Marketing teams can measure hundreds of things. Whether a given measurement is useful depends entirely on the business question it has been asked to answer.
Reach and impressions describe media delivery. Click-through rate describes response to an impression. Conversion rate describes movement from an eligible population to a defined action. Customer acquisition cost and customer lifetime value address the economics of winning and keeping customers. None of them substitutes for the others, and most of the reporting arguments I sat through over nine years on the account side came down to two people using one word for two different calculations.
This guide organises commonly used marketing metrics by funnel stage. For every KPI it sets out what the metric means, how to calculate it, the question it answers, the inputs it requires, its main limitations, and what to check before you interpret a change. For guidance on selecting KPIs, setting targets, choosing attribution methods and establishing reporting processes, see our guide to building a marketing measurement framework.
Marketing metrics by funnel stage: quick reference
| Funnel stage | Principal metrics |
|---|---|
| Awareness | Share of voice, branded-search growth |
| Reach and attention | Reach, impressions, frequency |
| Acquisition | Click-through rate, cost per click |
| Engagement | Engagement rate |
| Lead generation | Cost per lead, lead-to-customer conversion rate |
| Conversion | Conversion rate |
| Revenue | Revenue per visitor, average order value, return on ad spend |
| Retention | Repeat-purchase rate, retention rate, churn rate |
| Customer lifetime value | Customer lifetime value, CLV-to-CAC ratio |
| Efficiency | Customer acquisition cost, marketing efficiency ratio |
| Profitability | Contribution margin, marketing return on investment |
This mapping is diagnostic rather than universal. A free-trial registration is a lead-generation event for one company and a conversion event for another. An online purchase is the final conversion for a retailer and the opening move of a subscription relationship elsewhere.
Define the event, the eligible population, the reporting period and the data source before comparing any KPI against anything.
Awareness metrics
Awareness metrics indicate whether a brand is maintaining a visible presence in its market. On their own they do not establish that awareness changed, or that advertising caused additional sales.
Share of voice
Definition and formula
Traditional advertising share of voice compares a brand’s media expenditure with total media expenditure in a defined category, channel, market and period:
Share of voice = Brand media expenditure ÷ Total category media expenditure × 100
Nielsen defines share of voice in expenditure terms and specifies that its scope must state the category, market, channel and point in time. Nielsen also makes the limitation explicit: share of voice does not measure the impact of a campaign, only whether the campaign has the means to compete in the first place (1).
Some digital teams use the same term for a brand’s share of impressions, mentions or search visibility. Label those variants explicitly. They are not interchangeable with expenditure-based share of voice.
Funnel stage
Awareness.
Business question
How large is the brand’s advertising presence relative to the competitors included in the analysis?
Required inputs
- brand media expenditure;
- category media expenditure;
- a defined competitor set;
- channel, geography and reporting period;
- a consistent gross or net expenditure methodology.
Limitations
Reliable category-level expenditure is often unavailable for auction-based digital media and walled platforms. Cross-channel totals may combine data collected under incompatible methodologies.
Share of voice indicates competitive weight. It does not reveal whether anyone noticed the advertising, understood it or acted on it.
Interpretation warning
Share of voice can move because competitors changed their spending while your own budget stayed flat. Always examine the numerator and the denominator before concluding anything about your own performance.
Branded-search growth
Definition and formula
Branded-search growth tracks the change in searches or organic search activity containing the brand name and its agreed variants:
Branded-search growth = (Current-period branded search volume − Previous-period branded search volume) ÷ Previous-period branded search volume × 100
Google Search Console reports clicks, impressions and click-through rate at query level and supports query filtering, which makes it possible to define a branded-query set and track it consistently over time (2).
Name the metric after its actual input:
- branded-search impressions growth;
- branded organic clicks growth;
- branded query volume growth;
- branded share of search.
These are related measures. They are not the same measure.
Funnel stage
Awareness and active consideration.
Business question
Is demand involving the brand name rising or falling?
Required inputs
- a maintained list of brand names, misspellings and product brands;
- branded query impressions, clicks or external search-volume data;
- current and comparison periods;
- consistent geography, device and search-property filters.
Limitations
Search demand responds to offline campaigns, news coverage, promotions, distribution changes, service incidents and seasonality. Search Console reports activity that produced an eligible impression for your site, not the whole market’s search volume (3).
Interpretation warning
Do not read every branded search as positive awareness. Volume also rises during complaints, recalls, recruitment interest and customers simply trying to log in.
Reach and attention metrics
Reach, impressions and frequency describe delivery. They should never be presented as evidence of attention.
Reach
Definition
Reach is the estimated number of distinct people or accounts exposed to an advertisement at least once in a defined period. Google describes unique reach as the number of people shown an ad (4). Meta separates reach from impressions on the basis that one reached account can generate several impressions (5).
