advertising-hiring

Imported from mbfinotti/advertising-skills (skills/advertising-hiring/SKILL.md) via skills.sh. Install upstream with npx skills add mbfinotti/advertising-skills --skill advertising-hiring. Copyright stays with the author (MIT).

Advertising Hiring

Build the packet a marketing leader needs to recruit a paid-advertising hire: role scoping, interview loop and scorecard, work sample, and a ramp plan with staged spend authority. Scoping comes first: the field's own hiring managers trace mis-hires to it, not to interviewing technique - Duane Brown (Take Some Risk): "Bad hires are the results of bad job descriptions or interviews"; Constant Hire: "a mis-hire is usually a scoping [error]."

Titles blur, but a real functional ladder sits underneath them, each rung adding scope over the one below:

• Media buyer: channel execution.
• Performance marketer: adds funnel strategy, creative testing, measurement, business outcome.
• Growth marketer: the whole system, including conversion and lifecycle.
• Creative strategist: a separate role bridging creative and buying, not a senior buyer.

The frame borrows the scorecard tradition popularized by Geoff Smart and Randy Street's Who: mission, ranked outcomes, then competencies, all defined before anyone is evaluated.

This is the hirer's side only: it does not source candidates, post jobs, run applicant tracking, or write offer letters, and anything touching compensation terms, contracts, or employment law needs HR/legal review. It quotes published salary bands with source and date as scoping input, always labelled directional. The candidate side - breaking into and progressing in paid media - is the mirror-image sibling mbfinotti/advertising-skills@advertising-career; if the person you are talking to turns out to be a candidate, route them there.

Interview

Ask before designing. One question per message; offer multiple-choice answers where possible. Skip anything already answered.

• What stage and size is the company? (a) under 15 people / pre-launch, (b) 15-150, (c) 150+, (d) an agency hiring for its own bench.
• Is this the first dedicated paid-media hire, or an addition to an existing team? If existing: who is on it and which channels do they own?
• Which paid channels are live or planned? (a) one, (b) two, (c) three or more, (d) none live yet.
• What monthly ad budget will this hire manage - now and in 12 months? (a) under $10k, (b) $10-50k, (c) $50-250k, (d) $250k+.
• What segment? (a) B2C / DTC ecommerce, (b) B2B, (c) app or marketplace, (d) mixed.
• Is this an in-house brand-side hire or an agency-side one - and does your candidate pool skew agency-trained or brand-trained?
• How trustworthy is measurement today - conversion tracking validated, platform numbers reconciled against backend revenue? (a) yes, (b) partly, (c) no, (d) unknown.
• Who is on the interview panel, and how many rounds can you realistically run?
• By what date must this seat be producing? (a) inside a month, (b) this quarter, (c) no fixed date. A hard date promotes the options that cover the channels this week - a retained agency or a freelancer bridging while you hire - and demotes anything that adds an interview round.
• Do you want a one-off fix for the spend running today, or a capability that compounds? A one-off answer promotes outside help and a contract; a compounding answer promotes the in-house hire, the higher rung, and the full ramp.
• What is your effort ceiling - manager and interviewer hours per candidate, panel size, and how reversible the decision must stay (contract versus permanent headcount)? A low ceiling promotes the two-round loop and the own-work presentation; a high ceiling unlocks the live audit and the paid trial. No budget or approval for paid candidate work deletes the paid trial from the menu instead of ranking it last - say it is deleted, so it cannot resurface later as an unfunded round.

Workflow

Every ordering below is a default, not a law: it shifts with the company's context and with who actually executes the loop. Re-rank it against what you already know, since each of these collapses the effort of one option and moves it up:

• A bench of candidates from a previous search.
• An in-house recruiter.
• An agency already retained.
• A founder who will personally sit every round.

1. Run the Interview; collect every answer before designing anything.

2. Gate whether to hire at all. AJ Wilcox's published don't-hire conditions (scoped to one B2B social channel, but they generalize) are the clearest gate: monthly budget under about $5,000, marketing fundamentals not in place (audience, positioning, pricing, working site), or unwillingness to commit 3-6 months before judging a channel. When the gate fails, rank the alternatives by channel coverage bought per hour of your own time: agency > freelancer > self-manage, and wait only while the missing fundamental is what blocks. The axes disagree - the agency costs the most control and the least of your hours:

   • value (channels covered, creative bandwidth, cover for absence): agency > freelancer > self-manage
   • effort (your hours briefing, reviewing, coordinating): self-manage > freelancer > agency
   • compliance cost (contract review, third-party access to ad accounts and customer data): freelancer > agency > self-manage - the freelancer is not tied with the agency here: both need the same contract and the same account access, but only the individual engagement raises a worker-classification question, and only that one is hard to unwind after the fact. Full breakdown in references/role-ladder-and-leveling.md. When the gate fails on a missing fundamental, wait is not a rung on this list - the hire is deleted from the menu until positioning, pricing, or the site is fixed, and saying so stops it reappearing next month as an unscoped req.

