Your First 90 Days as a Company’s First Security Hire
You are the first. There is no team, no budget line you control, no documented estate, and at least two people who think your arrival means security is now entirely your problem. Somewhere there is a customer asking for ISO 27001, which is probably why you were hired. What you do in the first ninety days decides whether the next two years are possible.
The usual advice — run a maturity assessment, build a three-year strategy — assumes an organisation that already believes in the function. You do not have that yet. You have to earn it.
Days 1–30: find out what is true
Resist shipping anything. The instinct to arrive with a policy pack is strong and wrong — policy written before you know what people actually do describes a company that does not exist, and everyone will know it.
Build the asset and data inventory first
Everything downstream — risk, access, retention, breach scope, every questionnaire you will ever answer — is unanswerable without it. Two to three days of work that saves weeks. What systems exist, who owns them, what data they hold, and which of them anyone would notice going down.
Establish your four numbers
Not a maturity score. Four facts you can state in a sentence each, which an ICO enforcement notice earlier this year showed are what regulators reach for:
- What percentage of your estate is actually monitored — not "do we have logging", but would anyone notice within 24 hours if something went wrong on this system?
- How many systems are past end-of-life and unsupported.
- How old is your oldest open critical vulnerability.
- How many people hold administrative access to production, your identity provider and your cloud accounts.
Everyone is surprised by at least two of their numbers. Those four, with dates, are also the most credible thing you can put in front of a board in month two.
Interview widely, commit to nothing
Talk to engineering, support, sales and finance. Ask what they do today, what annoys them about security, and what they are quietly worried about. You will learn more about your real risks from a support engineer describing how they access customer data than from any scanner. You will also discover which of your future controls will be fought over, which lets you sequence them.
Days 31–60: fix something visible
Your credibility problem is not technical. It is that nobody yet knows whether you make things better or just slower. Spend month two proving the first.
Pick something that is genuinely a risk, genuinely fixable in weeks, and visible to people outside security. Good candidates: MFA everywhere including the service accounts people pretend do not exist; cleaning up administrative access so the number you measured in month one halves; a joiner-mover-leaver process that actually removes leavers.
Do not start with the thing that annoys engineers most. You may be right that shared admin credentials have to go, and that fight is coming — but having it in week five, before anyone has seen you add value, converts a technical argument into a political one you will lose. Sequence the unpopular work behind one visible win.
Wire up evidence collection now, not later
Roughly half the ongoing work of any compliance programme is producing evidence that a control operated: access reviews, patch status, backup tests, training completion. Done by hand it is a permanent tax paid in screenshots.
Connect the systems that already know the answer — cloud, identity provider, endpoint tooling, code repository, HR system — before you start collecting anything manually. The rule: if you will need the same evidence more than twice, wire it up instead. Doing this in month two rather than month eight is the difference between a programme you can run alone and one you cannot.
Days 61–90: make it a programme, not a person
Now write things down — because now they describe something real.
- A risk register of eight to twelve real risks, each with an owner and a decision. Not thirty imported from a template, all scored medium, which tells an auditor nobody looked.
- Policies that describe what now actually happens. Short. A policy nobody can follow is a finding waiting to be written up.
- A scope statement if certification is coming — the single biggest lever on what it costs. Scope the product and the teams that build it, not the whole company.
- A reporting rhythm: a monthly note to your sponsor, a quarterly item for the board. Start it before anyone asks, so the first report is not about an incident.
- An intake route for security questionnaires that is not your personal inbox. They will otherwise eat every week.
The things nobody warns you about
You have responsibility without authority. Every control you need touches someone else’s work and you cannot instruct any of them. The fix is not to argue harder — it is to bring the customer rather than the clause. “Annex A.8.2 requires privileged access restriction” loses. “The deal we are trying to close asked specifically about shared admin accounts, and we have four” wins.
You will be asked to approve things you cannot assess. Early on, say what you know and what you do not, and give a date for the rest. A first security hire who admits uncertainty keeps credibility; one who guesses confidently loses it the first time they are wrong.
The questionnaire queue is the silent killer. It arrives unpredictably, always urgently, always attached to revenue. Build an answer library from the very first one, so no question is ever answered twice from scratch. Publish a Trust Centre so the easy 60% never reaches you.
You need a sponsor, and the standard agrees. ISO 27001 clause 5 requires demonstrable leadership commitment, which is a polite way of saying this does not work without someone senior who will have the unpopular conversations. Identify that person in week one. If there isn’t one, that is the thing to fix before anything else on this list.
What success looks like at day 90
Not a certificate — that is months away and it is a milestone, not a goal. At ninety days you should be able to answer, from records rather than memory:
- What we have, where it is, and who owns it.
- Our four numbers, with the dates they were measured.
- The eight risks that matter and what we decided about each.
- One thing that is materially better than when I arrived, that someone outside security noticed.
- What we would do in the first hour of an incident, and who I would call.
That is a programme. Everything after it is execution.
Get your baseline in three minutes
20 questions, no signup wall. It scores where you stand against SOC 2 and ISO 27001 and maps each gap to the clause it belongs to — a reasonable day-one artefact for someone who has just inherited an undocumented estate.
Take the readiness check