Reach is normally supplied by the advertising or audience-measurement platform rather than calculated from raw impression logs.
Funnel stage
Reach and awareness.
Business question
How many distinct members of the measured audience had an opportunity to encounter the campaign?
Required inputs
- platform or audience-measurement reach estimate;
- campaign and reporting period;
- target-audience definition;
- geography and device scope.
Limitations
Reach is usually modelled or deduplicated using platform-specific identity signals. Adding reach figures from several platforms double-counts anyone exposed in more than one place.
Interpretation warning
Reach is not the number of people who paid attention, remembered the message, or belonged to the audience you intended to buy.
Impressions
Definition and formula
Impressions count how many times an advertisement or content item was served or displayed under a platform’s counting rules (6). Meta describes impressions as how often ads were on screen for the target audience (7). One individual can produce many impressions.
No further calculation is normally required:
Impressions = Count of qualifying ad displays
Funnel stage
Reach and awareness.
Business question
How much advertising delivery did the campaign generate?
Required inputs
- impression count from the relevant platform or measurement provider;
- placement and format definitions;
- reporting period;
- invalid-traffic and filtering rules where available.
Limitations
An impression is not necessarily viewable, attended to or remembered. Counting rules differ across search, social, display, connected television and everything else.
Interpretation warning
Impressions rise when budget rises, when media prices fall, when targeting broadens, and when the same people are served repeatedly. Read reach and frequency next to them.
I want to be blunt about why this section matters more than its position in the funnel suggests. Delivery metrics are the easiest numbers in marketing to present impressively, and that makes them the ones most likely to survive on a slide long after they have stopped meaning anything. In my years running enterprise accounts, the campaign types brands asked for most were not the ones with the best economics. They were the ones that looked most advanced when presented internally: the clever tab-title trick, the social proof badge, the personalised homepage slider. Those requests were driven by how well the work marketed inside the company, not by what the numbers said. Delivery metrics are the fuel for that, because a large impression count is impressive to almost everyone and diagnostic to almost no one.
Frequency
Definition and formula
Frequency is the average number of impressions delivered per reached person:
Frequency = Impressions ÷ Reach
Google defines frequency as how often people were shown an ad, and calculates on-target frequency by dividing on-target impressions by on-target reach (8).
Funnel stage
Reach and attention.
Business question
How concentrated or repetitive was delivery among the audience actually reached?
Required inputs
- impressions;
- deduplicated reach;
- consistent audience, channel and time period.
Limitations
Frequency is an average, and averages conceal shape. A campaign averaging four exposures per person contains people who saw it once and people who saw it twenty times.
Cross-platform frequency cannot be derived by summing platform totals unless the audience has been deduplicated first.
Interpretation warning
Neither a high nor a low value is universally good. The right frequency depends on objective, creative, purchase cycle, audience size, media environment and the period measured.
Acquisition metrics
Acquisition metrics measure whether delivered media and search visibility generate visits and other initial responses. They say nothing yet about lead quality, customer value or incremental demand.
Click-through rate
Definition and formula
Click-through rate is the proportion of impressions that produced a click:
CTR = Clicks ÷ Impressions × 100
This is the formula Google Ads uses in its own definition (9).
Funnel stage
Acquisition.
Business question
How often did an impression lead to a click?
Required inputs
- clicks;
- impressions;
- consistent campaign, format, placement and reporting period.
Limitations
A click tells you nothing about what happened after arrival, whether a conversion followed, or whether the visitor was a plausible customer.
CTR responds to ad position, format, audience, targeting, brand familiarity, creative, offer and each platform’s definition of a click. Comparisons between unlike channels are rarely meaningful.
Interpretation warning
A higher CTR can accompany worse business performance when the messaging attracts curiosity instead of qualified demand. Read it with post-click engagement, conversion and customer-quality measures.
Cost per click
Definition and formula
Average cost per click is the total cost of clicks divided by the number of clicks:
Average CPC = Total click cost ÷ Clicks
This follows Google Ads’ definition (10).
Funnel stage
Acquisition.
Business question
How much did the advertiser pay, on average, for each click?
Required inputs
- media cost;
- billable clicks;
- campaign, audience and period.
Limitations
CPC measures the price of traffic, not its value. Auction competition, targeting constraints, placement mix and bid strategy all move CPC independently of anything downstream.
Interpretation warning
Falling CPC is not automatically an improvement. A more expensive click can bring a better-qualified visitor, a higher conversion rate, a larger order and a customer who stays longer.
Engagement metrics
Engagement metrics indicate whether people interacted with a site, app, email or content experience after arriving. Every platform defines engagement differently, so the definition has to travel with the number.
