Ten things I do, in four families

Mail sits on a server nobody updates, files live in a share that only works from the office, and anyone working from home makes do with a USB stick.

• Migration to Microsoft 365, Google Workspace or AWS, chosen on what you already use rather than on what suits me
• Mailboxes, calendars, contacts and historical archives moved without losing permissions and rules
• Cutover run outside working hours or over a weekend, with the old mail readable for as long as needed
• SPF, DKIM and DMARC configured so your mail actually arrives
• Half a day with the people who will use it, the day after the switch

The server is out of warranty, the business application runs on a Windows version that stopped getting updates, and nobody wants to be the one who touches it.

• Honest sizing: how much machine you actually need, not how much can be sold
• Windows Server and Linux, installed and put into production
• Active Directory: domains, group policies, accounts and permissions put back in order
• Virtualisation with Hyper-V or VMware, storage, and consolidation of several physical machines onto one
• Documentation of how it is built, so whoever works on it next understands it straight away

There is no wi-fi in the warehouse, the line stutters at peak times, and nobody can say which cable goes where.

• Routers, switches and access points chosen for how many people you are and what you do
• Network segmentation: guests, production and administration kept apart
• VPN and secure remote access for people working off site
• Wi-fi coverage verified on site, not estimated from a floor plan
• An up-to-date network diagram, handed over to you

You have three quotes that say different things, nobody to explain them, and the feeling that someone is selling you more than you need.

• Audit of the existing infrastructure: what is there, what state it is in, what is at risk of stopping
• Gap analysis between where you are and what the business will need in two years
• Improvement plan with items in priority order and an estimate against each
• Total cost over time, not just the purchase price
• A critical read of quotes you have already received, if you want a second opinion

The backup always reports success. The real question is a different one: how many hours of work you lose and how many days you are down if the server does not come up tomorrow morning.

• Incremental backups to a local copy and a cloud copy, so one failure cannot take both
• A plan with RTO and RPO in writing: how long to get working again, and how much data can be lost
• Restore tests carried out properly at set intervals, and recorded
• Alerting when a job does not run, instead of the email nobody reads
• A one-page emergency procedure, readable by whoever happens to be in that day

An antivirus installed in 2019 and a belief that nothing happens to a company your size. Ransomware does not pick targets by turnover. It picks them by how easy they are.

• Vulnerability assessment: what is exposed to the outside and what can be reached from inside
• Firewalls, antivirus and EDR configured and kept current, not merely installed
• Ransomware and phishing defences, including copies that ransomware cannot encrypt
• Staff training: two well-spent hours stop more attacks than most products
• Incident handling when it does happen, and the GDPR obligations that come with it

The software vendor is waiting on the network people, the network people are waiting on the electrician, and the go-live date went past a month ago.

• A single point of contact who coordinates suppliers and owns the dates
• A plan built on real dependencies, not a list of tasks in a row
• Budget tracking, with variances flagged while they can still be corrected
• Risk management: what can fail, what it costs if it does, what we do about it
• Reporting you read in five minutes, with the decisions waiting on you at the top

The Excel file with half the company in it, opened by six people, with three versions going around and nobody sure which one is current.

• Requirements worked out by watching how you work now, not how you ought to work
• Web applications reachable from the office and from outside, with nothing to install
• Integration with ERP, CRM and external services through their APIs
• Maintenance and changes over time, because processes move
• Code and data stay yours, with the documentation for anyone else to pick them up

Every problem turns into a search: who do I call, what will it cost, when will they come. And whoever comes does not know your infrastructure, so they start from scratch.

• A named contact who knows your infrastructure and does not start over each time
• Proactive monitoring: I notice the full disk before you do
• Scheduled preventive maintenance, updates included
• Priority over one-off requests
• A monthly fee based on how many workstations and machines you actually have

A workstation that will not boot, an application that will not open, a network printer that has vanished. Small things that still cost someone half a day.

• Diagnosis and repair on Windows and Mac workstations
• Remote when remote is enough, on site when it is not, within 24 to 48 hours
• Mail, printing, shares and access problems
• New workstations set up and configured, with the data moved across
• For completeness: hardware repair and data recovery are not done in house, but I know who to send you to