3. Name the rung in writing before anything else, using the ladder in the same reference. Default rung for a first dedicated hire: the generalist operator (performance-marketer profile) who owns measurement and channel choice - never a single-channel specialist. Ordered by business outcome owned per week of ramp: performance marketer > media buyer > creative strategist > growth marketer. Value and effort disagree with that, which is the point - the default is neither the broadest seat nor the fastest one to fill:

   • value (breadth the seat can own): growth marketer > performance marketer > creative strategist > media buyer

   • effort (time-to-fill, ramp length, panel time, comp pressure): growth marketer > creative strategist > performance marketer > media buyer Conditions that move the default:

     • A working funnel that needs depth in one channel moves it down to a media buyer.
     • Spend past the ~$50k/month breakpoint, or 3+ paid channels, adds a dedicated buyer alongside rather than instead ("a generalist starts dropping balls", both Remoby).
     • Creative volume as the bottleneck opens a creative strategist as a separate req.
     • Conversion and lifecycle already in scope moves it up to a growth marketer.

     What this order starves: the growth marketer, top of the value order and top of the effort order at once, so a ratio ranks it last however wide the seat is. Promote it deliberately - not by waiting for the ratio to pick it - when conversion and lifecycle are already the constraint, when the seat is meant to outlast two or three channel cycles, or when the panel and the comp band can absorb a longer search. The same trap runs one rung down: the specialist buyer loses to the generalist on efficiency at every spend level, and only the breakpoint condition above ever promotes it. A funded headcount the comp band cannot reach is deleted from the menu, not ranked last - re-scope the req to a rung you can actually fill and say which rung you removed.

4. Write the role definition as a scorecard: one-line mission, 3-6 ranked measurable outcomes with dates, then 4-6 competencies. Use spend-under-management as a filter, never a level - pair any "manages $X/month" screen with "walk me through a campaign you killed and why" and an incrementality-versus-attribution probe (Constant Hire). Pressure-test the funded level against the published bands in the reference; quote every figure with source and date, labelled directional.

5. Build the question bank from references/question-bank-and-scorecard.md: judgment over recall - "Never cite 'best practices' or default methods as your rationale" (Navah Hopkins). Spend the interview minutes in this order, by judgment revealed per minute asked: two senior filters > exaggerator probes > killed-campaign and incrementality probes > named situational questions > commitment screen. Effort and value disagree, because the cheapest questions are not the deciding ones:

   • value (separates a senior operator from a fluent one): two senior filters > killed-campaign and incrementality probes > named situational questions > exaggerator probes > commitment screen
   • effort (minutes asked, plus the interviewer expertise needed to grade the answer): two senior filters > killed-campaign and incrementality probes > named situational questions > commitment screen > exaggerator probes The two senior filters are diagnosing a ROAS drop in the fixed order (measurement, then media, then creative, then post-click) and translating platform ROAS into MER and break-even reality. Promote the exaggerator probes to the screen when the pool is unvetted; demote them when a trusted referral or a past-search bench already covers provenance.

6. Map the loop: 2-3 rounds maximum (screen, time-boxed technical/case round, peer or founder round). Duane Brown cut his agency's loop from 3-4 interviews to 2 with standardized questions and reports better hires for it. Each competency is owned by exactly one stage, and each panelist has a lane. Say where this loop shape comes from: the published evidence behind it is search-heavy small and mid-sized agencies. Enterprise in-house and DTC paid-social loops are far less documented, so treat the round structure as transferable and the round content as needing a local check.

7. Fix scoring discipline before anyone interviews:

   • A role-specific behavioral description per anchor point - a bare number is banned.
   • Every score cites a quote or concrete example.
   • Interviewers submit scores independently before any debrief.

   Weight commercial reasoning and communication over technical recall; the only published rubric (Pare - vendor content, label it as such) weights Technical 40 / Communication 30 / Commercial insight 30.