Engagement rate
Definition and formula
In Google Analytics 4:
Engagement rate = Engaged sessions ÷ Total sessions × 100
GA4 defines an engaged session as one that lasts longer than ten seconds, contains a key event, or records at least two page or screen views (11).
Other platforms define engagement as a click, reaction, comment, video view or some other interaction. Do not combine these under one unlabelled KPI.
Funnel stage
Engagement and consideration.
Business question
What proportion of sessions met the organisation’s minimum definition of meaningful activity?
Required inputs
- sessions;
- engaged sessions;
- configured key events;
- engagement-time and page-view settings;
- consent and tracking coverage.
Limitations
The metric inherits the analytics platform’s session rules and your event implementation. A user can qualify as engaged without showing any purchase intent, and a useful single-page experience behaves quite differently from a multi-page journey.
Consent choices, blocked scripts and implementation errors all reduce observable sessions and events. Shopify notes that its session reporting depends on consent and that analytics platforms calculate sessions differently (12).
Interpretation warning
Find out which engagement condition moved before you interpret the change. A rise driven by automatically collected key events means something different from a rise in active reading or product exploration.
Lead-generation metrics
Lead-generation metrics need an agreed definition of a lead. Mixing raw form submissions, marketing-qualified leads and sales-accepted leads makes the trend line meaningless.
Cost per lead
Definition and formula
CPL = Lead-generation cost ÷ Leads generated
LinkedIn states the formula plainly in its campaign-performance guidance: ad spend divided by leads (13). It follows the same structure as cost per action, which Google defines as total cost divided by the number of defined actions (14).
Funnel stage
Lead generation.
Business question
How much did it cost to generate each lead, under the lead definition you chose?
Required inputs
- included campaign or marketing costs;
- number of leads generated;
- lead qualification definition;
- source and reporting period;
- deduplication rules.
Limitations
CPL ignores lead quality, sales acceptance, close rate, sales-cycle length, deal value and retention.
A low CPL is easy to manufacture. Remove enough friction from a form and you will generate a great many unsuitable and duplicate contacts very cheaply.
Interpretation warning
Always report the lead type. Cost per raw enquiry, cost per marketing-qualified lead and cost per sales-qualified lead answer three different questions.
Lead-to-customer conversion rate
Definition and formula
Lead-to-customer conversion rate = New customers from the lead cohort ÷ Leads in the cohort × 100
Salesforce expresses sales conversion rate as sales divided by leads (15).
Funnel stage
Lead generation through conversion.
Business question
What proportion of leads eventually became customers?
Required inputs
- a defined lead cohort;
- a customer or closed-won outcome;
- identity matching between marketing and sales records;
- an observation window long enough for the sales cycle;
- treatment of reopened, duplicated and disqualified leads.
Limitations
A same-period calculation misleads whenever the sales cycle crosses reporting periods. Dividing customers closed this month by leads generated this month compares two populations that have almost nothing to do with each other.
The result also depends heavily on what qualifies as a lead. Tighten the qualification rules and the rate rises without a single additional customer.
Interpretation warning
Use cohort-based reporting wherever you can. Follow leads generated in a period until they convert or reach a defined maturity point.
There is a further limitation the number cannot show you, and I learned it running RFPs. The formal evaluation scores your written answers, your references and your pricing. What actually moved those decisions, repeatedly, was introducing the team that would take the account if we won, in person wherever possible.
When the teams who will work together meet, and it is better if it is in person, everything changed.
After those meetings we would sometimes start hearing what had been said in the client’s own decision-making sessions. None of that appears in a conversion rate. The metric tells you how many leads closed. It will not tell you what closed them, which is why a falling rate is a reason to go and talk to the sales team rather than a finding in its own right.
Conversion metrics
A conversion is a predefined action, not necessarily a sale. It might be a purchase, an application, a booked meeting, a registration or a completed trial.
Conversion rate
Definition and formula
The general form:
Conversion rate = Conversions ÷ Eligible opportunities to convert × 100
For advertising, Google divides conversions by trackable ad interactions (16). For ecommerce, Shopify defines online-store conversion rate as the percentage of sessions that result in an order (12).
Possible denominators include ad interactions, clicks, sessions, users, leads, checkout starts and product-page visits. The denominator must always be stated.
Funnel stage
Conversion.
Business question
What proportion of the defined eligible population completed the target action?
Required inputs
- conversion count;
- stated denominator;
- event definition;
- attribution window where applicable;
- reporting period;
- valid-traffic and consent rules.
Limitations
Conversion rate rises while total conversions fall whenever traffic contracts. It also moves when channel mix, audience composition or the denominator itself changes.
Platform-reported conversions depend on attribution rules and may include modelled or view-through outcomes.
