facilit8 · Capability-Lens for Open Roles · Commercial Edition
Look Before You Backfill:
The Capability-Lens Playbook
for Open Commercial Roles
For HR and commercial leaders. Run every open sales, marketing or service role through the lens before you repost it. Let AI do the AI-able work. Free your people for the work only people can do — the accountabilities and the soft skills — and put the right person on it. A role-design tool, not a layoff tool.
Owner guide:
HR / People
Commercial
Both
Facilitator
v2 — facilit8 · Capability-Lens Series · 2026
Step 0
▾
Integrity Check — Run This Before You Use the Lens
5 questions · 2 minutes · tells you whether this is safe to use yet
The same tool that redesigns a role well can be misused to justify never filling it. These five questions tell you which one you're about to build. Answer honestly — the result is for you, not your board.
Q1
Is this role genuinely open — no incumbent, and nobody being quietly managed out behind it?
Q2
Do both an HR lead and the commercial manager own this review and the decision together?
Q3
Do you have a real picture of what the team actually does — not just the job description?
Q4
Is there real permission to reshape the role and develop people — not only to fill it or freeze it?
Q5
Have you agreed in writing that the lens informs the decision but never makes it?
The rule: the lens asks questions; it never says "don't hire." Every run gives four outputs — what AI can carry, the human work that's left, who's the right person for it, and a bridge with a decision date. People decide.
Prerequisite — The Scope Fence
Open roles only — never a role with a person attached
Open means no incumbent: no displacement, no consultation trigger. If someone is behind the vacancy — retiring, resigning under pressure, being restructured — stop. That is a different, governed conversation, and the wrong job for this lens. In Switzerland it can trigger consultation and mass-dismissal duties (Code of Obligations Art. 335d ff. and the Participation Act). Keep open-role and filled-role work as separate engagements.
Record this before you begin
"This covers an open role with no incumbent. It informs — it does not make — the decision, which [HR lead] and [commercial manager] own jointly."
1
Confirm the fence and name two owners
Confirm in writing the role is open. Name both owners: HR and the commercial manager. One owner drifts — commercial toward cost, HR toward process; the pair holds the line. Treat the Integrity Check above as a hard gate: red means stop, amber means fix the foundation first. Target: Day 2.
⚡ Smaller Mittelstand: Day 1–2
🏢 Works-council context: Week 1, loop in HR
Success looks like: Every open role has two named owners and a recorded scope-fence confirmation before recruiting spend starts.
2
Read the job description assuming it lies
The JD is a recruiting artefact, not the truth. It's stale and it misses the glue work that holds a team together. Spend 30 minutes with the people who sat next to the role: "What did this person actually do that isn't written down?" That answer is the real job. Target: Day 8.
Watch for: Analysis built on the JD alone inherits its flaws. No lived picture of the role means a confident, wrong output.
3
Run the role through the capability lens
Take the real role through the lens (next tab). It gives four outputs, never a verdict: (1) what AI can carry; (2) the human work that's left — the accountabilities and soft skills; (3) who's the right person for that work; (4) a bridge with a decision date. The caveat travels with every output. Target: Day 18.
Success looks like: Each role leaves the lens with four written outputs and a confidence note — and none of them reads "don't hire."
4
Decide together and set the sunset
The two owners pick a path and write down why. Every bridge gets a sunset date; on expiry the default is hire, not keep automating. Aim for a better team in six months — not a smaller one this quarter. Target: Day 25.
⚡ Smaller Mittelstand: Week 3–4
🏢 Works-council context: Week 4–6, HR sign-off
Reversibility warning: Not-hiring is cheap to decide and expensive to undo. When the lens is ambiguous, hire and keep reshaping — don't freeze.
A job is accountability, not a task list
Step 1High impactBoth
Don't read a role as tasks and tick off what AI could do. A job is a bundle of accountability. AI drafts, summarises, updates — but it can't be on the hook when the forecast is wrong or the customer's furious. Start with what the role is accountable for. Tasks come second.
