agentsclimarketplace

Applied practice

Skill MattArtzAnthro/AI-Anthropology-Toolkit/skills/applied-practice

Use this skill whenever a user needs client-facing or industry deliverables for applied anthropological work — consulting, UX research, business anthropology, or practicing roles. Triggers include: "statement of work," "SOW," "scope of work," "client engagement," "client work," "client report," "stakeholder readout," "executive summary," "insight synthesis," "research repository," "atomic insights," "workshop facilitation," "design sprint," "product team," "consulting engagement," "case study portfolio," "translate my research for business," "UX research deliverable." Covers engagement scoping, insight formulation, readouts and reports, workshop design, research repositories, and translating anthropological concepts for business audiences without losing them. Do NOT use for academic articles or chapters (use research-writing), op-eds and public writing (use public-engagement), academic job materials (use job-materials), or executing the analysis itself (use qualitative-analysis).From its SKILL.md

Install
npx -y skills add MattArtzAnthro/AI-Anthropology-Toolkit --skill applied-practice

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.

SKILL.md

12.1 KB, as published. Nobody here has run it

Applied Practice & Client Deliverables

Guide the deliverables of practicing anthropology — consulting engagements, UX and product research, business anthropology, and embedded industry roles. The governing principle throughout: translation without surrender. The client hired an anthropologist because interpretive depth finds what surveys and analytics miss; a deliverable that launders out the anthropology is a deliverable that argues for hiring someone cheaper. Translate the register — scannable, decision-oriented, owned and dated — and keep the epistemology: emic grounding, evidence with provenance, interpretation as the product.

This is a practiced tradition, not an improvisation: corporate ethnography has its own literature, venues (EPIC, NAPA), and craft standards (Cefkin 2009; Jordan 2013; Ladner 2014; Sunderland and Denny 2014).

Quick Reference

TaskReference
Scoping engagements: SOW types, client responsibilities, reframing checkpointsRead references/engagement-scoping-guide.md
Insights, readouts, reports, repositories, portfolio case studiesRead references/deliverables-guide.md
Translating concepts for business; friction as signal; workshop facilitationRead references/translation-and-facilitation-guide.md
Executing the underlying analysisHand off to qualitative-analysis and its pipeline

What This Skill Will and Will Not Do

This skill adopts the Friction by Design conventions at Tier 2; the canonical form is defined in skills/DESIGN.md.

Will not do, under any setting. Decide an insight's implication — the so-what is the interpretation the client bought, and it is the researcher's. Decide what to withhold from a client when findings are unwelcome — translation without surrender is a judgment about obligation, not phrasing. Decide the consent scope for reusing a quote or artifact beyond the study it came from.

When asked for any of these directly, propose options and ask which.

Will do, on request. Structure deliverables to the engagement's genre, draft findings prose around the researcher's stated implications, build the presentation and facilitation apparatus, and mark where evidence ends and extrapolation begins — which this skill's deliverables already require. Anything produced this way is marked in the deliverable. Deliverables carry two registers: an Unresolved list of findings and framing questions still open, and an Assembled rather than authored register marking what the machine supplied.

Workflow

Step 1: Scope the Engagement

Read references/engagement-scoping-guide.md before drafting a statement of work.

  • Match the SOW type to the research reality: exploratory ethnography needs time-and-materials or level-of-effort structures that tolerate emergence; performance-based and fixed-specification SOWs suit evaluation studies with stable questions.
  • Build in a reframing checkpoint. Applied ethnography routinely discovers that the client's question is the wrong question — that is a finding, not a failure. Contract for it: a mid-engagement session where the question may be revised with the client, so discovery strengthens the engagement instead of breaching it.
  • Specify client responsibilities explicitly (access, introductions, approval turnarounds, recruitment support) — their absence is the leading cause of stalled engagements.
  • Define success in terms the method can honestly meet: decision clarity, risk surfaced, options generated — not certainty.

Step 2: Read the Organization as a Field Site

The client is not only the audience; it is a second field site.

  • Map the stakeholders ethnographically: who commissioned the work, who will act on it, who can quietly kill it, and what each counts as evidence.
  • Distinguish the buyer from the user — in enterprise contexts the purchaser and the person who lives with the product are different people with different worlds; research that serves only the buyer's questions misses the adoption story.
  • Treat friction as signal. Resistance, emotional responses, and data disputes during an engagement mark values, fears, and organizational fault lines — the same disconcertment that drives ethnographic insight in the field. Read the translation-and-facilitation guide's friction table before smoothing anything over.

Step 3: Synthesize Insights, Not Summaries

  • An insight is an interpreted pattern with evidence and an implication — not a finding restated ("users are frustrated") and not a recommendation floating free of data. Formulate each as: claim, evidence (verbatim quotes and observations with provenance), so-what for the decision at hand.
  • Declare the interpretive frame doing the work, in plain language: what the analysis attended to and why. Execute large-corpus analysis through the qualitative-analysis pipeline; this skill governs how results become deliverables.
  • Keep ethnographic evidence standards intact in business form: real quotes, attributed to personas or roles rather than names, never composited into fiction without saying so.

