What Selene builds for a real role
A care home, a warehouse, an engineering team and a finance function do not hire the same way — and they should not be screened the same way either. Here is what comes out of a conversation with Selene across six very different roles, and one of them walked through end to end.
These are illustrations, not templates. There is no library of pre-built roles to pick from — every set of criteria below is the kind of thing Selene builds from what a hiring manager tells her about their role, at their company. Yours will look different, because your role is different. Candidate names and details are invented for the example.
Care Assistant, from one line to a ranked shortlist
High volume, tight eligibility rules, and a manager with no time. The hardest kind of role to screen well by hand — and the clearest example of what Selene is for.
You say it the way you'd say it to a colleague
No form, no required fields, no wizard. One sentence is a perfectly good starting point — Selene asks for what she still needs, one question at a time, and tailors those questions to the kind of role it is. For a frontline role she goes after shift patterns and volume; for an engineering role she'd be asking about your stack.
She also checks the rate against the market before saving it, and says so when it looks off. That is a cheap thing to hear at the start of a role and an expensive thing to discover three weeks into it.
Six kinds of criterion, three tiers, weighted
Selene turns the conversation into a structured set of requirements — each one typed, tiered and weighted, and every one of them editable inline. Must-haves stay deliberately few, because a list where everything is critical ranks nobody.
- Typed as skill, knowledge, experience, qualification, behaviour or eligibility
- Tiered must-have / important / nice-to-have, weighted 1–4
- Your reusable company criteria — right to work, location, values — applied automatically
- Change any of it and the scoring guide follows
Ask the disqualifying questions first
Some things are yes-or-no and there is no point scoring a CV against them. Selene puts those on your public application page as eligibility questions, so the answer arrives with the application instead of surfacing in a phone call a week later.
You share that link wherever you already advertise — Selene doesn't post your role or distribute it for you. Every applicant who follows it lands in one pipeline.
Every applicant read the same way
Applicant number 180 is measured against exactly the same guide as applicant number 3 — and the scorer never sees who any of them are. No name, no raw CV, and stated date of birth, sex, nationality and religion stripped before scoring runs.
What comes back is a score per criterion with the evidence behind it, the gaps it couldn't fill, and how confident it is. Where it isn't sure, it says so rather than rounding up.
How the blind scoring works →A ranked list, and a reason for every name on it
You get an ordered shortlist with a recommend-or-decline call on each person, and can open any score to see what it was based on. Share a read-only version with your registered manager or owner — reasoning included, no account needed to open it.
Selene recommends. You decide, interview, and make the call — and if you override a score, the reason is stored with it.
The same method, a completely different guide
What changes per role isn't the rigour — it's what gets asked, what counts as a must-have, and where the weight goes. A sample of what that looks like elsewhere.
High volume, fast turnover
Selene asked about: shift patterns, pick rates, and whether forklift tickets are funded in-house — because that decides whether certification is a must-have or a bonus.
Depth over keyword bingo
Selene asked about: the stack, the scale it runs at, and whether this hire is meant to lead — the answers move leadership from nice-to-have to critical, and change the whole guide.
Qualification-gated, systems-heavy
Selene asked about: the tools and the transaction volume — a five-entity consolidation in NetSuite is a different job from bookkeeping in Xero, and the same job title covers both.
Thin CVs, real signals
Selene asked about: covers per day, whether they'd run a shift alone, and how much of the team is under 21 — because supervising sixteen-year-olds on a Saturday is the actual job.
Evidence, not adjectives
Selene asked about: deal size and sales cycle — a nine-month £200k enterprise cycle and a two-week SMB close reward almost opposite backgrounds, and the CV language is identical.
Almost certainly not on this page
Which is the point. Selene builds the guide from what you tell her matters — not from a template you have to bend your role to fit. Describe it in a sentence and see what comes back.
Set up a role →The guide changes. The rigour doesn't.
Scored blind
The scorer never receives a name or a raw CV, and stated age, sex, nationality and religion are stripped before it runs — whether it's a care assistant or a CTO.
Evidence attached
Every criterion carries what was found, what was missing, and how confident the judgment is. A low-confidence score says so instead of rounding up.
Pinned to a version
Each score records which version of the guide produced it. Edit the guide and affected scores show as stale rather than being quietly rewritten.
See what she builds for yours
Describe a role in a sentence. You'll have requirements and a weighted scoring guide to argue with before you've finished your coffee.