IT project manager: application projects that are scoped, tested and put live without unpleasant surprises for the business.
The IT project manager runs application projects from requirements gathering to go-live: an ERP roll-out, a new business tool, an interface between two pieces of software, a process redesign. Much of the time goes on minutes, test books, defect-tracking files and data migrations prepared by hand. Automation gives that time back for what makes a project succeed: understanding the business and holding to decisions.
Between business users, IT, software vendors and providers.
The IT project manager turns a business need into a solution that works: scoping, writing or commissioning the specifications, coordinating the vendor or integrator, organising acceptance testing and preparing the cut-over. The role reports to the steering committee and bridges users who talk about processes and technical teams who talk about configuration and interfaces. After go-live, the project manager hands over to support and application maintenance, with documentation that has to hold up.
A typical day.
- MorningReview open acceptance-test defects and chase the vendor on expected fixes.
- MorningRun a workshop with users to describe a process and its special cases.
- MiddayWrite up the workshop minutes and update the functional specifications.
- AfternoonPrepare a data migration batch and check its quality with the business.
- Every weekUpdate the schedule, budget consumed and risks, and prepare the review with IT.
- Every monthPresent progress to the steering committee and have scope decisions approved.
What the role is measured on.
- Adherence to schedule and milestones
- Adherence to budget and effort
- Open defects before go-live
- Quality of migrated data
- User adoption of the tool
- Incidents after go-live
Indicators are named, never quantified: levels depend on each company, and the diagnostic measures them.
The problems that come up most.
Poorly expressed needs
Users describe how they work today, not what they need. Special cases emerge during testing, when they are most expensive to handle.
Acceptance testing run in spreadsheets
Scenarios, results, defects and screenshots circulate in files and e-mails. Nobody can say for certain what has been tested.
Underestimated data migration
Data in the old tool is incomplete, duplicated or coded differently. Cleansing happens late, by hand, at the pace of load trials.
Project reporting that takes time
Schedule, budget, risks and defects are consolidated by hand before each committee, from tools that do not talk to each other.
Forgotten regression tests
A change fixes one point and breaks another. Without replayable tests, each release means rechecking everything or taking the risk.
Documentation that is out of date
Decisions made in workshops, configuration rules and interfaces are not recorded as the project goes. Application maintenance inherits a tool it has to rediscover.
One role, several realities.
- Fashion and luxury
The IT project manager rolls out tools shared between boutiques, e-commerce, wholesale and workshops, often across several countries and languages. The product master and customer data concentrate the migration and quality challenges.
- Retail
Projects often touch the POS, e-commerce, WMS or ERP, with store-by-store roll-outs. Peak trading periods impose short cut-over windows and prepared rollbacks.
- Cosmetics and beauty
The IT project manager runs projects that span R&D, regulatory affairs, the plant and marketing: formula management, product data sheets, exchanges with retailers. Compliance data demands rigorous acceptance testing.
- B2B distribution and wholesale
Projects cover the ERP, WMS, EDI and pricing terms. Every interface with a customer or supplier must be tested with them before cut-over, or orders may be blocked.
- Construction and property
The IT project manager equips branches and sites: quotes, job tracking, mobile field tools, accounting. Adoption by busy, scattered teams is as big a challenge as the technology.
- Industry
Projects connect the ERP, production tracking, maintenance and quality, sometimes with the machines. Cut-over is planned with production, and migrating bills of materials and routings needs particular care.
- Pharma and healthcare
Any system covered by good practice follows a validation approach: specifications, risk analysis, qualification, documented testing, change control. The IT project manager produces and maintains these deliverables with quality assurance.
Automate the task, keep the judgement.
What gets automated
- Workshop minutes and first drafts of specifications written by an AI agent from recordings and notes, reviewed by the project manager.
- Acceptance-test campaigns tracked in a single tool, with scenarios linked to requirements, replayable regression tests and traced defects.
- Quality checks on migration data run at every load trial, with the gap report sent to data owners.
- Project reporting — schedule, effort, budget, risks, defects — fed automatically by the tracking tools for the steering committee.
- Functional and technical documentation kept up to date from specifications, configuration and recorded decisions.