Interpretation warning
Never report “conversion rate” without naming the conversion and the denominator. Purchase conversion per session and lead conversion per click are different KPIs that happen to share a label.
Revenue metrics
Revenue metrics connect customer actions to commercial value. They need consistent treatment of taxes, shipping, discounts, refunds, cancellations and currency conversion.
Revenue per visitor
Definition and formula
Revenue per visitor = Revenue ÷ Unique visitors
The session-based alternative:
Revenue per visit = Revenue ÷ Visits or sessions
Adobe lists both as separate default calculated metrics, revenue divided by unique visitors and revenue divided by visits, alongside revenue per order (17).
Funnel stage
Conversion and revenue.
Business question
How much revenue did the average measured visitor or visit generate?
Required inputs
- revenue under a documented definition;
- unique visitors or visits;
- identity and session rules;
- reporting period;
- treatment of refunds and delayed transactions.
Limitations
Visitor identity fragments across devices, browsers and consent states. Revenue can also arrive after the original visit, or through a different channel entirely.
The arithmetic mean is sensitive to a small number of very large orders.
Interpretation warning
Do not compare revenue per visitor against revenue per session. One visitor generates several sessions, so the two denominators cannot produce comparable results.
Average order value
Definition and formula
At its simplest:
AOV = Order revenue ÷ Orders
Adobe defines revenue per order as revenue divided by completed orders (17). Shopify’s reporting definition is gross sales minus discounts divided by orders, and it explicitly excludes post-order adjustments such as edits and exchanges (18). Two credible platforms, two different revenue bases. Document which one you are using.
Funnel stage
Revenue.
Business question
What was the average monetary value of a completed order?
Required inputs
- order revenue;
- completed order count;
- documented treatment of discounts, taxes, shipping, gift cards and refunds;
- currency conversion rules;
- reporting period.
Limitations
AOV is an order metric, not a customer metric. One customer places many orders.
It also rises when your lower-value customers stop buying, which is not usually what anyone means when they report an AOV improvement. And it says nothing about margin.
Interpretation warning
Read order volume and contribution margin alongside it. Lifting AOV through heavy discounting or expensive incentives can reduce profit while the KPI improves.
Return on ad spend
Definition and formula
ROAS = Revenue attributed to advertising ÷ Advertising spend
Expressed as a percentage:
ROAS percentage = Attributed conversion value ÷ Ad cost × 100
Google Ads reports this as conversion value per cost, defined as conversion value divided by cost (19).
Funnel stage
Revenue and media efficiency.
Business question
How much attributed revenue was recorded for each unit of advertising expenditure?
Required inputs
- advertising cost;
- attributed conversion value;
- attribution model and conversion window;
- transaction values;
- treatment of refunds, cancellations and offline sales.
Limitations
ROAS is built on revenue, not profit. It flatters low-margin products and penalises high-margin ones.
Platform-reported ROAS is also attribution-dependent. Two platforms will happily claim the same sale, and attributed revenue is not incremental revenue.
Interpretation warning
Do not compare ROAS figures until attribution windows, conversion definitions and value rules have been aligned. Use contribution margin or incremental profit for decisions about overall profitability.
Retention metrics
Retention metrics have to reflect the actual customer lifecycle. Monthly measurement suits a frequent subscription and tells you very little about a product repurchased once a year.
Repeat-purchase rate
Definition and formula
A customer-level formulation:
Repeat-purchase rate = Customers who made a subsequent purchase ÷ Purchasing customers in the eligible cohort × 100
Shopify defines its returning-customer rate as returning customers divided by total customers who placed orders (12).
Funnel stage
Retention.
Business question
What proportion of eligible purchasers came back and bought again?
Required inputs
- customer-level transaction history;
- first and subsequent order dates;
- customer identity matching;
- eligible cohort and observation period;
- treatment of subscriptions, exchanges and duplicate accounts.
Limitations
The metric is extremely sensitive to the observation window and the natural repurchase cycle. A customer who has not repurchased in thirty days is not lapsed in a category people buy from twice a year.
Returning-customer rate and cohort repeat-purchase rate can use different denominators. Keep the platform’s exact label when you report its built-in metric.
Interpretation warning
Compare cohorts at the same age. A six-month-old cohort has had six times the opportunity of a one-month-old cohort.
Customer retention rate
Definition and formula
Retention rate = (Customers at end of period − New customers acquired during period) ÷ Customers at start of period × 100
Stripe uses this formulation for customer retention rate (20).
Funnel stage
Retention.
Business question
What proportion of the opening customer base was still active at the end of the period?
Required inputs
- customers at the start of the period;
- customers at the end;
- new customers acquired during the period;
- a definition of an active customer;
- reporting period.