8. Design the work sample from references/work-sample-design.md, picking the format by signal returned per candidate-hour, net of who drops out rather than do it: live screen-share audit > data-table case > own-work presentation > paid trial > job shadow. The multi-hour unpaid take-home is off the list entirely - it is the one format whose funnel cost exceeds its signal. The axes disagree, so the strongest format is not the cheapest one:

   • value (how much of the candidate's actual method you see): paid trial > live screen-share audit > data-table case > own-work presentation > job shadow

   • effort (scheduling, panel hours, prep, budget approval, calendar drag): paid trial > job shadow > live screen-share audit == data-table case > own-work presentation

   • candidate drop-off (seniors who decline rather than comply): unpaid take-home > paid trial > job shadow > live screen-share audit == data-table case > own-work presentation

   • compliance cost (pay terms and worker classification for real work; access to live accounts and customer data): paid trial > job shadow > live screen-share audit > own-work presentation > data-table case The audit and the data-table case tie twice for the same structural reason, not for want of a decision: both consume one hour of an already-scheduled interview and one prepared thing (a panelist who can grade live, or a table built in advance), and neither asks the candidate for an hour outside the loop, which is what drives drop-off. They do not tie on compliance, so that line breaks them: an own-work presentation invites a candidate to walk you through a former employer's account, which is a real exposure that a synthetic table never carries - ask for it anonymized, and score the anonymizing. What this order starves: the paid trial, first on value and first on effort, so it can never win a ratio - and the in-interview audit takes the top slot every time despite showing less of the candidate's real method. Promote the paid trial explicitly for a senior hands-on seat, for a creative strategist whose portfolio cannot be trusted to be their own work, or when a candidate already contracting elsewhere makes a week of their time cheap to buy. Delete rather than demote everywhere your constraints bind:

     • No budget or approval for paid work removes the paid trial and the job shadow from the menu.
     • No panelist able to grade an audit in real time removes the live audit.
     • Accounts that cannot be exposed to an outsider remove it too.

     Record each removal in the packet by name - a format left ranked last comes back as a round nobody scoped. Pay for any real account work, use synthetic or anonymized data everywhere else, and build in deliberate ambiguity scored as a top answer whichever format survives.

9. Build the ramp from references/ramp-and-spend-authority.md: all access on day one, and an explicit spend-approval threshold in week one. Stage the authority ladder:

   • Shadow, weeks 1-2.
   • Co-sign within caps, weeks 3-6.
   • Independent within a cap, by day 60.
   • Full authority, by day 90, contingent on a validated measurement audit.

   The first win is a tracking fix or a contained optimization, never a scaling bet. Label the 30-60-90 milestones as this skill's own synthesis rather than a published standard.

10. Emit the packet (next section) section by section for user validation, grounding each section in references/worked-examples.md.

11. Iterate until every item in Pass Threshold holds.

12. If your harness has persistent memory, memorize the approved packet - rung, level, competencies, stage map, spend ladder - so screening runs and the eventual ramp start from it instead of re-interviewing. Without memory, tell the user to keep the packet document as the standing input for future runs.

The Hiring Packet

Deliver every engagement as this artifact. Fully worked versions live in references/worked-examples.md.

HIRING PACKET - <company>, <role>, <date>, v<n>
Rung decision  : ladder rung + why | first-hire vs specialist | segment + context fit
Role definition: mission | 3-6 ranked outcomes with dates | 4-6 competencies
Requirements   : must-haves (5-7 max) | nice-to-haves | disqualifiers separate |
                 spend screen paired with kill-decision + incrementality probes
Comp input     : published band quoted with source + date, directional | flag: HR/legal
Stage map      : 2-3 rounds | stage -> interviewer -> competencies (one owner each)
Question bank  : per competency - behavioral + situational + probes | the two senior filters
Scorecard      : behavioral anchors | cited evidence per score | weights labelled sourced or
                 own convention
Work sample    : format picked off the step-8 order, with the reason it moved | 60-90 min in
                 interview, or paid if real account work | deliberate ambiguity |
                 rubric written first | debrief agenda
Ramp plan      : day-one access list | week-one spend threshold | shadow -> co-sign ->
                 independent-within-cap -> full authority | measurement audit before targets |
                 watch areas

B2B and B2C

The hiring process itself is identical across segments - scorecard, loop, work sample, staged ramp - and say so when asked. What changes is which instincts and metrics the role must own, and those instincts do not transfer by default.

