Marketing metrics the managers guide to measuring marketing performance

1. 1Fix the unit of analysis first. The passages tie financial performance to three outcomes: profitability, return on investment (Marketing ROI, MROI), and customer equity. Write one line for each stating what counts as the firm's bottom-line result, so every later metric maps to one of them.
2. 2Size the spend before judging the return. The passages say marketing, with sales force costs, is 10 percent or more of operating budgets at many public firms. Pull total marketing plus sales force cost for the period as the denominator for every ROI statement.
3. 3Choose KPIs with a stated link to profit. Per the Google foreword, set key performance indicators, then measure against them on a regular schedule. Reject any KPI (awareness, shipments, leads) that you cannot connect to profit or customer equity in one sentence.
4. 4Measure profitability at the customer-account level, not only in aggregate. Quelch calls for analyzing the profitability of individual customer accounts. Where the business sells through retailers, use SKU-level profit measures such as GMROII and direct product profitability (DPP), which combine movement rates, gross margins and inventory cost.
5. 5Quantify how marketing spend contributes to profits (MROI). State the spend, the incremental outcome attributed to it, and the method of attribution. Name any competitor reaction or cross effect you net out. The residual price elasticity logic (own elasticity + competitor reaction × cross elasticity) shows that a price move's net volume effect differs from the naive own-price effect.
6. 6Triangulate before reporting a result. Compare survey or attitude data with sales revenue, shipments and retail sales, since these can move in different directions and timing differs (toy shipments precede advertising-driven purchase). Only report a financial outcome that holds across at least two sources.
7. 7Separate leading from lagging indicators. Treat pipeline conversion (percentage of leads who ultimately buy) and attitudes as leading. Treat profit, ROI and customer equity as lagging. Report them in separate columns so a lead indicator is never presented as a financial result.
8. 8Run a multi-metric check against single-metric flattery. The passages warn that one metric can look very good while the company's true value is destroyed. Review several metrics even though the picture is less precise (less reliable), because you are less likely to miss a major problem (more valid). Then expand programs that work and pull back on those that do not.

The passages supply no firm-level financials, so this walks through their logic rather than invented numbers. Procter & Gamble's CMO said marketing is a $450 billion industry whose decisions get less data and discipline than $100,000 decisions elsewhere in the business. Take a firm in that position that reports only awareness gains. Step one: record total marketing and sales force cost as the investment base. Step two: break profitability out by customer account, as Quelch urges, to see which accounts the spend actually earns on. Step three: triangulate. If awareness rose but retail sales and shipments did not, treat the awareness gain as unconfirmed rather than as a return. Step four: if a price cut was part of the program, net out the competitor's reaction using E1 + E2×E3, so the volume gain is not overstated. Step five: report MROI beside the lagging profit figure and cut or expand programs accordingly. The boards Quelch describes want exactly this: a marketer who speaks the language of productivity and ROI and accepts accountability.

• If a metric improves but you cannot trace it to profit, ROI or customer equity, then label it a leading indicator and withhold any financial-performance claim.
• If survey attitudes, shipments and retail sales disagree, then do not report ROI until you can explain the timing gap (for example, shipments ahead of advertising and purchase).
• If one metric looks excellent, then test additional metrics, accepting less precision, because a single flattering metric can hide destroyed value.
• If a price change drives your volume result, then net out competitor reaction and cross elasticity (E1 + E2×E3) before crediting the program.
• If a program is working against its KPIs, then expand it. If it is not, then pull back, and decide on data rather than opinion.

• Reporting top-line sales as financial performance. The tell: no profit, margin or cost figure appears anywhere in the report.
• Using aggregate profitability only. The tell: you cannot say which customer accounts or SKUs earn money, so unprofitable accounts hide inside a healthy total.
• Crediting a program with all observed change. The tell: competition and cross effects are never mentioned, and the marketers say outside factors make results impossible to monitor.
• Treating the hierarchy of effects as one-way, so awareness is assumed to cause usage. The tell: the causal flow may run the other way (owners tend to like the brand they own), yet the report assumes a single direction.
• Relying on a single metric. The tell: one number is celebrated while no second metric has been checked for a hidden problem.

