Contents
In short. Start with the question, not the acronym. CPM, CTR and CPC show what attention costs; conversion rate, CPL and CPA what an inquiry costs; CAC what a new customer costs. Average order value, contribution margin and LTV show what a customer brings in, and CAC payback when the money comes back. An ad platform answers the first questions reliably; the rest need your CRM and financial data. Before comparing any number, confirm its numerator, its denominator, the period and the rule that credited the sale to a source.
Four levels of metrics: from attention to cash
Every term follows one pattern: what it is, the formula, a worked example and a common mistake. Definitions were checked against help pages from Google Ads, Google Analytics, Meta, Yandex, HubSpot and an OpenStax textbook; where platforms differ, the entry says so.
The glossary links our other guides: lead cost versus customer cost (“A cheap lead can still be an expensive customer”), affordable lead cost, cohort conversion and lead quality.
Scroll right to see all columns
| Level | Metrics | Question | Data source |
|---|---|---|---|
| Attention | CPM, CTR, CPC | Cost of an impression and a click | Ad platform |
| Inquiries | Conversion rate, CPL, CPA, attribution, MQL, SQL | Cost of an inquiry; which are in play | Platform, analytics, CRM |
| Customers | CAC, full CAC, AOV, cohort, retention, churn | Cost of a buyer; do they stay | CRM and accounts |
| Money | ROAS, contribution margin, ROMI and ROI, MER, LTV, CAC payback, LTV:CAC | What a customer brings; when cash returns | CRM, accounts, finance |
The worked company: from impression to customer
Every calculation uses one made-up company; the figures are illustrative, not benchmarks. Assumptions: one month of ads in one channel, leads observed for 60 days, new customers for 12 months. All calculations were rerun in code and recorded in the verification ledger.
| Item | Value |
|---|---|
| Impressions | 200,000 |
| Clicks | 3,000 |
| Sessions from ads | 2,600 |
| Unique leads | 120 |
| MQLs | 60 |
| SQLs | 30 |
| New customers | 12 |
Illustrative example. Share of the previous step: CTR 1.5%, sessions 86.7% of clicks, leads 4.6% of sessions, MQLs 50%, SQLs 50%, customers 40% of SQLs.
Scroll right to see all columns
| Input | Value |
|---|---|
| Ad spend | $6,000 |
| Platform conversions | 150, including repeat forms and calls |
| Other acquisition costs | $3,600 |
| New customer revenue in month one | $21,600 from 18 orders |
| Contribution share after variable costs | 40% |
| Contribution per customer after month one | $80 a month on average |
| All revenue and all marketing spend for the month | $72,000 and $16,000 |
CPM: what a thousand impressions cost
CPM (cost per mille) is ad spend per thousand impressions. In Google Ads it is also a bidding method in which you pay per thousand impressions on the Display Network.
Formula: spend / impressions × 1,000. Meta’s example: $50 on 10,000 impressions is a $5 CPM.
Example: $6,000 / 200,000 × 1,000 = $30.
Mistake: treating an impression as a person. One person can see the ad several times, which is why Meta also reports cost per 1,000 accounts reached. A low CPM is the price of attention, not of buyers: compare it within one format and audience.
CTR: does the ad catch the people who see it
CTR (clickthrough rate) is the share of impressions that led to a click, as Google Ads defines it. Meta reports a link clickthrough rate and explains how clicks (all) differ from link clicks, so check which CTR you are looking at.
Formula: clicks / impressions × 100%.
Example: 3,000 / 200,000 = 1.5%.
Mistake: comparing CTR across platforms. Google says a good CTR is relative to what you advertise and on which network. A headline promising what the site does not deliver lifts CTR and lowers the share who inquire.
CPC: what a click costs
CPC (cost per click) is what the platform charges for a click. Google Ads separates the maximum CPC, your bid ceiling, from the actual CPC charged, which is often lower.