Limitations
“Active” can mean a paid subscription, a purchase inside a window, or any product use at all. Each definition produces a different rate.
The standard formula is most intuitive for contractual and recurring relationships. Non-contractual retailers usually learn more from cohort repurchase rates.
Interpretation warning
Customer retention and revenue retention are not the same measurement. Losing one large customer barely moves customer-count retention and can devastate retained revenue.
Customer churn rate
Definition and formula
Customer churn rate = Customers lost during period ÷ Customers at start of period × 100
Stripe presents this as the most common and straightforward churn formula, and stresses defining both the customer and the period before calculating (20).
Funnel stage
Retention.
Business question
What proportion of the opening customer base stopped being customers during the period?
Required inputs
- customers at start of period;
- customers lost;
- cancellation or inactivity definition;
- reporting period;
- treatment of pauses, involuntary churn and reactivations.
Limitations
Customer churn hides changes in account value. You can lose many small accounts and most of your revenue is fine. You can lose two large ones and the customer-count churn barely moves.
I have watched the second version happen. We lost one of our biggest clients because the CTO chose a vendor he had a personal relationship with. Our coverage of that account was strong at CEO and CMO level and essentially absent one seat over. Nothing in the churn number would have warned anyone, because until the moment it happened there was nothing to count. What we changed afterwards was the stakeholder map rather than the metric: CTO meetings became standard on enterprise accounts.
For the critical customers, we were playing the CTO-to-CTO card and sending emails from our own CTO.
Monthly and annual churn percentages should also never be compared directly without converting them onto a consistent basis and accounting for when customers actually left.
Interpretation warning
Retention rate is not reliably 100% − churn rate. The two diverge through differences in definition, timing, treatment of new customers, and whether the metrics are customer-, logo-, subscription- or revenue-based.
Customer lifetime value metrics
Customer lifetime value is a modelled estimate rather than an observed number. Its usefulness depends entirely on the margin, retention, purchase and discount-rate assumptions behind it.
Customer lifetime value
Definition
CLV estimates the present value of the future profit expected from a customer over the relationship.
Academic CLV models incorporate future customer margin, retention or survival probability, discounting, time horizon and acquisition cost. Gupta and colleagues define CLV as the present value of future profits and demonstrate that different contractual and non-contractual settings require materially different modelling approaches (21).
A general expression:
CLV = Sum of expected future customer contribution, discounted to present value − Acquisition cost
A simplified historical calculation:
Historical customer value = Total customer contribution margin to date ÷ Number of customers
That second one is not a forecast until you add assumptions about future behaviour.
Funnel stage
Customer lifetime value and long-term profitability.
Business question
How much economic value is a customer or segment expected to generate over the chosen horizon?
Required inputs
Depending on the model:
- order or subscription revenue;
- gross or contribution margin;
- purchase frequency;
- retention or survival probability;
- customer-service and retention costs;
- discount rate;
- forecast horizon;
- acquisition cost;
- customer or cohort identity.
Limitations
There is no universal CLV formula. Contractual subscriptions, non-contractual retail and multi-product financial relationships need different assumptions.
A revenue-based “lifetime value” will always be substantially higher than a margin-based CLV. Historical averages overstate future value whenever acquisition quality or retention is deteriorating, which is precisely when someone is most likely to quote them.
Interpretation warning
Label the measure precisely: historical revenue per customer, predicted lifetime revenue, gross-margin CLV or contribution-margin CLV. Never compare models built on different economic definitions.
CLV-to-CAC ratio
Definition and formula
CLV-to-CAC ratio = Customer lifetime value ÷ Customer acquisition cost
CAC should represent the cost of acquiring new customers, and CLV the value generated by a comparable customer population. Shopify defines CAC as acquisition-related expenditure divided by new customers and frames the CLV-to-CAC relationship as a way of weighing acquisition cost against longer-term customer value (22).
Funnel stage
Customer lifetime value and unit economics.
Business question
How much expected customer value is generated relative to the cost of acquisition?
Required inputs
- CLV under a documented margin and horizon definition;
- CAC for the same customer segment or cohort;
- aligned dates and acquisition channels;
- consistent currency and cost treatment.
Limitations
The ratio distorts badly when a fully loaded CAC meets a revenue-based CLV, or when a short-period CAC meets a mature-customer CLV estimate.
A high ratio is not automatically good news. It can indicate strong economics, or timid acquisition spending, or costs that were never fully counted.
Interpretation warning
There is no universal target ratio. The right economics depend on gross margin, cash flow, CAC payback period, growth stage, capital constraints, retention risk and how much you trust your own forecast. Our guide to CLV-to-CAC and profitable growth goes further into the unit economics.
Efficiency metrics
Efficiency metrics compare outputs against resources consumed. They are not measures of incrementality and they are not measures of profit.