What stays human
- Scoping and arbitration of requests with the business and the steering committee.
- The go-live decision, based on test results and remaining risks.
- Supporting users and managing the relationship with the vendor or integrator.
How to automate, case by case.
Each case starts from a question leadership teams ask. The starting point, what gets automated, the tools involved, and what stays in the teams’ hands.
How do you automate acceptance testing and regression tests for a business application?
- Today
- Acceptance testing is run in a shared spreadsheet: scenarios, results and defects, with screenshots sent by e-mail. With every new vendor release, users repeat the same tests by hand, or skip them for lack of time.
- What gets automated
- Scenarios are described in a test management tool and linked to requirements. Critical journeys and interfaces are automated as replayable tests, run on every release with a gap report. Defects are created directly from the failed test, with the steps and data used.
- What stays human
- Key users approve the scenarios and sign off acceptance, and the project manager decides on defect severity.
- test management tool
- test automation tool
- ticketing tool (ITSM)
How do you automate data migration when changing ERP?
- Today
- Items, customers, suppliers, open balances and history must move from the old ERP to the new one. Extracts are reworked in spreadsheets, code mappings are kept by hand, and every load trial reveals new errors.
- What gets automated
- The migration is built as a replayable flow: extraction from the old system, written transformation rules and mapping tables, quality checks, loading into the new ERP. Each trial produces a rejection report by object and by owner. An AI agent helps match duplicates and propose code mappings, approved by the business.
- What stays human
- Data owners approve each migration trial and decide which data is not migrated.
- old and new ERP
- data integration tool (ETL)
- data quality tool
- spreadsheet
How do you write functional specifications faster from business workshops?
- Today
- The project manager runs the workshops, takes notes, then spends evenings tidying them up and turning them into specifications. Points discussed get lost between the workshop and the document.
- What gets automated
- With participants’ consent, workshops are recorded and transcribed. An AI agent produces the minutes, the list of decisions and open questions, then a first draft of the specifications structured by process, with the business rules and special cases mentioned. It flags contradictions with specifications already approved.
- What stays human
- The project manager reviews, completes and has the specifications approved by the business.
- video-conferencing tool
- AI agent
- document management
How do you automate IT project reporting for the steering committee?
- Today
- Before each committee, the project manager gathers the schedule, effort consumed, the provider’s invoices, open defects and risks, each in a different tool. The same charts are rebuilt every time.
- What gets automated
- Project indicators are fed automatically by the project management tool, the ticketing tool, time tracking and the ERP for invoices. The committee pack is generated with progress by milestone, remaining work, budget consumption and the trend in defects. An AI agent proposes a summary of points of attention.
- What stays human
- The project manager chooses the messages and the decisions to request from the committee.
- project management tool
- ticketing tool (ITSM)
- ERP
- reporting tool (BI)
How do you automate defect tracking during acceptance testing and after go-live?
- Today
- Defects arrive by e-mail, in meetings or by phone, are copied into a file and passed to the vendor in its own tool. Statuses diverge between the two, and users chase without knowing where their request stands.
- What gets automated
- A single queue receives defects through a form that requires the useful information: process, steps, data, severity. An AI agent detects duplicates and proposes a qualification. Tickets are synchronised with the vendor’s tool, statuses update on both sides and users are notified of every change.
- What stays human
- The project manager decides on defect severity and priority with the business and the vendor.
- ticketing tool (ITSM)
- vendor’s tool
- middleware and APIs
- messaging
How do you make a go-live and its cut-over plan reliable?
- Today
- The cut-over is prepared in a document listing the steps, owners and times. On the day, checks are done by phone, and a forgotten step or skipped check is discovered after restart.
- What gets automated
- The cut-over plan is described step by step in a shared tool, with its dependencies, owners and pass criteria. Checks that can be automated — volumes loaded, interfaces active, access opened, first flows through — run and report their result. The rollback decision point is defined in advance, with its criteria.
- What stays human
- The cut-over committee decides to proceed or roll back at the planned decision point.
- project management tool
- monitoring tool
- test management tool
- messaging
How do you automate tracking of an IT project’s budget and effort?
- Today
- Internal effort is reported irregularly, days billed by the integrator arrive with the invoices and change orders are added as the project goes. The project manager discovers the overrun when asking for extra budget.
