Why a method
Guessing is expensive. A method finds the cause first.
Faced with "it is slow", guessing spends time and money. A method finds the cause before the fix.
Someone says: my PC is slow. What do you do? You could guess. Reinstall Windows: a day gone. Buy more memory: money spent. Replace the machine: more money. And it may still be slow, with nobody knowing why. Or you could work to a method. Ask what is slow and since when. Look at which part is at its limit. Test one cause at a time. You find the cause, and then the cheapest fix that removes it.
The method
Six steps, always in the same order
Identify, theorise, test, fix, verify, document.
Good technicians are not faster guessers. They follow a method, and it has six steps. Identify the problem: find out exactly what is wrong. Establish a theory of the probable cause. Test the theory. If the test fails, go back and form another, or pass the fault to someone who can. Plan the fix and carry it out. Verify that everything works, and stop it happening again. Document what you found and what you did.
Identify the problem
Who is affected tells you where to look
One person or many? One program or everything? The answer points at the fault.
The first two questions cost nothing and remove most of the possibilities. How many people, and how much is failing? One person and one program: look at that program, or their settings. One person and everything: look at that computer. Many people and one program: the fault is behind it, in a server or service. Many people and everything: it is something they all share, usually the network.
Form a theory
Halve the problem with every test
Test in the middle. Each answer clears half of what is left.
A website will not load, and eight things lie between this PC and the site. Checking them one by one could take seven tests. Test in the middle instead. The switch answers, so the first four are fine, and four suspects are left. Nothing answers beyond the firewall, so the last two are cleared. Two left. The router answers. That leaves only the firewall.
Test the theory
One change at a time, cheapest test first
Each question splits the causes. Change one thing at a time.
Testing a theory is a series of small questions, each with two answers. Take a PC that will not turn on. Are there any lights or fan noise? If not, the fault is power. Does a lamp work in that outlet? If the lamp fails, it is the outlet. If it works, swap the power cable, and then suspect the supply. If there are lights, is anything on the screen? No picture points to the monitor cable, the memory or the graphics.
Evidence
The log has usually recorded the cause
Find the failure in the log, then read upwards to what came before it.
Before forming theories of your own, read what the machine wrote down. This PC restarted without warning. The log has a critical event at two seconds past nine fifteen: the system rebooted without shutting down. That is the symptom, not the cause. So read upwards. Seven seconds earlier, an error: a bad block on a drive. And before that, a warning from the same drive. The drive is failing.
Evidence · Going deeper
Symptoms and first checks
Slow: Task Manager. Blue screen: the stop code. Clicking drive: back up now.
Finish the job
Verify it, prevent it, write it down
The user confirms the fix. The ticket records symptom, cause and cure.
Applying the fix is step four of six. Next, verify. Have the user confirm that everything works: all of it, not only the part you touched. Then prevent. Here the stuck print job came from an old driver, so the driver was updated. Last, document. The ticket records the symptom, the cause and the cure, so the next person to meet this fault solves it in minutes.
Professional practice
The fix is half the job
After the fault come records, people, safety and privacy: the other half of the work.
Tickets
Every request becomes a ticket
A ticket has one owner, a priority and notes anyone could pick up.
Support work runs on tickets, and each one has the same life. It is new when the user reports the fault. It is triaged: given a category and a priority. It is assigned to one named person. While it is in progress, notes are added as the work is done. If it is beyond that person, it is escalated, with every note attached. It is resolved when the fix is applied. And closed when the user confirms.
Tickets · Going deeper
Setting the priority
Impact: how many people. Urgency: how much work has stopped.
Change management
A change needs a plan, an approval and a way back
Request, assess, approve, schedule, implement, verify, with a rollback plan.
Fixing a fault is one thing. Changing a system that works is another, and it follows a stricter path. A request states what will change, and why. The risk is assessed: what could break, and for whom. A change board approves it. It is scheduled for a quiet window. It is implemented, having been tested beforehand. And it is verified. If verification fails, the rollback plan, written in advance, puts everything back.
Communication
Say what happened, what you did and what comes next
Plain words: what happened, what I did, what comes next.
The last skill is explaining. Here is one fix, described twice. In jargon, it is accurate, and it tells the user nothing. In plain words, start with what happened: your computer lost its network address. Then what you did: I have renewed it. Then what comes next: I will look in tomorrow to make sure it stays.
Safety
A spark too small to feel can destroy a chip
A chip can be damaged by 100 volts. You feel nothing below 3,000.
The parts inside a computer are far more fragile electrically than they look. A sensitive chip can be damaged by a hundred volts of static. You cannot feel a static spark below about three thousand volts. And walking across a carpet in dry air can charge you to thirty-five thousand. So the precautions are fixed habits: power off and unplug, wear a wrist strap, hold boards by their edges, and keep parts in antistatic bags.
Safety · Going deeper
Safety and disposal
Lift with the knees. Never put water on an electrical fire. Wipe drives before disposal.
Privacy
You will see private data. The job is to not use it.
Personal, health and payment data are protected by law. See only what the job needs.
A technician can open almost anything, so the law and the employer both set limits. Personal data, such as names with identity numbers, is protected by privacy law. Health data has stricter rules of its own. Payment card data is governed by the card industry's standard. And credentials are covered by a simple policy: never ask for a password. The conduct that follows is the same for all four.
Privacy · Going deeper · 1 of 2
If you find something you should not
Stop, report, preserve. Do not investigate.
Privacy · Going deeper · 2 of 2
Licensing: software is licensed, not owned
Every install needs a licence. The EULA says who may use it and where.
When a company buys software, it buys permission to use it, on terms. This company has the program installed on fifty machines. It owns forty licences. Ten copies are unlicensed, and an audit would find them. The terms depend on the type of licence: one person or one machine, an agreed number across a company, a subscription for each named user, or open source, free to use under conditions.
Lesson complete
You finished “Troubleshooting and professional practice”
Nicely done.