Build an IT Department From Scratch: A Manufacturer’s Playbook
In this article
- Start with an honest assessment, not a shopping list
- The sequence that works: foundation before applications
- Setting up IT infrastructure for a company: decisions that are hard to reverse
- ERP comes after the foundation, not before
- Write the policies while the department is still small
- Building the team: structure for a mid-sized manufacturer
- What the IT manager in a manufacturing company owes the CEO and board
- Mistakes to avoid when you build an IT function from zero
- Key takeaways
- Frequently asked questions
- How long does it take to build an IT department from scratch?
- What should a new IT manager do first in a company with no IT?
- Which IT policies does a manufacturing company need first?
To build an IT department from scratch, work in a fixed order: assess what exists, secure the foundation (identity, network, backups, endpoints), stand up core services, then implement the ERP and its integrations, and only then formalize policies, SLAs and a team structure. Companies that start with the ERP or with hiring usually rebuild the foundation later, at a higher cost and with the business already depending on it.
I joined NAF Company, a feed manufacturer in Al-Qassim, Saudi Arabia, in October 2023 as IT Manager. There was no central infrastructure, no ERP and no IT team. Three years later the plant runs on servers, cloud services, a network and endpoint security that I designed and that have sustained 99.9% uptime since go-live, an SAP Business One core integrated with the MES, Jisr HR and a custom feed formulation system, a written set of governance and security policies, and a multi-disciplinary team with SLAs. I report directly to the CEO and the Board Chairman, so every step below was also a step I had to explain and defend.
Start with an honest assessment, not a shopping list
The first deliverable of a new IT function is a written picture of the current state. Not a proposal, not a quotation: an inventory. Until you know what the company already owns, who holds the passwords and where the data physically lives, every purchase is a guess. The inventory I built when I started covered what I now recommend to anyone starting from zero:
- Devices and ownership: every computer, printer, switch, router, access point and industrial PC, who uses it and whether it is on any contract.
- Accounts and credentials: email, internet lines, domain registrar, cloud subscriptions, and who holds admin access. In companies with no IT function these are usually spread across finance, an external technician and a former employee.
- Data: where production records, formulas, invoices and HR files actually sit (local disks, shared drives, personal mailboxes, paper), and whether anything is backed up.
- Processes that run on manual entry: which numbers are typed twice and who reconciles them. At a feed plant the obvious one was production data being re-keyed for administration.
- Vendors and contracts: internet, telephony, software subscriptions, maintenance agreements, with renewal dates.
- Visible risks: shared passwords, unpatched machines, a single internet line, a server room without cooling or access control.
This assessment is also the baseline you will be measured against: when I present uptime or cost reduction to the board, the comparison point is this document.
The sequence that works: foundation before applications
The most common question I get is “what do I do first?” Here is the order I followed, and why each phase has to come before the next.
- Assessment and quick wins. Inventory as above. Fix anything dangerous immediately: unique admin accounts, a first backup of critical data, shared passwords replaced. Do not buy anything large yet.
- Identity, network and endpoint security. A central directory for users and devices, a network with separation between offices, guests and the production floor, and endpoint protection on every machine. Everything that follows depends on knowing who a user is and which device they are on.
- Servers, cloud services and backups. Decide what runs on-site and what runs in the cloud, set up the server room properly, and put a tested backup and restore routine in place before the first business system is installed. “Tested” means you have restored from it.
- Core business services. Company email on your own domain, shared files with permissions that mirror the organization, printing, and a helpdesk channel so requests stop arriving as phone calls to one person.
- ERP and integrations. Only now select and implement the ERP. Users have accounts, the network can carry the traffic, and the data to migrate has been located and cleaned. At NAF this phase delivered SAP Business One integrated with the MES, Jisr HR and the feed formulation system.
- Governance, policies and SLAs. Write the IT governance, data protection and cybersecurity policies while the function is still small enough that you know every exception. Publish service levels for the helpdesk and critical systems.