• ▸Period and program: name the campaign or budget line and the dates covered.
• ▸Total marketing + sales force cost: enter the full spend, the denominator for all ROI claims.
• ▸Profit outcome(s): state profit, margin or customer-account profitability attributed to the program, with the attribution method.
• ▸MROI statement: write incremental profit contribution relative to spend, in one line.
• ▸Customer-account / SKU profitability: list the top and bottom accounts or SKUs (GMROII or DPP where retail applies).
• ▸Leading indicators: record lead-to-sale conversion, awareness or attitude data, kept separate from financial results.
• ▸Triangulation check: note whether survey data, shipments and sales agree, and explain any timing gap.
• ▸Competitor reaction / cross-effect adjustment: record own, reaction and cross elasticities, and the residual E1 + E2×E3.
• ▸Second-metric validity check: name at least one additional metric reviewed for hidden problems and what it showed.
• ▸Decision: expand, hold or pull back, with the data point that justifies it.

1. 1Sort every lever into one of two kinds of relationship before you fund it. Identity relationships are arithmetic, such as revenue less costs equals profit. Empirical relationships must be estimated, such as the effect of advertising, sales force, or trade promotion spending on unit sales. Draw the Figure 13.4 chain for your business: lever → unit sales (empirical) → net profit (identity). Estimate only the empirical links.
2. 2Build a line-item inventory of the whole marketing budget. Marketing spending usually includes advertising and non-price promotion. It sometimes includes sales force spending and may include price promotions. Fix the definition once and keep it constant, or comparisons across periods and companies break.
3. 3Classify each line as fixed, variable, or stepped. Variable selling costs usually track revenue, so express them as a percentage of revenue rather than a cost per unit. Treat stepped costs, such as an agency contract covering up to three campaigns a year, as fixed, provided you know where the step falls. Rebates and quota bonuses are hard to model, so list them separately.
4. 4Decide expense versus investment for each line. Amortize true investments, such as a special allowance for financing receivables from new distributors. These raise working capital rather than current-period cost. Treat long-term-impact advertising as a period expense, which is the book's recommendation.
5. 5Set a baseline before you judge any lever. Incremental sales equal total sales minus baseline sales. Use a control group or experiment if you can. Otherwise use prior sales adjusted for expected growth and seasonality, or a regression that includes controllable and uncontrollable factors such as competitor spending. In a regression, the intercept is often read as the baseline.
6. 6Model the lever's path to volume with the multiplicative identity: Forecast Unit Sales = Consumer Prospects × Awareness × Availability × (Trial Rate × Trial Units + Repeat Rate × Repeat Units). Tie each lever to the term it moves. Advertising moves awareness, distribution and channel spend move availability, and promotion and pricing move trial and repeat. Source each input from a simulated test, survey, management judgment, or an empirical model.
7. 7Check the effect on the denominator of return. Compute ROI as net profit ÷ investment. Marketing decisions change net profit and also inventories, receivables, and plant. Use a period-end view where seasonal swings would make an annual average hide the real asset load.
8. 8Stress-test the budget for risk. Variable selling costs stay lower if sales miss, so they are less risky than fixed ones. Run the plan at below-forecast sales and see how much cost stays committed. Track marketing as a percentage of sales (marketing spending ÷ revenue), but read it against what is normal for the product, strategy, and market.

The source gives formulas but no full case, so the figures here are hypothetical and show only the mechanics. A firm contracts an agency for up to three campaigns a year. The contract is a stepped cost, treated as fixed up to campaign three, and a fourth campaign adds an incremental cost. The firm also pays a sales commission as a percentage of revenue, which is variable. Say it launches a product with a $4M marketing budget. A line-item review finds $3M fixed or stepped and $1M variable. If sales come in 30% under forecast, the variable $1M shrinks automatically and the $3M does not. That is the budgeting risk the book describes. To judge the advertising, the firm compares post-campaign sales with a baseline built from last year's sales adjusted for growth and seasonality. The difference is incremental sales, and incremental sales × margin, minus the spend, gives the profit effect. It then checks the forecast identity: if awareness rose but availability stayed low, the multiplicative form shows the gain being capped by distribution. The sales force follows the same logic. Break-even employees equal revenue × margin net of commission ÷ cost per staff member.