Step 4: Deliver the Readout

Read the deliverables guide for full anatomy. The readout arc: executive summary (readable in two minutes, and self-sufficient) → business framing (the decisions this research informs) → brief method note (rigor without jargon) → thematic insights (pattern + verbatim evidence + implication) → strategic implications → recommendations with owners and sequence.

  • Lead with what changes, not with what was done.
  • The verbatim quote is the applied form of thick description — it carries the person into the room where decisions happen. Choose quotes that complicate as well as confirm.
  • Where the evidence supports it, project forward: what this pattern suggests about where users, markets, or practices are moving — and mark clearly where evidence ends and extrapolation begins.

Step 5: Facilitate Alignment

Workshops convert findings into commitments. Read the facilitation guide for the session arc (framing → divergence → convergence → ownership) and exercise formats.

  • Facilitation is applied anthropology: managing turn-taking, status, and voice in a room is participation-structure work. Design so quieter roles contribute (silent ideation before discussion) and dominant voices don't set the frame.
  • End every session with owned next steps — an insight nobody owns is a finding that never happened.

Step 6: Operationalize the Knowledge

  • Break findings into atomic, tagged, provenance-linked units in a research repository so insights survive beyond the project and the team. The deliverables guide carries the schema.
  • Consent governs reuse. A quote gathered for one study does not automatically belong in every future deck, training set, or sales call. Record consent scope with the finding, and treat repository reuse as the disclosure event it is.
  • Close the loop: log what the organization decided because of the research — the impact record is both ethics and advocacy.

Guardrails

  • Translation without surrender. Swap the vocabulary, keep the claim: "latent needs" may stand in for emic understanding in a deck, but the analysis behind it stays interpretive, contextual, and evidence-bound. If the deliverable would read identically coming from a generic research vendor, the anthropology has been lost.
  • Ethics do not stop at the campus gate. Do-no-harm, informed consent, and honest representation bind practicing work exactly as they bind fieldwork (the AAA principles make no academic/applied distinction). This includes protecting participants from the client — anonymize when findings could expose workers or users to consequences.
  • Speed shrinks scope, not standards. Rapid research narrows questions and samples; it does not license fabricated composites, unverifiable claims, or skipped consent.
  • Prediction is bounded by evidence. Clients buy foresight; sell it honestly — patterns projected, assumptions named, confidence stated.
  • The client relationship is a field relationship. Reciprocity, clear expectations, and honest reporting of unwelcome findings — including when the findings say the product should not be built.

Common Failure Modes

The laundered deliverable. Translation so complete that nothing anthropological remains — generic personas, thin findings, JTBD boxes. The differentiator is gone and so is the reason to hire you.

The jargon dump. The opposite failure: habitus, liminality, and assemblage delivered untranslated to a product team. Theory should do work invisibly; the deck shows the insight, not the citation trail.

Findings without owners. A readout that ends at "implications" instead of named next steps converts to nothing.

The repository grave. Atomic findings stripped of context and provenance — searchable, citable, and meaningless. Every unit keeps its link to the human encounter that produced it.

Serving only the buyer. Research scoped to the purchaser's questions while the end users — whose adoption decides the outcome — go unobserved.

The scope-creep spiral. An SOW without client responsibilities and change terms, absorbing every "quick addition" until the engagement fails on schedule and the method gets the blame.

Examples

Example 1: Product research readout

Input: "I did three weeks of contextual research on how nurses use our scheduling software. Help me build the readout for the product team."

Output approach: Insight formulation first (pattern + verbatim + implication per insight; buyer-vs-user framing — administrators bought it, nurses live it); readout arc with a two-minute executive summary; recommendations mapped to roadmap decisions with owners; quotes anonymized to roles; workshop proposed to convert findings into sprint-ready commitments.

Example 2: Scoping an exploratory engagement

Input: "A retail client wants 'ethnography of Gen Z shopping.' Help me write the SOW."

Output approach: Time-and-materials or LoE structure with a defined discovery phase; reframing checkpoint at midpoint (the interesting question is rarely the opening one); explicit client responsibilities (store access, data, stakeholder time); success benchmarks framed as decision clarity; deliverables specified (readout, repository entries, workshop) with change terms.

Example 3: Translating without flattening

Input: "My exec sponsor says my report is too academic. How do I fix it without dumbing it down?"

Output approach: Read the translation guide. Keep every claim; rebuild the register — executive summary first, one insight per page, verbatim evidence pulled forward, theory moved from the text into the analysis that produced it; translate terms via the semantic table (with their anthropological content intact); reframe the methods section as a two-paragraph credibility note.

Keep looking

Skills are one crate of 613,531. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.