Customer acquisition cost
Definition and formula
CAC = Total acquisition-related sales and marketing costs ÷ New customers acquired
Shopify defines CAC as the total cost of acquiring a new customer and specifies that the denominator holds first-time customers rather than retained ones (22).
Funnel stage
Acquisition efficiency.
Business question
How much did the company spend, on average, to acquire each new customer?
Required inputs
The numerator may include advertising and media, agency and affiliate fees, sales and marketing salaries, commissions, creative production, acquisition software, introductory incentives and allocated overhead.
The denominator requires first-time customers, a customer identity rule, and an acquisition period or cohort.
Limitations
“Paid-media CAC”, “marketing CAC” and “fully loaded sales-and-marketing CAC” are three metrics wearing one name.
Blended CAC cannot tell you which channel caused an acquisition. Channel-attributed CAC inherits every assumption in the attribution model and should never be treated as causal acquisition cost.
Interpretation warning
Align spend with the customers it had time to produce. Where the sales cycle is long, dividing this month’s spending by this month’s new customers pairs costs and outcomes from unrelated cohorts.
For a fuller treatment of definitions and attribution limits, see our guide to calculating and diagnosing customer acquisition cost.
Marketing efficiency ratio
Definition and formula
MER = Total revenue ÷ Total marketing expenditure
Some organisations call this blended return on advertising or marketing spend. HubSpot defines it as total revenue divided by total marketing spend across the same period and distinguishes it from campaign-level ROAS (23).
Funnel stage
Overall marketing efficiency.
Business question
How much total revenue did the business generate for each unit of total marketing expenditure?
Required inputs
- total revenue;
- total marketing expenditure;
- an aligned reporting period;
- documented inclusion of staff, media, agency, technology and production costs;
- a consistent gross or net revenue basis.
Limitations
MER does not isolate revenue caused by marketing. It contains demand from existing customers, organic demand, distribution, pricing, seasonality and everything else the business did that quarter.
It also cannot tell you which campaign or channel moved it.
Interpretation warning
MER improves when marketing spend is cut, including when the cut is quietly destroying next year’s demand. Examine revenue, customer acquisition, retention and contribution margin before you celebrate.
Profitability metrics
Revenue efficiency is not profit. Profitability metrics bring in at least part of the cost of delivering the product or service.
Contribution margin
Definition and formula
Contribution margin = Revenue − Variable costs
As a percentage:
Contribution margin ratio = Contribution margin ÷ Revenue × 100
ACCA defines contribution margin as total revenue less total variable costs, with the contribution-to-sales ratio calculated as total contribution divided by total sales revenue (24). That makes it structurally different from a revenue-only measure such as ROAS.
It is worth knowing that contribution margin is a management-accounting measure rather than a standardised financial-reporting line. Because organisations differ in which costs they treat as variable, and in whether items such as commissions, labour and warehousing are included, comparisons across companies are only meaningful where the calculation basis is consistent (25).
Funnel stage
Profitability.
Business question
How much revenue remains after variable costs to cover fixed costs, marketing and profit?
Required inputs
- net revenue;
- cost of goods sold where variable;
- transaction and payment costs;
- fulfilment and shipping costs where variable;
- sales commissions;
- promotional or service costs tied to the transaction;
- returns and refunds.
Limitations
Organisations classify costs differently. Some put fulfilment, customer service or promotional discounts inside contribution margin; others report them in a later contribution layer.
A product-level contribution margin usually excludes acquisition and fixed operating costs entirely.
Interpretation warning
Document which costs are included, and hold the same contribution layer constant when comparing products, channels or periods.
Marketing return on investment
Definition and formula
A profit-based marketing ROI:
Marketing ROI = (Incremental contribution generated by marketing − Marketing cost) ÷ Marketing cost × 100
Google’s advertising ROI guidance follows the same general structure: revenue attributable to advertising, less overall costs, divided by overall costs (26).
Funnel stage
Profitability and overall performance.
Business question
What financial return remained after the cost of marketing and the variable cost of the resulting sales?
Required inputs
- incremental revenue attributable to marketing;
- contribution-margin rate or incremental variable costs;
- marketing expenditure;
- a defined evaluation period;
- treatment of delayed customer value.
Limitations
The hard part is estimating incremental revenue, meaning the revenue that would not have arrived without the marketing activity. Platform attribution does not establish that counterfactual and was never designed to.
Short evaluation periods omit repeat purchases and longer-term effects. Long ones accumulate assumptions and external influences.
Interpretation warning
Do not call revenue divided by advertising spend “ROI”. That is ROAS. ROI requires a profit or contribution measure after the relevant costs.