• If a cost moves with revenue, model it as a percentage of revenue and treat it as lower-risk. If it holds regardless of sales, treat it as fixed risk and size it to the downside forecast.
• If an expenditure builds working capital, such as distributor financing or inventory, amortize it and include it in ROI's investment base. If it is advertising, expense it in the period.
• If you cannot run a control experiment, build the baseline from growth- and season-adjusted history, and use regression with competitor spending included when history is noisy.
• If one term of the unit-sales identity (awareness, availability, trial, repeat) is far below the others, fund the weak term before adding spend to the strong ones. The multiplication makes the weakest term the cap.
• If you compare marketing % of sales across firms or periods, first confirm that all firms define marketing spending the same way (sales force and price promotions in or out). Then judge the figure against product, strategy, and market norms, not a universal target.

• Treating accounting identities as things to estimate, or empirical effects as things to calculate. The tell is a model with regression coefficients on profit arithmetic, or a flat assumed lift on advertising.
• Claiming lift without a defensible baseline. The tell is incremental sales that equal total sales, or a result that moves when you change the baseline period.
• Budgeting all marketing as variable. The tell is a plan that assumes costs scale down when sales miss, while contracts, agency retainers, and stepped commitments stay in place.
• Ignoring the investment side of ROI. The tell is a promotion or channel program credited with profit gains while inventory and receivables rise unmentioned, or annual averages that hide seasonal peaks in assets.
• Comparing marketing spending ratios without a common definition. The tell is a peer benchmark that includes sales force and price promotions on one side but not the other.

• ▸Lever and budget line: name the lever (advertising, sales force, trade promotion, price promotion, channel) and the budget line it sits in.
• ▸Cost behavior: mark it fixed, variable (as % of revenue), or stepped, and note where any step or bonus threshold falls.
• ▸Expense or investment: state the treatment, and give the reason if it affects working capital such as receivables or inventory.
• ▸Baseline sales: state the method (control group, adjusted history, or regression) and the period used.
• ▸Expected incremental sales: enter total minus baseline, and name the term it moves in the unit-sales identity.
• ▸Identity-to-profit link: enter margin × incremental sales minus the spend, using only accounting identities.
• ▸Marketing as % of sales: enter spending ÷ revenue, with what is included and the norm you compare it to.
• ▸Downside test: show committed (fixed and stepped) cost if sales miss forecast.
• ▸ROI check: show net profit ÷ investment, including any change in inventory or receivables caused by the lever.

1. 1Set the decision first. Hearts-and-minds metrics track trends in attitudes and behaviors, so name the question (is awareness the bottleneck, is belief, is satisfaction?) before you write any survey item.
2. 2Map the metrics to the Hierarchy of Effects: awareness, then attitudes and beliefs, then purchase and experience, then loyalty. Pick at least one metric per stage so you can see where customers stall. Treat the sequence as an assumption, not a law; the sources note it is often violated and can be reversed.
3. 3Write the Awareness, Attitudes, and Usage (AAU) questionnaire using the passage's question types. Awareness and knowledge: 'Have you heard of Brand X?' and 'What brand comes to mind when you think "luxury car?"' Attitudes and intentions: 'Is Brand X for me?', a 1-to-5 rating such as 'Is Brand X for young people?', and the strengths and weaknesses of each brand. Usage: purchase habits.
4. 4Fix the prompted versus unprompted question for every awareness and knowledge item and record it in the metric definition. Unprompted recall (top of mind: first brand to consider) and prompted recognition ('Have you heard of Brand X?') measure different things and cannot be compared. Top of mind can also be skewed by recent advertising or experience.
5. 5Survey large samples of households or business customers. Add the 'who' cuts the passage names: heavy versus light category users, geography, demographics, psychographics, media usage, and purchase of other products. Use them to see who accepts and who rejects the brand.
6. 6Add satisfaction and willingness to recommend on a 1-to-5 scale. Report the share meeting your satisfaction goal, and read the extremes separately. 5s tend to return and may evangelize; 1s are unlikely to return and may talk negatively to prospects.
7. 7Attach a comparator to every reading: prior periods, different markets, or competitors. A level without a benchmark is not interpretable.
8. 8Link the perception readings to behavior (sales, share, retention) and watch for dips. A falling satisfaction rating warns of later problems in sales and profitability, so escalate before the sales numbers move.