Formula: spend / clicks; Google’s example is clicks at $0.20 and $0.40, an average of $0.30. Also, CPC = CPM / (1,000 × CTR).
Example: $6,000 / 3,000 = $2.00; check: $30 / (1,000 × 1.5%) = $2.00.
Mistake: chasing cheap clicks, which often come from broader, less relevant keywords or audiences. A $1.20 click at a 1% lead rate gives a $120 lead; a $2.00 click at 4% gives a $50 lead.
Conversion rate: where people drop out
Conversion rate (CR) is the share of people or interactions that completed the next target action. The denominator is what matters. Google Ads divides conversions by trackable ad interactions, and the rate can exceed 100% when several conversions count per interaction. Google Analytics reports session and user key event rates.
Formula: target actions / entries into the step × 100%.
Example: 120 leads from 3,000 clicks is 4%; from 2,600 sessions, 4.6%. Both are correct. Lead to customer: 12 / 120 = 10%.
Mistake: not naming the steps (click to lead, lead to paid order) or measuring sales by calendar month, which mixes leads of different ages; see our guide on cohort conversion.
CPL: what one inquiry costs
CPL (cost per lead) is ad spend per inquiry. Meta defines it as amount spent divided by leads and may use statistical modelling when some lead events cannot be measured directly.
Formula: spend / unique leads.
Example: $6,000 / 120 = $50.
Mistake: never agreeing on what counts as a lead. Repeat submissions, a second call from the same person and spam make CPL look cheaper: count unique contacts in the CRM. A cheap lead that never buys is the most expensive kind; see “A cheap lead can still be an expensive customer”.
CPA: what the action you configured costs
CPA (cost per action or acquisition) is ad spend per action configured in the platform: a form, a call, a sign-up or a purchase. Without the action’s name, CPA tells you nothing.
Formula: cost of conversions / conversions, as Google Ads calculates average CPA.
Example: 150 conversions, including repeat forms and calls: $6,000 / 150 = $40, against a CRM-based CPL of $50. The 25% gap comes only from counting rules.
Mistake: reading CPA as acquisition cost. Google’s Target CPA bidding seeks as many tracked conversions as possible at the target cost; if any form counts, it finds cheap forms. With enough volume, send later stages to the platform.
Attribution: which channel gets the sale
Attribution is the rule for crediting an inquiry or sale to sources on the customer’s path. First touch gives all credit to the first source; last touch (last click) to the last.
Models in tools. Google Analytics 4 attribution reports offer data-driven and two last-click models; first click, linear, time decay and position-based have been unavailable since November 2023, though the source that first acquired a user remains visible as First user source.
Example: an article found in search, a social ad a week later, then a brand search ad click and a purchase. First touch credits organic search, last touch the brand ad. The CRM records 12 sales; two ad platforms report 9 and 7, a total of 16.
MQL: which leads go to sales
An MQL (marketing qualified lead) is a contact that marketing, using agreed criteria, has qualified as ready for sales, as HubSpot’s lifecycle stages describe it. The business sets the criteria: location, budget, need, complete contact details.
Formula: MQLs / all leads.
Example: 60 of 120 leads, a 50% rate; ad spend per MQL $6,000 / 60 = $100.
Mistake: marketing grades its own leads and reports on its own grades. Loosen the criteria and MQLs rise while sales stay flat; our guide on lead quality shows how both teams grade one sample.
SQL: which leads sales accepts
An SQL (sales qualified lead) is a contact the sales team, after checking, has accepted as a potential customer. In HubSpot the next stage is an opportunity, a contact with an associated deal, and a customer has at least one closed deal.
Formula: SQLs / MQLs.
Example: 30 of 60 MQLs (50%); 12 of those 30 buy (40%). Ad spend per SQL: $6,000 / 30 = $200.
Mistake: each salesperson applies different rules. An SQL records a judgment, not a payment: rising SQLs with flat sales point to a loose definition or a longer cycle.