Ask: "What breaks, and who carries it, if this had no human owner for three months?" That's usually more than the JD admits.
Hunt for the glue work the JD never mentions
Step 2High impactCommercial
The best commercial work is the least documented: sensing a renewal is wobbling, killing a tempting-but-wrong campaign, being the person everyone routes the angry customer to. That's glue work — judgment, trust, memory. It's invisible to a JD. Map it before you ask what AI can take, or the lens will quietly cut the most valuable part.
Success looks like: The team can point to the role's glue work and agree it stays human — named before any automation question.
Ask the four questions
Step 3Both
Run the real role against four questions. (1) Carry: what can AI do? (2) Human work: what's left — the accountabilities and soft skills? (3) Right person: who's best placed to own that work? (4) Protect: what must never be automated? Four answers — never a recommendation not to hire.
Capacity check: A proper run takes a few focused hours across the owners and the team. Twenty minutes gets you the JD's answer, not the real one.
Make the caveat travel with the number
Step 4High impactHR
The master risk: a conclusion outlives its caveats. The tidy line gets screenshotted into a board deck; the "this JD is unreliable" footnote drops off. Build the caveat in so it can't be separated — confidence band plus the top two reasons it's wrong, in the same box. If it can be quoted without the caveat, the format failed.
Why it matters: A confident output on a bad input produces industrialised error — one stale JD becoming an org-wide pattern of wrong, certain-looking calls.
Same Engine, Opposite Outcomes
| The misuse | The good version | Why the difference holds |
|---|---|---|
| Declares the role mostly automatable; recommends no backfill | Gives four options; a person chooses | Output is questions, never a verdict |
| A single score that travels to the board alone | Every output carries its confidence and top two flaws | The caveat can't be stripped off |
| Quietly applied to filled roles and people | Open roles only — filled roles are a separate track | The scope fence is structural |
| A bridge that silently becomes permanent | Every bridge has a sunset; default on expiry is hire | Drift has a forced stop date |
| Automates away the work where people build skill | Protects the work where capability is grown | Capability outlasts this year's saving |
| Priced on headcount saved | Priced on the quality of the decision | The model removes the incentive to cut |
The One-Page Lens Worksheet
Run this for each open role
Four questions, one caveat — and deliberately no score
Fill this in with both owners, from the real role — not the JD. There's nothing to total and no box that reads "don't hire," by design. If you start inventing a score, stop: that's the misuse version returning.
1 · Carry — What routine, codifiable work can AI take on?
2 · Human work — What's left that only a person can own — accountability, judgment, relationships, the soft skills?
3 · Right person — Who's best placed for that work, and what support and bridge (with a decision date) gets them there?
4 · Protect — What must never be automated? (Empty list = look harder.)
1 · Carry — What routine, codifiable work can AI take on?
2 · Human work — What's left that only a person can own — accountability, judgment, relationships, the soft skills?
3 · Right person — Who's best placed for that work, and what support and bridge (with a decision date) gets them there?
4 · Protect — What must never be automated? (Empty list = look harder.)
Mandatory caveat — travels with every output
"Based on a JD we know is imperfect. Confidence: low / medium / high. Top two reasons it's wrong: [reason], [reason]. Informs a decision a named person owns — it does not make one."
Sales — let AI do the codifiable work, people do the trust
Per open roleHigh impactCommercial
Give AI the codifiable load: outreach drafts, CRM hygiene, first-draft proposals. What's left is human — qualification, discovery, trust, the negotiation. In Mittelstand B2B, that's where deals are won. Shape the role around that work, and put your strongest relationship people on it.
Success looks like: Reps spend most of their time on relationships and judgment, with the routine handled in the background.
Marketing — automate production, keep the judgment human
Per open roleCommercial
AI drafts and reports fast. The human work is positioning, brand judgment, and knowing what not to publish. Shape the role around that — directing and editing AI, owning the calls that protect a precision brand. The right person is whoever has that judgment.
