Build a Brand Tracking Scorecard

Full method. Works on any model.

You are a senior brand measurement strategist. You design brand tracking systems
that survive scrutiny: every signal maps to a decision, and no proxy is ever
treated as the truth. Brand health is latent, so you treat every observable
measure as an imperfect proxy for what people remember, feel, prefer, and say.

CONTEXT
Business and offer: {{BUSINESS_AND_OFFER}}
Business model and market: {{MODEL_AND_MARKET}}   (e.g. B2B SaaS, DTC, local; niche or mass)
Brand and distinctive assets: {{BRAND_ASSETS}}    (names, misspellings, handles, taglines, campaigns, key people)
Competitor set: {{COMPETITORS}}
Data sources available today: {{DATA_SOURCES}}     (survey capability, social listening, search console, CRM, web and video analytics, backlink index, reviews)
Key decisions this scorecard must support: {{DECISIONS}}
Maturity and constraints: {{MATURITY}}            (audience size, survey budget, analytics coverage)

If BUSINESS_AND_OFFER, MODEL_AND_MARKET, DATA_SOURCES, or DECISIONS is missing or
vague, do not build yet. Ask me clarifying questions one at a time, waiting for my
answer before asking the next, up to five in total, then build with what you have.
Do not invent a business model or assume data you were not told exists; work only
from the sources listed in DATA_SOURCES.

METHOD

1. Frame the four brand concepts in this business's own language:
   - Awareness: how many people have heard of or recognize the brand.
   - Engagement: how many people actively interact with it.
   - Affinity: how many people like, prefer, or return to it.
   - Sentiment: what people say and how they describe it.
   You MUST cover all four. A small brand can have low reach yet unusually high
   affinity, so awareness alone is never enough.

2. Inventory every observable signal in DATA_SOURCES and map each to exactly one
   primary concept. If a concept has no viable proxy in DATA_SOURCES, do not
   fabricate one: mark it as a coverage gap in the output and name the cheapest
   instrument that would close it. Starting palette:
   - Awareness: unaided category recall, aided recognition, branded search impressions, social amplification rate, reach, broad share of voice.
   - Engagement: branded search clicks, direct, organic, and permissioned email traffic, content completion events, conversation rate, relevant mentions.
   - Affinity: returning visitors, CRM fan and superfan counts, subscriber growth, time spent with brand led content, relevant linking root domains, non brand search click through rate (controlled), followers.
   - Sentiment: open ended and prompted brand attributes, positive, neutral, and negative mention share, topic specific share of voice, review sentiment.

3. Score each candidate from 0 to 2 (weak, plausible, strong) on six axes: concept
   fit, actionability (does a decision follow?), stability (survives small
   denominators and controls?), coverage of the target market, distinctiveness
   (adds a unique signal, not a duplicate), and operational cost. Reject any
   candidate with no plausible decision attached. Reject vanity metrics.

4. Select the scorecard. Start with ONE absolute metric and ONE relative KPI per
   concept, for eight signals total. Add a second pair only when it captures a
   meaningfully different behavior with its own decision. Never exceed two metrics
   and two KPIs per concept. Choose the highest value combination that covers all
   four concepts without overrepresenting one channel, not simply the highest
   individual scores.

5. For each selected signal, build the measurement chain, keeping metric, KPI, and
   goal distinct: state the business outcome in plain language; name the absolute
   metric (a quantity that rises or falls, such as branded impressions); name the
   relative KPI (a rate, ratio, or share, such as share of voice); set a target and
   review date; record what the proxy cannot prove. Then specify the plumbing:
   source system, exact query or segment definition, population covered, refresh
   cadence, baseline period, known confounders, owner, action if it rises, action
   if it falls.

6. Attach decision rules so movements translate to responses:
   - Awareness down and confirmed: do more (expand distribution and exposure).
   - Awareness stable but engagement down: do different (test positioning, message, creative).
   - Engagement up but affinity flat: do bigger (build deeper content, community, repeat experiences).
   - Sentiment turning negative: do other work (read verbatims, group complaints, route product, service, or policy failures to the owning function before touching messaging).
   Add cross metric flags: mentions up while sentiment down may be complaint driven,
   so investigate before celebrating; branded impressions up while branded clicks
   down means checking search result changes and intent mix.

GUARDRAILS
- Ask unaided recall before aided recognition; a brand list contaminates recall.
- Survey two populations: customers for sentiment and recommendation intent, a
  representative target market sample for awareness and recognition.
- Benchmark share of voice against competitors, but for amplification, followers,
  mentions, and conversation, prefer the brand's own historical baseline.
- In small or thinly discussed markets, use self comparison; unaided awareness may
  be near zero even when the business is healthy, so lean on behavioral proxies.
- Keep CRM fan scores separate from sales lead scores; a fan can have no purchase
  intent, and a hot lead little brand affinity. Treat a recommendation score as a
  directional proxy only, and diagnose product and service before blaming brand.
- Never equate impressions with attention, likes with affinity, or comments with
  positive engagement. Flag small denominators; extreme swings may be a few actions.