Choosing between attribution, experiments and marketing-mix modelling is a question about choosing attribution and experimentation methods rather than a question about this formula.
How to interpret marketing KPIs without relying on generic benchmarks
No marketing benchmark travels safely across sectors and business models.
Before judging whether a number is good, segment it by the factors that actually determine its economics:
| Metric | Important context |
|---|---|
| CTR and CPC | Channel, placement, format, audience, match type and brand familiarity |
| CPL | Lead definition, qualification level, market and sales motion |
| Conversion rate | Conversion event, denominator, traffic source, device and customer type |
| CAC | Included costs, attribution approach, sales cycle and customer segment |
| AOV and revenue per visitor | Product mix, promotions, taxes, returns and seasonality |
| ROAS | Attribution window, margin, repurchase behaviour and incremental demand |
| Retention and churn | Contract type, repurchase cycle, cohort age and customer segment |
| CLV | Margin definition, horizon, retention assumptions and discount rate |
| MER | Revenue basis, marketing-cost coverage, growth rate and business maturity |
| Contribution margin | Variable-cost definition, fulfilment model and product mix |
Market is the context people underestimate most, so let me give you the version I paid for personally. I sold in Istanbul for years before moving to Paris, and I carried a habit across that I should have left behind. In Istanbul you build a negotiation buffer into every offer, because the client will negotiate hard and you need the room.
In Istanbul, brands are willing to negotiate heavily on pricing. In Europe, they basically reject. I stopped sharing offers with a buffer.
European buyers, in my experience, do not counter. They decline. The buffer that was essential in one market simply priced me out of the other, and none of that difference is visible in a conversion rate or a CPL. Two markets, one offer, the same metric definition, and opposite readings. Any benchmark that crosses a border carries that problem invisibly.
The most defensible comparisons usually start with:
- the company’s own historical performance under consistent definitions;
- matched customer or campaign cohorts;
- the same channel, audience, product and period;
- expected economics based on margin and customer lifecycle;
- independently measured incremental effects, where the decision requires causal evidence.
Metric-definition checklist
Before a KPI reaches a report, record:
| Field | Example |
|---|---|
| Metric name | Purchase conversion rate |
| Formula | Purchase sessions ÷ eligible sessions |
| Numerator | Sessions containing a completed purchase |
| Denominator | Eligible online-store sessions |
| Included population | UK website visitors |
| Exclusions | Internal traffic, bots, test orders |
| Data source | Web analytics and transaction system |
| Attribution rule | Session containing the purchase |
| Reporting window | Calendar month |
| Owner | Ecommerce analytics |
| Known limitations | Consent loss and cross-device identity |
| Related diagnostic metrics | Checkout completion, AOV, revenue per visitor |
This is what stops two teams using one label for two calculations, and it makes a change in methodology visible instead of mysterious.
Final takeaway
A marketing KPI is useful only once its formula, denominator, economic scope and limitations are understood by the people reading it.
Use delivery metrics to understand exposure, acquisition metrics to understand response, conversion metrics to understand action, retention metrics to understand continuity, and margin-based metrics to understand economic value.
Then use the broader marketing measurement framework to decide which of these should become organisational KPIs, how they should be validated, and which measurement method should inform each decision. If your work involves personalised experiences specifically, our reference on personalisation KPIs and formulas covers the metrics that sit alongside these.
Frequently Asked Questions
How do you calculate customer acquisition cost?
Divide total acquisition-related sales and marketing costs by the number of new customers acquired in the same period. The denominator should contain first-time customers only, not retained ones. The harder question is what belongs in the numerator: paid-media CAC, marketing CAC and fully loaded sales-and-marketing CAC produce very different figures from the same business, so state which one you mean. Where the sales cycle is long, align the spend with the cohort it had time to generate rather than dividing this month's cost by this month's customers.
What is the difference between ROAS and marketing ROI?
ROAS is revenue attributed to advertising divided by advertising spend. Marketing ROI is a profit measure: incremental contribution generated by marketing, less marketing cost, divided by marketing cost. ROAS can look strong on a product that makes no money, because it never touches the cost of goods. Revenue divided by ad spend should not be labelled ROI, which is one of the most common reporting errors in marketing.
Which marketing metrics belong at each funnel stage?
Awareness uses share of voice and branded-search growth. Reach and attention use reach, impressions and frequency. Acquisition uses click-through rate and cost per click. Engagement uses engagement rate. Lead generation uses cost per lead and lead-to-customer conversion rate. Conversion uses conversion rate. Revenue uses revenue per visitor, average order value and return on ad spend. Retention uses repeat-purchase rate, retention rate and churn. Efficiency and profitability use CAC, marketing efficiency ratio, contribution margin and marketing ROI. The mapping is diagnostic rather than fixed: the same event can be a lead for one business and a conversion for another.