Nuance: "Faster drafting" is a capability, not a strategy. Automate production but never the judgment, or you ship more that says less.
Service — automate the routine, keep the saves human
Per open roleHigh impactCommercial
Let AI deflect routine tickets and draft standard replies. The human work retains revenue: de-escalation, complex problems, the loyalty moments. In B2B manufacturing, service is where accounts are saved or lost. Shape the role around the hard cases and route them to the person best at them.
Success looks like: The team spends far more of its time on retention and complex saves; routine queries resolve on their own.
Don't let "reshape" lose by default
Every reviewBoth
Cost (cut a salary) and bridge (delay) are easy to measure. "Work matched to the right person" isn't — so it tends to lose. Counter it: state the reshape option first, give it an owner, and put a rough value on the upside. Measure only what's easy and you'll only ever choose what's cheap.
Watch for: "We'll reshape it later" is how reshape dies. No owner, no date — then it's a polite no.
The point of the lens: an open role is the cheapest moment to develop people. No incumbent, real work to grow into. Ask "who's the right person for the human work, and what would help them?" before "who do we recruit?"
Free people for the work that matters
Per open roleHigh impactHR
AI carries the routine load. That frees people for the accountabilities and soft skills only they can do. So treat the open role as a chance to give that work to the right person — and develop them into it — rather than recruiting on reflex.
Make the growth plan real, not implied
A growth plan that isn't funded, scheduled and visible is just extra work with a compliment attached. Name the person, the capability, the mentor, what AI carries, and the review date. Unfunded "development" becomes overload — and the person leaves.
Use AI to remove the drudgery, not the work
Bridge phaseCommercial
Healthy: AI takes the low-judgment, high-volume tasks so people do the human work. Unhealthy: the team quietly shifts to just checking AI output — a reorg nobody approved. Name which one you're doing, out loud, and check back.
Watch for: A "humans review the machine" drift de-skills the team and removes your redundancy. That's a reorg — treat it as one, with consultation.
Keep the muscle alive
OngoingFacilitator
The trap: automate a skill, stop practising it, lose it — then the next review "proves" you can't do it, so you automate more. Break the loop. Keep people doing the judgment work. AI is a floor under the load, not a replacement for the craft.
Dependency note: Building around a tool creates a quiet dependency on its price and quality. Keep the human skill alive so a vendor change isn't a capability crisis.
Protect the roles where people learn the work
Per roleHigh impactHR
Some roles look the most automatable precisely because they're where people build judgment, relationships and accountability. Automate them away and you lose the place capability is grown. In the Swiss Lehre system it's also part of who you are as an employer. Protect these as capability investments, not costs.
Success looks like: Learning roles are ring-fenced from automation logic and discussed as investment, not overhead.
Hard carve-out: safety-, compliance-, quality- and relationship-critical work gets no score and no automation call — it's marked "human-required, not assessed." Naming what you won't touch is what makes the rest credible.
The relationship and the close (Sales)
Human-requiredCommercial
In Mittelstand B2B, deals run on years of trust and reading the room. The relationship is the asset, not a task. AI can prepare the person; it can't replace them. Over-cut the legible work and under-value the trust, and you make the most expensive mistake the lens allows.
High risk if ignored: Hollow out relationship ownership to save cost and you learn its value only when the account leaves.
Brand judgment and what not to publish (Marketing)
Human-requiredCommercial
A tool will happily produce a clever campaign that's subtly wrong for a conservative brand in a small market. Positioning, tone, and what not to say are human-required — the cost of a misstep is trust, not impressions. Protect brand stewardship; let AI speed up production beneath a person who owns the brand.
Nuance: Producing assets is easy to automate and easy to over-cut. Knowing which asset is right is where the value sits.
