system stays live anyway.

The safer path is incremental. You move pieces out of Notes in a defined order, each piece verifiable on its own, with a rollback point behind every step. That means you need a real inventory first, not a guess. You need to know which databases matter, who uses them, what talks to them, and which business rules are buried in code and in people's heads.

Notes adds its own wrinkles. Security is often enforced by ACLs, reader fields and view-level show-when formulas rather than by a clean permission model. Data lives in documents with rich text and attachments that do not map neatly to tables. Mail-in databases, scheduled agents and LotusScript libraries create side effects that are easy to miss. Inventory has to cover all of it.

What a good solution looks like

A good modernization plan treats each Notes application as a candidate for one of four moves: rehost, refactor, replace or retire. Rehost means you keep the logic and move the runtime, for example running the same application on a supported Domino version or a hosted platform. Refactor means you rebuild the parts that hurt while keeping the data model. Replace means you buy or build something that does the job better. Retire means you shut it down and migrate the few records people actually need.

Not every database deserves the same treatment. Some are used daily by a whole department. Others get opened twice a year to pull a report. Sorting applications into these buckets early stops you from spending months on something you should just archive. The decision should be based on usage, business criticality, integration surface and how much unique logic is locked inside.

The target architecture matters too. Moving off Notes usually means a relational database, a web front end and an identity provider. Document-style data becomes tables or JSON. Rich text becomes a controlled editor format. Agents become scheduled jobs or event handlers. ACLs and reader fields become role-based access control. Getting these mappings agreed before you write code prevents rework later.

The plan should also sequence the work. Start with the easiest, lowest-risk application to prove the approach and the tooling. Then move to harder ones with the lessons learned. Keep the old system readable during transition so people can check results. Every step should have a defined rollback: the old database stays in read-only mode until the new one has run clean for an agreed period.

How to build it

Step one is inventory. List every NSF database and mail-in database on every server. For each, record the owner, the business function, the user count, the last modified date, the integrations (who reads or writes it), and any scheduled agents. Pull design elements: forms, views, agents, script libraries, outlines and framesets. This is mechanical work, and it is where most surprises surface.

Step two is recovering business rules. Read the LotusScript and formula code, then sit with the people who use the application. Ask what the fields mean, which statuses matter, what happens when someone submits a bad record, and which reports get printed. Write those rules down in plain language. The code tells you what the system does; the users tell you what it is supposed to do. Both matter.

Step three is ranking risk. For each application, estimate the cost of an outage and the likelihood of one. A database with a failing agent and no owner is a different risk from a stable read-only archive. Rank the list and use it to set the order of work. The highest-risk items go first, not the easiest ones.

Step four is choosing the migration pattern. For each application, pick rehost, refactor, replace or retire, with a rough cost and effort for each option. Write a step by step sequence with rollback points. For a refactor, that might mean exporting documents to the new store, building the new screens, running both systems in parallel, then cutting users over in groups. For a retire, it means exporting the records people need, archiving the NSF, and removing access on a set date.

Step five is execution in small slices. Migrate data first and verify counts, attachments and field mappings. Then build the new workflow, test it against real historical records, and cut over one user group at a time. Keep the Notes database available read-only for a defined window so people can compare. Only then decommission.

Risks and how to manage them

The biggest risk is missing hidden logic. Reader and author fields, hide-when formulas, computed subforms and agent schedules often carry rules nobody documented. Mitigate this by testing the new system against a sample of real documents, including edge cases like returned mail, cancelled requests and records with missing fields. If the output differs, find out why before you cut over.

Identity and access is the next trap. Notes ACLs and reader fields do not map one to one onto role-based access in a web application. Map every role explicitly, and test with real user accounts, not admin accounts. Watch for shared IDs, group documents and nested groups. A user who could see a document in Notes must either see it in the new system or be told why not.

Data fidelity and attachments are a quiet source of pain. Rich text with embedded images, OLE objects and file attachments may not survive a straight export. Test the conversion on a sample and check that attachments open. Decide early whether old attachments are migrated or left in an archive.

Finally, plan for the old system to stay around longer than you think. Keep the servers patched and monitored during the transition. Set a real decommission date and communicate it. Without a date, read-only mode drifts into permanent use and you end up maintaining both systems.

How to measure success

Measure by application, not by project mood. Track how many databases are fully migrated, how many are in read-only transition, and how many remain untouched. A shrinking untouched list is the clearest signal.

Track data quality after migration: record counts reconciled against source, attachment counts, and the number of failed conversions. Track user adoption: logins to the new system, support tickets mentioning the old one, and the number of people still requesting Notes access. Falling Notes access requests means the cutover is holding.

Track operational cost and risk. Count the servers, agents and licenses still needed for Notes, and watch that number drop with each wave. Keep a list of open rollback points and close them as each application proves stable.

If you want help getting the inventory and the sequence right before committing to code, the Cloudverg Legacy modernization sprint is a fixed scope, three week engagement at $8,500/month, and it produces exactly that: an inventory, recovered business rules, a ranked risk list, options with costs, a migration sequence with rollback points and a fixed quote for the first phase. It is one option among others. You can also run the inventory yourself with a small team and a spreadsheet, and that works fine if you have the time.

Frequently asked questions

Can we modernize Lotus Notes without rewriting everything?

Yes. Treat each database as a separate decision and move it in planned steps. Most applications can be rehosted, refactored, replaced or retired on their own schedule, with the old system kept read-only until the new one proves stable.

How long does a Notes migration take?

It depends on how many databases you have and how much hidden logic sits inside them. A single small workflow can move in weeks. A portfolio with integrations and custom agents takes months, which is why sequencing by risk matters more than a single end date.

What should we do first?

Build a complete inventory of every NSF database, its owner, users and integrations. Then recover business rules from code and from users, and rank the risk. That list tells you what to migrate first and which applications you can simply retire.

Monthly plan

Want help with this? Legacy modernization sprint

$8,500/month

3 weeks · Starts within 10 business days

Three weeks to understand an old system nobody wants to touch, rank the risks and agree a step by step plan that avoids a big bang rewrite.

Renews monthly. Cancel any time.
What is included

Related guides

More custom software guides   Custom software

Smallest way to start

Get a senior architect on your problem this week

Ninety minutes a month on the cloud, AI or security question that is blocking you, a pre-read of your material and a written summary within two business days.

$450/month

Renews monthly. Cancel any time.