CAC: what a new customer costs
CAC (customer acquisition cost) is all acquisition costs divided by new customers for the same period or lead group. Shopify defines it as the total cost of acquiring a customer: marketing software and staff, advertising, new-customer discounts, content and sales expenses that support marketing.
Formula: (advertising + other acquisition costs) / new customers.
Example: $6,000 of ads plus $3,600 of other costs (agency and sales time, creative, tools) is $9,600; $9,600 / 12 = $800.
Mistake: dividing by every buyer, including repeat customers and those who came without ads: CAC looks lower. With a long cycle, calculate by lead cohort and fix the cost list in advance.
Full CAC or ad spend per customer: is CAC understated
Ad spend per customer is advertising alone divided by new customers; full CAC adds every other acquisition cost. Reports often label both as CAC.
Formula: ad spend / new customers, against all acquisition costs / new customers.
Example: $6,000 / 12 = $500 against $800: full CAC is $300, or 60%, higher.
Mistake: comparing $500 with a customer’s full value and overestimating headroom. Growing budgets need more sales time, so full CAC rises faster than the ad-only figure. Set limits on full CAC; turn only its ad part into an affordable lead cost.
AOV: what one purchase brings in
Average order value (AOV) is revenue divided by orders for the same period, as Shopify gives it. Shopify’s example counts shipping; Google Analytics 4 calculates purchase revenue net of refunds. Pick a rule and keep it.
Formula: revenue / orders.
Example: 12 new customers placed 18 orders worth $21,600: AOV $1,200, revenue per customer $1,800.
Mistake: dividing revenue by customers and calling it AOV. AOV rises with bundle discounts or one large order while margin may fall: read it with the contribution share, and with the median order if sizes vary widely.
ROAS: does advertising pay for itself on revenue
ROAS (return on ad spend) is attributed revenue divided by ad spend. Google Ads’ example: $5 in sales per $1 of ads, a target ROAS of 500%. Meta divides purchase conversion value by amount spent. Some teams use the inverse, ad spend as a share of revenue, which carries the same information.
Formula: attributed revenue / ad spend.
Example: $21,600 / $6,000 = 3.6, or 360%; the inverse is 27.8%.
Setup mistake: the platform reports whatever value you configured. Google lets you assign one value to every conversion or send different values. An assigned lead value flowing into ROAS is not revenue.
Contribution margin: what is left from a sale
Contribution margin is the amount by which the price exceeds the sale’s variable costs, as the OpenStax managerial accounting textbook defines it. Variable costs rise with every order: goods or delivery, shipping, payment and marketplace fees, discounts. Contribution pays for acquisition and fixed costs; the rest is profit.
Formula: price - variable costs; ratio = contribution / price. Textbook: a $100 price, $20 variable cost, $80 contribution, 80% ratio.
Example: $21,600 × 40% = $8,640; after the $9,600 full CAC, the group’s first month is minus $960.
Mistake: using a gross margin without shipping and fees, or deducting rent and salaries. A company average hides that ads may mostly sell low-margin products.
ROMI and ROI: did we earn more than we put in
ROI compares what you earned with what you invested; ROMI applies it to marketing, and both are defined in different ways. Google Ads compares ad profit with costs: $200 of ads added to $600 of product cost, ($1,200 - $800) / $800 = 50%. The OpenStax textbook divides divisional income by average capital assets.
Formula in this glossary: ROMI = (contribution - marketing spend) / marketing spend.
Example: one data set, three answers: 44% on contribution with ads only, -10% on full CAC and 13.9% following the Google Ads example. The calculations are under the chart.
Mistake: comparing ROI across reports without labelling numerator and denominator. A one-month ROMI ignores repeat purchases: minus 10% in month one becomes payback by month two, while a positive ad-only ROMI can hide a loss after other costs.