De-escalation and the loyalty moment (Service)
Human-requiredCommercial
The angry customer, the high-stakes complaint, the account on the line — human-required. Routine deflection is fine; routing a furious key customer to a bot turns a recoverable problem into a lost contract. Protect these moments and design the tool to hand off to a person early.
Success looks like: Routine queries resolve without a person; high-stakes moments reach a skilled human fast, by design.
Accountability and institutional memory (All)
Human-requiredBoth
Across all three functions, the connective tissue is human-required. Someone has to be on the hook when it goes wrong, hold the context that isn't written down, notice what no dashboard flags. AI does tasks; it can't carry accountability. These are exactly the parts a task-by-task lens makes invisible.
Watch for: No "don't automate this" list means the lens didn't look hard enough. An empty list is a red flag, not a clean bill of health.
The tool refuses to decide
StructuralBoth
The key control: the lens outputs questions, never "don't hire." No verdict, no ranking that implies one, no number presented as a decision. That's the structural line between a role-design tool and a layoff engine. Let it conclude, and you build the bad version by accident.
Why structural beats policy: "Please add the caveats" fails under pressure. A tool that can't produce a caveat-free verdict holds when no one's watching.
Architect the scope fence so it can't be crossed
StructuralHR
Open-role review and filled-role analysis are separate engagements — not a toggle. The tool refuses individual and filled-role data. That defuses the displacement and consultation exposure, and protects HR from "while you're at it, look at the team we already have."
Capacity check: Budget pressure pushes open → filled. The fence has to be architecture, not willpower.
The pair must be real: Co-ownership only holds if HR can actually halt a decision. Where HR is outranked and would rubber-stamp, the safeguard is fiction — fix the authority first, or don't run the lens.
Mandatory equal-treatment review before anyone acts
StructuralHRCompliance
Recommendations can skew by age or cluster on certain roles — real non-discrimination exposure. Under Swiss law that includes the Gender Equality Act (GlG) and the personality-protection duties in the Code of Obligations. Run a mandatory equal-treatment review before acting. HR owns this gate; it can't be skipped for speed.
Not legal advice: This is a governance prompt, not a legal opinion. Confirm specific exposure with Swiss employment counsel before acting.
Forced sunset on every bridge
StructuralBoth
Every bridge has an expiry date, and on expiry the default is hire — not keep automating quietly. That stops the common failure: a temporary bridge becoming a permanent reorg nobody decided on. Extending past the sunset takes a fresh, explicit decision by the same two owners. Drift can't happen by inertia.
Success looks like: No bridge runs without a visible sunset, and every expiry forces a real decision.
Price on decision quality, and keep the liability clean
StructuralHRCommercial
The commercial model is a control. facilit8 prices on the quality of the decision, never on headcount removed — so there's no incentive to cut. The liability hygiene is explicit: decision-support only, the client owns every decision in writing, and the lens is never the sole basis for not backfilling. This lives in the engagement letter, not just the conversation.
Why it's here: If the fee grows when headcount shrinks, every other safeguard swims against the incentive. Align the money with the good version.
25
Specific actions across 7 tabs
4
Outputs per role — never a "don't hire"
5
Structural guardrails that beat policy
1
Scope fence: open roles only, by design
facilit8 · Commercial Transformation enabled by Technology
Decision-support only, and not legal advice — confirm employment-law questions with Swiss counsel. The client owns every hiring decision in writing; the lens is never the sole basis for not backfilling a role. Liability terms live in the engagement letter, not on this page.
Decision-support only, and not legal advice — confirm employment-law questions with Swiss counsel. The client owns every hiring decision in writing; the lens is never the sole basis for not backfilling a role. Liability terms live in the engagement letter, not on this page.
An open commercial role? Run it through the lens.
A 30-minute conversation with David can show you how the capability-lens framework applies to your next open role — and how to build a repeatable process around it.
Want this playbook by email?
Leave your email and I'll send you a copy — no spam, unsubscribe anytime.