The Wal-Mart passage in the source shows why perception data needs behavioral data beside it. The ShopperScape findings report that 3 of 5 consumers shopped for gifts at Wal-Mart in the holiday season, that U.S. households buy 22% of their groceries there, and that a quarter of shoppers are spending more of their clothing budget there than a year ago. Those figures are share of wallet, a behavioral outcome. They do not say why shoppers moved. An AAU study on the same shoppers would ask: 'Have you heard of Wal-Mart?' (awareness), 'What are the strengths and weaknesses of each retailer?' (beliefs), and a 1-to-5 satisfaction and willingness-to-recommend rating (affect). Compare the results with the prior period and with competitors. If the quarter increasing clothing spend rate satisfaction at 5 and the rest sit lower, the attitude data explains the behavior. If clothing spend rises while satisfaction falls, you have a leading-indicator warning that the share gain may not hold. This is an illustrative use of the passage's example; the source gives no attitude figures for Wal-Mart.

• If awareness (prompted) is low, then fix reach and advertising before working on beliefs; if prompted is high but unprompted or top of mind is low, then the problem is saliency, not recognition.
• If awareness is high but 'Is Brand X for me?' and attribute-belief ratings are weak, then the problem is perception, not exposure. Change the message or product, not the media weight.
• If satisfaction ratings dip, then treat it as a warning for future sales and profitability and investigate now, even if current sales and share look healthy.
• If you compare any attitude metric across periods, markets, or competitors, then keep question wording and the prompted/unprompted format identical; otherwise do not compare.
• If you need a loyalty read, then do not rely on share of requirements alone. The authors call it a less-than-perfect measure of loyalty, so add satisfaction and willingness to recommend.

• Mixing prompted and unprompted awareness: the trend line jumps when the questionnaire changes, not when the market does. Tell: awareness moves sharply with no change in advertising or distribution.
• Reading a single number without a comparator: a satisfaction or awareness score is quoted as 'good' or 'bad' with no prior period, market, or competitor beside it. Tell: nobody can say whether the number is improving.
• Treating the Hierarchy of Effects as a strict sequence: you assume customers cannot buy before forming attitudes. Tell: usage data shows buyers whose awareness or attitude scores are low, or purchase preceding belief.
• Averaging satisfaction and missing the extremes: the 1s (who will not return and may spread negative comments) and the 5s (who may evangelize) are hidden by the mean. Tell: the average is stable but complaints or referrals are shifting.
• Using behavior as a proxy for attitude: a share-of-wallet or share-of-requirements figure is read as loyalty or liking. Tell: the report explains what customers did but cannot say why.

• ▸Decision this tracker informs: one line naming the choice (for example, fix awareness vs. fix beliefs).
• ▸Stage and metric: for each Hierarchy of Effects stage (awareness, knowledge, beliefs, intentions, usage, loyalty), name the metric you will track.
• ▸Question wording and format: write the exact question and mark it prompted or unprompted; do not change it between waves.
• ▸Sample and segments: sample size and the 'who' cuts (heavy/light users, geography, demographics, media usage, other purchases).
• ▸Comparator: choose prior period, different market, or named competitor, and record the benchmark value.
• ▸Satisfaction (1-5) and willingness to recommend: record the share meeting your satisfaction goal and the counts at 5 and at 1.
• ▸Current reading vs. comparator: enter the value and the gap for each metric.
• ▸Behavioral cross-check: the sales, share, or retention figure to set beside each attitude reading.
• ▸Action trigger: the dip or gap that starts an investigation, the owner, and the date.