- Team structure and continuous improvement. Hire into the gaps you now understand, split responsibilities, and move to Industry 4.0 work: automating workflows that still depend on manual entry.
Phases overlap in practice, but the dependencies do not: the ERP will be only as reliable as the identity, network and backup layers underneath it.
Setting up IT infrastructure for a company: decisions that are hard to reverse
Most infrastructure choices can be revisited. A few are expensive to undo once the business depends on them. These are the ones I spent the most time on.
| Decision | Options | What drives the choice in a factory |
|---|---|---|
| On-site servers vs cloud | All on-site, all cloud, or a mix | Production systems (MES, formulation, SCADA interfaces) must keep working when the internet line does not; collaboration and backup copies benefit from the cloud. I run a mix for exactly that reason. |
| Network design | Flat vs segmented (office, guest, servers, production floor) | Segment from day one. Redesigning a plant network after the MES and PLCs are connected is disruptive and risky. |
| Identity | Local accounts vs a central directory | Central from the start. Joiner and leaver processes, ERP access and security policies all hang off it. |
| Internet connectivity | Single line vs primary plus failover | Once the ERP is live, a single line is a single point of failure for invoicing and dispatch. |
| Backup | Local only vs local plus off-site copy, with restore tests | Off-site copy and scheduled restore tests. This is the control that turns a ransomware incident into a bad day rather than a shutdown. |
The 99.9% uptime we have sustained is not the result of expensive hardware. It comes from decisions of this kind, made deliberately before users depended on the systems, and then from operating discipline: monitoring, patch windows and restore tests that actually happen.
ERP comes after the foundation, not before
The ERP is where the business feels the pain (duplicate entry, slow month-end, stock figures nobody trusts), which pushes new IT managers to start there. Resist it until the foundation phases are done, then move fast. At NAF the ERP phase was an end-to-end SAP Business One implementation, with three integrations that mattered more than any single module: the MES on the production floor, Jisr HR for people and payroll data, and the custom feed formulation system that turns a recipe into production orders. The measurable result is that production data now syncs with administration in real time, and the manual reconciliation between production and administration ended entirely. I describe that project in the SAP Business One manufacturing case study.
Two lessons from the client’s side apply to any company doing this for the first time. Own the requirements: the vendor configures what you specify, and a vague specification produces a system that reflects the vendor’s assumptions, not your plant. And design the integrations before the modules: decide which system is the master for items, batches, employees and cost centers before anyone configures a screen. If you are still choosing a platform, my comparison of SAP Business One vs Odoo for Saudi companies covers licensing, localization and manufacturing depth from the buyer’s side.
Write the policies while the department is still small
Governance feels premature when the department is one person. It is not: the policies you write in year one define what “normal” looks like before habits set in. I authored NAF’s IT governance, data protection and cybersecurity policies with two constraints: alignment with the requirements that apply to companies connected to the Saudi financial market and with current compliance standards, and brevity, so that department heads would actually read them. The minimum set I recommend for a manufacturer:
- Access control and acceptable use: who gets which access, how it is requested, how it is removed when someone leaves.
- Information classification and data protection: what counts as confidential (formulas, pricing, personal data), where it may be stored, how long it is kept.
- Cybersecurity baseline: authentication and multi-factor, patching, endpoint protection, email security, incident reporting.
- Backup and recovery: what is backed up, how often, where the copies are, and the restore test schedule.
- Change management: how changes to the ERP, network and production systems are requested, tested and approved.
- Third-party access: how external support connects, for how long, and who approves it.
In Saudi Arabia, the National Cybersecurity Authority’s Essential Cybersecurity Controls are the natural reference for the cybersecurity baseline even for companies not formally in scope; aligning your policy headings with them saves work if you are assessed later.