What is the difference between cost per lead and cost per acquisition?
Cost per lead divides lead-generation cost by the number of leads, where a lead is someone who has expressed interest. Cost per acquisition divides cost by the number of completed defined actions, which for most businesses means a customer or a purchase. A low CPL alongside a high CPA usually means the lead definition is too loose, and the two figures should never be compared with each other or against benchmarks built on the other denominator.
Is retention rate just 100% minus churn rate?
Not reliably. The two can diverge because of differences in definition, timing, the treatment of new customers acquired during the period, and whether the metric counts customers, logos, subscriptions or revenue. They also answer different questions when account values vary: losing one large customer barely moves customer-count retention while doing serious damage to retained revenue. Calculate and report both rather than deriving one from the other.
References
- Nielsen. Need to know: What is share of voice? March 2025. https://www.nielsen.com/insights/2025/what-is-share-voice/
- Google. Performance report: overview and basic configuration, Search Console Help. Accessed 27 July 2026. https://support.google.com/webmasters/answer/7576553
- Google. What are impressions, position and clicks?, Search Console Help. Accessed 27 July 2026. https://support.google.com/webmasters/answer/7042828
- Google. Unique Reach: Definition, Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/9012727
- Meta. About reach metrics, Meta Business Help Centre. Accessed 27 July 2026. https://www.facebook.com/business/help/710746785663278
- Google. Impressions: Definition, Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/6320
- Meta. Impressions, Meta Business Help Centre. Accessed 27 July 2026. https://www.facebook.com/business/help/675615482516035
- Google. Frequency: Definition, Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/59384
- Google. Clickthrough rate (CTR): Definition, Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/2615875
- Google. Average cost-per-click (Avg. CPC): Definition, Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/14074
- Google. [GA4] Engagement rate and bounce rate, Analytics Help. Accessed 27 July 2026. https://support.google.com/analytics/answer/12195621
- Shopify. Analytics data points (fields) reference, Shopify Help Center. Accessed 27 July 2026. https://help.shopify.com/en/manual/reports-and-analytics/shopify-reports/report-types/analytics-fields
- LinkedIn. How to analyse your campaign performance, LinkedIn Marketing Solutions. Accessed 27 July 2026. https://business.linkedin.com/marketing-solutions/success/best-practices/analyze-your-performance
- Google. Cost per action: Definition, Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/13278730
- Salesforce. What is a sales conversion rate, and how is it measured? 21 January 2025. https://www.salesforce.com/sales/analytics/sales-conversion-rate/
- Google. Conversion rate: Definition, Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/2684489
- Adobe. Default calculated metrics, Adobe Analytics documentation. Updated 26 May 2026. https://experienceleague.adobe.com/en/docs/analytics/components/calculated-metrics/calcmetrics-reference/default-calcmetrics
- Shopify. Sales reports, Shopify Help Center. Accessed 27 July 2026. https://help.shopify.com/en/manual/reports-and-analytics/shopify-reports/report-types/default-reports/sales-report
- Google. Conversion value per cost: Definition, Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/13405059
- Stripe. Retention rate vs. churn rate: what businesses need to know. 8 November 2024. https://stripe.com/resources/more/retention-rate-vs-churn-rate-what-businesses-need-to-know
- Gupta, S., Hanssens, D., Hardie, B., Kahn, W., Kumar, V., Lin, N., Ravishanker, N. and Sriram, S. Modeling Customer Lifetime Value. Journal of Service Research, 9(2), 139-155, November 2006. https://journals.sagepub.com/doi/10.1177/1094670506293810
- Shopify. Customer acquisition cost: how to calculate and reduce it. 29 July 2024. https://www.shopify.com/uk/blog/customer-acquisition-cost
- HubSpot. Marketing efficiency ratio: how to calculate and improve yours. 2 January 2026. https://blog.hubspot.com/marketing/marketing-efficiency-ratio
- Association of Chartered Certified Accountants. Cost-volume-profit analysis, technical article. Accessed 27 July 2026. https://www.accaglobal.com/gb/en/student/exam-support-resources/fundamentals-exams-study-resources/f5/technical-articles/CVP-analysis.html
- Cafferky, M. E. Estimating Retail Breakeven Using Markup Pricing. Management Accounting Quarterly, Institute of Management Accountants, Winter 2017, 18(2). https://www.imanet.org/-/media/b11fc3a98e17436a875ab7eb2204ffaf.ashx
- Google. Return on investment (ROI), Google Ads Help. Accessed 27 July 2026. https://support.google.com/google-ads/answer/14090
Use this guide as a source
If it settled an argument in your reporting, cite it, and add Herm as a preferred source so the next answer you get from search or a model comes from work with its definitions attached.