Constant Hire, a DTC recruiter, refuses B2B searches outright: "A buyer who has run paid search and lead generation for a SaaS company optimizes toward pipeline, not first-order profitability. Those instincts do not transfer well." The Remote Reps states the mirror: a buyer from B2B or brand work "will face a steep learning curve" in a DTC account, where campaigns "live and die on creative performance, audience segmentation, and the ability to read data quickly and act on it."

• B2C/DTC: the role owns first-order economics - ROAS against break-even, contribution margin, CAC, creative-testing velocity. The creative strategist as a distinct, senior, expensive role is largely a DTC phenomenon (US creative strategy directors run $180,000-$250,000+; Foxwell Digital, 2026, directional).
• B2B: the role owns pipeline - cost per qualified lead and cost per sales-qualified lead over long cycles, where clicks cost several times more but cost-per-SQL can still win. Budget floors are higher and channel verdicts take 3-6 months (AJ Wilcox).

Treat a segment-crossing candidate as a probe-and-weight case, never an automatic rejection.

Pass Threshold

Iterate until every item holds. Each item states whether it rests on sourced practitioner guidance or on this skill's own convention.

• The rung is named in writing, with the first-hire-versus-specialist call, before the role definition is drafted (sourced: every practitioner verdict on mis-hires traces to scoping).
• Every ranked order - the rung, the work-sample format, and the hire-at-all alternative - follows these rules (this skill's rule; the orders are defaults, and an unexplained pick is indistinguishable from taste):
  • The option picked records which condition moved it off the default, or states that nothing did.
  • Every option your constraints rule out is written down as deleted by name, rather than left ranked last.
  • The high-value, high-effort option each order starves - the growth-marketer rung, the paid trial - is named with the condition that would promote it.
• Spend-under-management appears only as a filter, and every spend screen is paired with a kill-decision question and an incrementality-versus-attribution probe (sourced: Constant Hire).
• The loop runs at most 3 rounds (sourced: Duane Brown, DigitalSamIAm, and the documented community consensus):
  • Any exercise fits inside a 60-90 minute session, or is paid.
  • No multi-hour unpaid take-home anywhere.
• Every question maps to exactly one competency and probes reasoning, not recall - a "best practices" model answer fails the bank (sourced: Navah Hopkins, Kasim Aslam).
• Every anchor point carries a role-specific behavioral description and every submitted score cites evidence; the numeric weights are labelled as this skill's own convention rather than a published standard (see the scorecard reference).
• The work sample uses synthetic or anonymized data, its rubric exists before the first submission, and it contains deliberate ambiguity scored as signal (sourced: Jason Parks, Steve Gibson; the rubric-first rule is general work-sample guidance).
• Certifications appear nowhere in competence scoring; if partner compliance requires them, the req says so explicitly so it is not mistaken for a skill bar (sourced: uniform practitioner testimony).
• The ramp grants all access on day one, sets the spend threshold in week one, stages authority shadow -> co-sign -> independent-within-cap -> full, and schedules the measurement audit before any performance target (sourced: eCommerce Placement for the two rules; the ladder itself is this skill's synthesis, labelled as such).
• Every salary figure carries its source and date and is labelled directional; turnover is quoted as the 18-45% range with named sources, never as a single industry rate (this skill's rule, from the documented spread).
• A segment-crossing candidate (DTC-to-B2B or the reverse) gets the instincts-transfer probe - weighted, not auto-rejected (sourced: Constant Hire, The Remote Reps).

Common Failure Modes

Invocation Examples

• "We're a DTC brand spending $80k/month on two channels with no dedicated buyer - scope the role and build the interview loop."
• "I need interview questions and a scorecard for a senior performance marketer joining our B2B growth team."
• "Design a work sample for a media buyer that doesn't require an unpaid take-home."
• "Our new paid media manager starts Monday - write the 30-60-90 with spend-authority staging."

Reference

• See mbfinotti/advertising-skills@advertising-career for the mirror-image candidate side - breaking into and progressing in paid media.
• See mbfinotti/advertising-skills@ad-account-diagnostic and mbfinotti/advertising-skills@ad-attribution-gap for the domain depth behind the two senior filters - the ROAS-drop tree and the attribution probes.
• See mbfinotti/advertising-skills@cac-roas-benchmark and mbfinotti/advertising-skills@ad-spend-guardrails for the MER, CAC, and break-even math candidates are tested on.
• See mbfinotti/advertising-skills@ad-conversion-tracking for the measurement audit the ramp gates full spend authority on.