- What gets automated
- Internal time, ERP orders and invoices and change orders are attached automatically to the project and its work packages. A board compares actual with planned, calculates the remaining work declared per package and projects the cost at completion. Variances beyond a set threshold trigger an alert to the project manager and the CIO.
- What stays human
- The project manager re-estimates the remaining work, and the steering committee decides on scope or budget trade-offs.
- time tracking tool
- ERP
- project management tool
- reporting tool (BI)
How do you automate functional and technical documentation for a tool after delivery?
- Today
- At the end of the project, documentation is written in a hurry and is often incomplete. Configuration rules, interfaces and workshop decisions are not gathered anywhere, and application maintenance rediscovers them incident after incident.
- What gets automated
- Documentation is built during the project: approved specifications, exported configuration, interface descriptions from the integration layer, recorded decisions. An AI agent assembles these into a guide structured by process and updates it with every approved change. Users and application maintenance can query it in natural language.
- What stays human
- The project manager approves the documentation handed over, and application maintenance becomes responsible for it after the transfer.
- document management
- AI agent
- middleware and APIs
- ticketing tool (ITSM)
How do you speed up choosing business software: requirements and comparison of responses?
- Today
- The requirements document is written from generic templates, vendors’ responses arrive in different formats, and the comparison is done in a spreadsheet filled in by hand. Demonstrations show what each vendor chooses to show.
- What gets automated
- Requirements are drawn from the described processes and weighted with the business. Vendors answer in a common grid, and an AI agent reads their responses and documents to flag requirements covered as standard, by configuration, by development or not at all. Demonstrations follow set scenarios drawn from the company’s real cases.
- What stays human
- The selection committee, with the business and IT, chooses the solution.
- document management
- AI agent
- spreadsheet
- reporting tool (BI)
How do you automate user training and help when a new tool starts?
- Today
- Training is given just before start-up, on generic cases. In the first weeks, the same questions keep coming back to the project manager and key users, who answer them one by one.
- What gets automated
- Training materials are produced from the acceptance-test scenarios and real cases, by user profile. An AI agent answers questions in messaging from the approved documentation and materials, and passes to a key user what it cannot handle. The questions received are analysed to correct the documentation or the tool.
- What stays human
- Business key users support teams on the ground and approve the agent’s answers.
- AI agent
- document management
- messaging
- ticketing tool (ITSM)
What teams look to improve.
- How do you gather user requirements better?
- By starting from real processes rather than desired screens: what events trigger the work, what data comes in, what decisions are made, what cases fall outside the rule. Users’ current files and workarounds say a lot about what the tool will need to cover.
- How do you organise effective acceptance testing?
- By linking each scenario to a requirement, preparing representative test data including special cases, and tracking defects in a single tool with their severity. Scenarios that can be automated become replayable tests for later releases.
- How do you make a data migration succeed?
- By starting early: data inventory, mapping rules, cleansing at the source, then several controlled load trials. Data owners approve each trial, and the final migration replays a procedure that has already been proven.
- Can AI write the specifications?
- It can produce a first draft from workshop minutes, described processes and existing documents, and flag ambiguous or contradictory points. Approving the rules and decisions remains with the project manager and the business.
- How do you decide on a go-live?
- On criteria set in advance: blocking defects closed, critical scenarios passed, migration approved, users trained, rollback prepared. The steering committee decides on these criteria, not just on the planned date.
- How do you avoid regressions with each new release?
- With a base of automated regression tests on critical processes and interfaces, run on every release delivered by the vendor or application maintenance. Manual tests remain for what changes.
- How do you prepare the hand-over from project to application maintenance?
- By delivering up-to-date documentation, replayable tests, the list of known defects and the configuration rules, then organising a support period after go-live. Decisions made during the project must be easy to find.
- How do you get teams to adopt a new tool?
- By involving key users from scoping, training on real cases from their daily work and providing help that is available when needed. The questions received after start-up show what to improve first.
Describe what’s holding your teams back.
A first conversation to qualify your need. Then a scoped diagnostic: the scope, the solution, the estimate.
Discuss your projectReply within 48 business hours.