OUTPUT FORMAT
A. Concept framing: one line each for awareness, engagement, affinity, sentiment.
B. Scorecard table: Concept | Metric | KPI | Target and review date | Source |
   Query or segment definition | Confounders | Owner | Action if up | Action if down.
C. Scoring rationale: one sentence per signal on why it beat the alternatives.
D. Rejected candidates: what you dropped and why (redundant, unactionable, noisy).
E. Dashboard layout: four sections, each showing current value, target, change
   versus prior period, trend line, and a record level drill down, with annotations
   for campaigns, launches, crises, spend changes, and instrumentation changes.
F. First 30 days: baselines to establish, instruments to stand up, review cadence
   (quarterly is a practical starting point).
G. Assumptions to verify: list every assumption you made about data availability,
   tool capability, population size, or survey feasibility, phrased so I can
   confirm or correct each one before an analyst starts the build. Include any
   coverage gaps from step 2 here.

SELF CHECK before finishing:
- Facts to verify: are queries and segments defined outside the dashboard so the
  logic is auditable? Is every survey wave using the same population and wording?
  Are named statistics attached to a real source, or dropped?
- Failure modes to avoid: no concept left uncovered; no single proxy treated as
  conclusive; no tile that produces no decision; paid, technical, and acquisition
  effects not mislabeled as brand; small movements not mistaken for trends; the
  scorecard compact enough that one reviewer can explain any movement and name the
  next action without opening five tools.
If any check fails, revise before presenting.

For the most capable models. Goal and quality bar up front.

You are a senior brand measurement strategist. Design a brand tracking scorecard
that survives scrutiny, then return it.

GOAL
Turn scattered signals into one compact, auditable scorecard where every number maps
to a brand concept, a decision, and an owner. Your first line of output is the
scorecard itself (concept, metric, KPI, owner, action if up, action if down); the
scoring rationale, rejected candidates, dashboard layout, first 30 days, and
assumptions to verify come after.

CONTEXT
Business and offer: {{BUSINESS_AND_OFFER}}
Business model and market: {{MODEL_AND_MARKET}}
Brand and distinctive assets: {{BRAND_ASSETS}}
Competitor set: {{COMPETITORS}}
Data sources available today: {{DATA_SOURCES}}
Key decisions this scorecard must support: {{DECISIONS}}
Maturity and constraints: {{MATURITY}}

PRINCIPLES (non negotiable)
Brand health is latent, so treat every observable as an imperfect proxy, never as
truth. Cover all four concepts (awareness, engagement, affinity, sentiment) in this
business's own language; map each signal to exactly one concept. Cap the system at
one to two metrics and one to two KPIs per concept, adding a second only when it
captures a meaningfully different behavior with its own decision. Every signal names
its source, exact query or segment definition, owner, and a specific action for both
a rise and a fall. Ask unaided recall before aided recognition; survey a target
market sample for awareness and a customer sample for sentiment. Benchmark share of
voice against competitors; benchmark amplification, followers, mentions, and
conversation against the brand's own history. Keep CRM fan scores separate from lead
scores. Attach decision rules so movements translate to responses (awareness down,
expand exposure; engagement down, test positioning; affinity flat, build depth;
sentiment negative, route product or service failures before touching messaging),
plus cross metric flags for mentions up while sentiment down and impressions up
while clicks down.

QUALITY BAR
Excellent output covers all four concepts, treats no single proxy as conclusive,
lists rejected candidates with reasons so the scorecard is defensibly compact,
flags any concept with no viable proxy as a coverage gap plus the cheapest
instrument to close it, and stays compact enough that one reviewer can explain any
movement and name the next action without opening five tools.

BOUNDARIES
Do not invent data, statistics, or a business model; work only from DATA_SOURCES.
Do not equate impressions with attention or likes with affinity. Do not pad with
generic advice or build tiles that produce no decision. If BUSINESS_AND_OFFER,
MODEL_AND_MARKET, DATA_SOURCES, or DECISIONS is missing or vague, ask one focused
question instead of guessing.

Five lines. Speed over rigor.

Build a compact brand tracking scorecard for {{BUSINESS_AND_OFFER}} in {{MODEL_AND_MARKET}},
using only {{DATA_SOURCES}} and supporting {{DECISIONS}}.
Cover all four concepts (awareness, engagement, affinity, sentiment) with one metric and one KPI each.
For every signal give the source, exact query or segment, owner, and the action to take if it rises or falls.
Quality bar: no proxy treated as truth, and a reviewer can explain any movement and name the next action without opening other tools.