Building the team: structure for a mid-sized manufacturer
Hire for the gaps you have proven, not for an org chart you copied. By the time the foundation and ERP phases are done you know exactly where the hours go, and that is what the structure should reflect. The team I built and lead has three disciplines:
| Discipline | Owns | Typical SLA items |
|---|---|---|
| Infrastructure | Servers, cloud services, network, backups, endpoint security, plant connectivity | Availability of core systems, backup success and restore tests, patch windows |
| Applications | ERP configuration, integrations (MES, HR, formulation), reports and dashboards, user permissions | Turnaround on report and permission requests, integration error resolution, month-end support |
| Helpdesk | First-line support, onboarding and offboarding, device lifecycle, user training | Response and resolution times by priority, first-contact resolution |
Two practices made the structure work: SLAs published to the business rather than kept inside IT, and security in everyone’s job description. That security-first culture is cheaper to establish at hire number one than to retrofit at hire number ten.
What the IT manager in a manufacturing company owes the CEO and board
Reporting directly to the CEO and the Board Chairman changed how I ran the function. Executives want to know whether the plant can ship, whether the numbers are trustworthy, whether the company is exposed, and what IT costs relative to what it delivers. So the reporting I built answers exactly that: availability of core systems (the 99.9% figure), roadmap status against plan, security posture in plain language, and operating cost against budget, where proactive vendor renegotiation and consolidation of overlapping services cut our IT operating costs by 15%. The executive dashboards the CEO and Board use are built on SQL queries and Crystal Reports inside SAP Business One, so the system that runs the plant also reports on it.
Mistakes to avoid when you build an IT function from zero
- Buying before assessing. Hardware bought in week one is usually the wrong size, because the requirements were not yet known.
- Starting with the ERP. It is the most visible project and the worst first project; it inherits every weakness of the layer underneath it.
- Leaving the production floor out of scope. Industrial PCs, PLC interfaces and the MES are on your network whether or not they are on your org chart. Segment and inventory them from the start.
- Backups that were never restored. A job that reports success is not evidence of recoverability. Schedule restore tests and record them.
- Vendor-owned credentials. Domain, DNS, cloud tenant and firewall admin access belong to the company. Recover them during the assessment phase, before a dispute makes it difficult.
Key takeaways
- Begin with a written inventory of devices, accounts, data, manual processes, vendors and visible risks; it is both your plan and your baseline.
- Follow the dependency order: identity and network, servers and backups, core services, then ERP and integrations, then policies and team structure.
- Make the hard-to-reverse infrastructure decisions (segmentation, failover, backup strategy, identity) before users depend on the systems; that is where uptime comes from.
- Implement the ERP as an integration project, with production-floor and HR data flowing into it, not as a standalone accounting install.
- Report to leadership in business terms: availability, roadmap progress, security posture and cost against budget.
Frequently asked questions
How long does it take to build an IT department from scratch?
Expect the foundation (assessment, identity, network, servers, backups, core services) to take months rather than weeks, and the ERP phase to be a project of its own on top of that. Policies and team structure mature over the first one to two years. The exact timeline depends on plant size, data quality and how many legacy processes have to be replaced.
What should a new IT manager do first in a company with no IT?
Produce a written inventory of every device, account, data location, vendor contract and manual process, and fix the immediately dangerous items: shared admin passwords, missing backups of critical data, and credentials held outside the company. Do not commit to major purchases until that assessment is complete and leadership has agreed on the sequence of phases.
Which IT policies does a manufacturing company need first?
Access control and acceptable use, information classification and data protection, a cybersecurity baseline (authentication, patching, endpoint and email security, incident reporting), backup and recovery, change management, and third-party access. Keep each short enough for department heads to read, and align the security sections with the National Cybersecurity Authority’s controls.
If you are building an IT function from zero, or inheriting one that was never designed, I can help you run the assessment, sequence the roadmap and manage the ERP and integration phases from the company’s side. See my digital transformation services and ERP implementation services, or get in touch to talk through your starting point.