1. 1List the behaviors you will track and define each in one line: initial trial, repeat purchase, purchase frequency, purchase quantity, brand switching, and online clickthrough and conversion. Each needs a written definition and a time window before you collect anything. Without them, period-to-period comparisons mean nothing.
2. 2Split every behavior by customer versus non-customer. The source warns that their responses may be very different and that causality between awareness, attitudes, and usage runs both ways: people who own a brand may be predisposed to like it. Never pool the two groups into one average.
3. 3Measure loyalty as several separate dimensions, not one score. Rarely switching, willingness to pay a price premium, and willingness to recommend can each be present or absent. Record them in separate columns.
4. 4Test whether low switching is real preference or inertia and habit. Compute willingness to search: the percentage of customers who leave the store without the product if their favorite brand is unavailable. The balance of the population is willing to substitute. A high rate suggests loyalty. A low rate with low switching suggests inertia.
5. 5Benchmark only against category-specific norms. Loyalty to matches is not comparable to loyalty to baby formula. Cut results by demographic group too, because older consumers show the highest loyalty rates.
6. 6Before reading any change, check for methodology shifts: mail versus telephone, paid versus unpaid respondents, changed question framing. If the data jumps, rule these out first and adjust the norms you use to call a result good or bad. Then separate seasonality and random noise from signal.
7. 7Triangulate survey-reported behavior against sales revenue, distributor and retail sales, and company shipments. They can move in different directions, and the gap often reveals category dynamics. Retailer shipments can run well ahead of consumer purchase, as with toys.
8. 8Treat recalled behavior as weak evidence. Consumers often don't know what they will do in new circumstances and recall past behavior poorly, especially for low-involvement items. Prefer observed purchase and online data over stated intent.
9. 9Use implicit and neuro measures, such as response latencies and the Implicit Association Test, only as an early-stage aid. The source still recommends traditional metrics to monitor campaign performance.

The source's pricing example shows how purchase quantity by variant drives a behavior-based average. A beverage line sells 10k units of one variant at $8, 40k units of another at $12, and 2k units of a third at $20, for 52k units in total. The unit shares are 19.23%, 76.92%, and 3.85%. The weighted average price per unit is ($8 × 0.1923) + ($12 × 0.7692) + ($20 × 0.0385), which comes to about $11.54. A simple average of the three list prices would be $13.33, so the customer mix pulls the realized price down. Read as behavior, 77% of units go to the $12 variant, so shifts in what customers choose move the average without any price change. Track unit mix per period and per customer group. If the average falls while list prices hold, customers are trading across variants, not paying less.

• If customers rarely switch but willingness to search is low, then call it inertia or habit, not loyalty, and don't assume a price premium or advocacy.
• If a tracking metric moves sharply between periods, then check for changes in survey method and seasonality before acting on it.
• If survey attitudes and sales or shipments diverge, then trust neither alone. Reconcile them against retail sales and shipments to find where the timing or channel gap sits.
• If you compare loyalty or repeat rates across products, then stop and re-benchmark within each category's own norms.
• If a neuro or implicit measure suggests a behavioral effect, then confirm it with traditional campaign metrics before reallocating spend.

• Pooling customers and non-customers. The tell: attitude scores look high and seem to explain usage, but the causal direction is untested.
• Equating low switching with loyalty. The tell: customers stay but won't pay a premium, won't recommend, and accept substitutes when the brand is out of stock.
• Comparing loyalty across categories. The tell: a matches brand is judged against a baby formula brand, or the reverse.
• Reading noise as a trend. The tell: a swing coincides with a switch from mail to phone, from paid to unpaid respondents, or with a season.
• Relying on stated future behavior. The tell: survey intent differs from observed purchases, or respondents can't recall low-involvement purchases.

• ▸Behavior and definition: name the behavior (trial, repeat, frequency, quantity, switching, clickthrough, conversion) and its exact definition and time window.
• ▸Segment: customer or non-customer, plus the demographic cut.
• ▸Observed value vs. prior period: the measured figure for this period and the last.
• ▸Method check: any change in survey mode, question framing, or respondent type, and the adjustment made to the norm.
• ▸Seasonality or noise call: signal, seasonal, or noise, with a one-line reason.
• ▸Willingness to search: percentage who leave without the product when their favorite is unavailable, and the percentage who substitute.
• ▸Loyalty dimensions: rarely switches, pays a premium, recommends. Mark each yes or no.
• ▸Category norm: the benchmark for this category only, and the gap to it.
• ▸Triangulation: matching sales, shipment, or retail data, and whether it agrees with the survey.

1. 1Split the construct into four named concepts before choosing any number: breadth of distribution, depth of distribution, in-store execution, and sales force productivity. The source warns that concepts, operational definitions, and metrics are easy to blur, so write them in separate columns.
2. 2Write an operational definition for each concept, a precise numerical statement of what you mean. For example, 'our product is available' becomes a stated measure of availability, not a feeling. The source's own pairing is distribution tracked by All Commodity Volume (ACV) and Total ACV. Use the Chapter 6 definitions of these for your calculation.
3. 3Name the decision each metric serves and the person who can act on it. The source's test is that a metric is valuable only if it improves a decision. A junior marketer who cannot move brand value can still move whether the product is on retailers' shelves, so distribution measures belong on his dashboard. Drop any metric nobody can act on.
4. 4Assemble a small dashboard rather than a single number. Pair a coverage metric (ACV or Total ACV) with sales force measures (organization, performance, compensation) and logistics measures from Chapter 6. Read them together, because the source says no single metric is likely to be perfect and recommends triangulating.
5. 5Decide how precisely each metric needs measuring by checking how much it swings. A volatile metric, like sales of a fashion item, repays a precise reading. A stable one, like safety matches, does not, because last year's figure already gives the same inventory decision. Spend measurement effort only on the volatile ones.
6. 6Check validity before you report: does the number translate the idea you wrote down? Ask whether the metric measures availability and sales force effectiveness as you defined them, or only something nearby, such as activity.
7. 7Compare like with like across regions and time periods. The source cites comparability as a core purpose of metrics, so use the same definition in every region and period. Then attach financial consequences, since sales force costs are a material share of operating budgets and marketers are held accountable for financial results.
8. 8Review on a fixed cadence. Where a distribution or sales force metric moves, trace the cause by decomposing it into parts, as the source's 'system of metrics' approach recommends, and act on the weakest component.

The source contrasts two marketers, and the contrast shows how to scope this construct. A CMO may find brand value invaluable for arguing a budget increase with C-suite colleagues. A junior marketer cannot move brand value but can influence whether the product is on retailers' shelves, so distribution measures (ACV, Total ACV) are what he should own. Apply the volatility test to the same shelf question. For a fashion item with unpredictable consumer reaction, a precise sales reading is 'exceptionally valuable' for stock planning, so you track sell-through at fine grain. For safety matches, where sales are stable, a precise reading would not change the inventory order, so you do not spend effort refining it. The result is a division of labour: the junior marketer tracks shelf presence on a dashboard with a precision tuned to each product's volatility, while the CMO tracks the brand-level metrics. Each number is justified by the decision it informs.

• If a metric does not change any decision the owner can make, remove it from the dashboard; the source defines a metric's value by its ability to improve decisions.
• If a product's sales are volatile (fashion-like), invest in precise, frequent distribution and sales data; if they are very stable (matches-like), accept last year's estimate.
• If one metric (e.g., ACV) looks healthy, check a second lens (sales force performance or logistics) before concluding; no single metric is likely to be perfect.
• If your team uses one word for the concept and the metric (e.g., 'distribution'), stop and write the operational definition first, then the metric name.
• If you must defend sales force or channel spend, express it in financial terms, because marketers are now accountable for the financial ramifications of their decisions.

• Naming the concept as the metric: the team says 'distribution' when it means ACV. The tell is that two people quote different numbers under the same name.
• Reporting a single coverage number as proof of channel health. The tell is strong availability metrics alongside weak sales force or in-store results nobody has examined.
• Measuring precisely what never changes. The tell is a polished report on a stable item whose reading would not alter the inventory order.
• Missing data treated as measured data. The source warns that many metrics rely on approximate, incomplete, or unavailable data. The tell is a figure presented without any note on its data gaps.
• Resisting accountability by blaming outside factors such as competition. The tell is that nobody can state the financial impact of channel and sales force decisions.

• ▸Concept: breadth, depth, in-store execution, or sales force productivity (pick one per card).
• ▸Operational definition: state in numerical terms exactly what you mean by the concept.
• ▸Metric(s): e.g., ACV, Total ACV, or a Chapter 6 sales force or logistics measure.
• ▸Decision it informs: the specific action the metric should change.
• ▸Owner: the person who can actually move this number.
• ▸Volatility: does the metric swing widely or stay stable? Set precision and frequency accordingly.
• ▸Companion metric: the second lens you triangulate against.
• ▸Data gaps: what is approximate, incomplete, or unavailable, and how it is handled.
• ▸Financial link: how a change in this metric ties to cost or profit.

1. 1Define the market before you measure it. Write down the addressable market, the competitors you are judged against, and the period. The source says marketers must understand their addressable markets quantitatively, so market performance has no meaning without this boundary.
2. 2Pick a small set of key performance indicators for market position, following the Google rule: determine your KPIs, then measure against them on a regular schedule. Draw them from the Chapter 2 cluster the book names: customer perceptions (hearts and minds), market share (markets), and competitive analysis. Use one or two from each so you see both the attitude and the outcome.
3. 3Write each metric's construction on one line: what it quantifies, the data source, and the formula. The book defines a metric as a measuring system that quantifies a trend, dynamic, or characteristic. Two teams using the same label with different constructions cannot compare observations across regions and time periods.
4. 4Measure against competitors, not only against your own history. Marketers who say competition makes results hard to monitor are describing the reason to track competitive position, not an excuse to skip it. Record your result and the competitive context together, so a flat number in a shrinking market reads differently from a flat number in a growing one.
5. 5Separate customer-behavior drivers from market outcomes. Market performance is the aggregate result of customer behaviors and channel performance. Log the upstream behavior and channel measures next to the position metric, so a change in position traces to a cause (customer or channel) instead of a guess.
6. 6Review the metrics as a set on a dashboard, not one at a time. The authors recommend viewing market dynamics from multiple perspectives and triangulating, because each metric has pros, cons, and nuances. Mark any metric that disagrees with the others and investigate before acting.
7. 7Close the loop with the Google rule: optimize and expand programs that are working, and pull back those that are not. Tie each position metric to a named program and an owner, so a movement leads to a resourcing decision.
8. 8Link the position metrics forward to financial results. The book places market performance as an intermediate link between marketing activity and finance (Chapters 3 and 11). Record how a share or perception change is expected to show up in revenue and profit, then check whether it did.

The passages give no market-share dataset, so this example uses the evidence they do contain. Procter & Gamble's Chief Marketing Officer says marketing is a $450 billion industry in which decisions are made with less data and discipline than $100,000 decisions elsewhere in the business. Measuring market position is the repair the authors prescribe: gather and analyze basic market data, measure the core factors that drive the business model, and optimize resource allocation across increasingly fragmented media. Apply it to a brand. Define the addressable market and named competitors, then set three KPIs, one each for perception, market share, and competitive standing. Review them quarterly on one dashboard. Where a program's metrics rise against competitors, expand it, and where they stall, pull back, as Google's "data beats opinion" rule directs. The test is whether the dashboard changes a spending decision. If the numbers are collected but no decision moves, the measurement has not yet become accountability.

• If a market-position metric moves and you cannot name the customer-behavior or channel driver, then hold the budget decision and trace the driver first.
• If two metrics on the dashboard disagree (for example, perceptions improve while share falls), then triangulate with a third view before acting on either.
• If a program's position metrics keep improving against competitors, then expand it. If they do not, then pull back, in line with the optimize-and-expand rule.
• If someone argues that competition makes results impossible to monitor, then measure competitive position directly and report results relative to it.
• If a metric has no written construction (formula, data source, period), then do not compare it across regions or time until one exists.

• Reporting your own numbers with no competitive frame. The tell: a share or sales figure is presented with no market or competitor context, and explanations lean on "the market was tough."
• Tracking one metric as if it were market performance. The tell: the dashboard shows only share or only awareness, and no one can say how the others moved.
• Inconsistent metric construction. The tell: the same label gives different values from different teams, or trend lines break when the data source changes.
• Measuring without deciding. The tell: the dashboard is reviewed regularly but no program has been expanded or cut because of it.
• Treating market position as the end result. The tell: no stated link from position changes to revenue and profit, so marketing spend cannot be defended to finance or the board.

• ▸Market definition: state the addressable market, segment, geography, and period being judged.
• ▸Competitor set: list the named competitors your position is measured against.
• ▸Perception metric (hearts and minds): name the customer-perception measure, its construction, and its data source.
• ▸Market share metric (markets): name the share measure, its formula, and its data source.
• ▸Competitive position metric: name how you compare with each competitor, and the comparison basis.
• ▸Behavior and channel drivers: list the customer-behavior and channel measures expected to explain movement in the position metrics.
• ▸Current vs. prior period reading: record each metric's value now, last period, and the competitor context.
• ▸Disagreements across metrics: note any metrics that conflict and the third view used to triangulate.
• ▸Program decision and owner: state which program to expand, hold, or pull back, and who owns it.
• ▸Expected financial link: state how the position change should appear in revenue and profit, and when you will check it.