MER: is marketing working overall
MER (marketing efficiency ratio) is all revenue for a period divided by all marketing spend for the same period. HubSpot’s blog contrasts it with ROAS: ROAS helps allocate spend across channels, MER shows whether total spend generates revenue. There is no universal good MER; compare with your own history and subtract refunds first.
Formula: all revenue / all marketing spend.
Example: $72,000 / $16,000 = 4.5.
Mistake: putting only ads in the denominator and calling it ROAS. MER rises with repeat purchases even when new ads stop paying back, so read it with CAC and the share of new buyers. HubSpot warns that long B2B cycles distort MER on closed revenue.
Cohort: how to compare periods fairly
A cohort is a group of customers sharing a characteristic, such as the month of first inquiry, tracked as one unit. In Google Analytics the inclusion criterion decides who enters the group and the return criterion who remains.
Rule: compare groups at the same age.
Example: 120 September leads produced 60 MQLs, 30 SQLs and 12 customers in 60 days. October at day 30 is compared with September at day 30, not day 60.
Mistake: comparing a young group with a mature one and concluding that sales fell. Small cohorts swing randomly: with 120 leads, one customer shifts the rate by almost a percentage point.
Retention: do customers come back
Retention is the share of a group that bought again or stayed active after a set time. Mixpanel distinguishes returning in exactly the period (on) from that period or later (on or after), its default.
Formula: returning members / group size.
Example: 50 of 200 March customers bought again in month three or later: 25%.
Mistake: not stating the rule and window: exactly-month-three retention can never exceed month-three-or-later. For rare purchases such as renovation, low retention is expected; for subscriptions it is the core metric. Read it with group revenue or contribution.
Churn: how much of the base is lost
Churn is the share of customers who stopped buying or cancelled during a period. In Salesforce’s example, 500 customers fall to 450, 10% churn. For the starting base, retention equals 100% minus churn.
Formula: customers lost from the starting base / starting base.
Example: 20 of 400 subscribers cancel in a month: churn 5%, retention 95%.
Mistake: mixing in customers acquired during the period; in Salesforce’s example this moves a quarterly rate from 15% to 16.5%. Customer churn is not revenue churn. Without subscriptions, first define a lost customer, such as 180 days without a purchase.
LTV: what a customer is worth in total
LTV (lifetime value) is the money a customer brings over the relationship or a chosen period. In Google Analytics 4 it is purchase revenue, not contribution; for acquisition decisions use contribution LTV over a period you actually observed.
Formula. Shopify gives three versions: AOV × purchase frequency × lifespan; the same × gross margin; and for subscriptions, monthly recurring revenue per customer × gross margin / monthly churn.
Example: month-one contribution is $8,640 / 12 = $720 per customer, then $80 a month: $1,600 over 12 months, or $4,000 of revenue. The subscription formula for the churn example: $200 × 40% / 5% = $1,600.
Mistake: comparing revenue LTV with CAC. Forecasts are sensitive to churn: at 4% instead of 5% the formula gives $2,000, a quarter more. Shopify notes such formulas assume constant margin and retention.
CAC payback: when the money comes back
CAC payback is the number of months until a customer’s cumulative contribution covers the cost of acquiring them.
Formula. Wall Street Prep, for subscriptions: sales and marketing expense / (new monthly recurring revenue × gross margin); a $560 CAC with $50 a month at an 80% margin pays back in 14 months. With irregular purchases, compare cumulative contribution with CAC month by month.
Example: with an $800 CAC, cumulative contribution reaches $720, $800 and $880 after months one to three: payback in 2 months.
Mistake: measuring on revenue, so $1,800 of month-one revenue seems to cover CAC at once. Payback ignores cash timing when ads are paid upfront and customers pay in instalments. A long payback means growth needs a cash buffer.
| Item | Cumulative contribution | Full CAC |
|---|---|---|
| Month 1 | $720 | $800 |
| Month 2 | $800 | $800 |
| Month 3 | $880 | $800 |
Illustrative example: $720 of contribution in month one, then $80 a month; contribution matches CAC in month two.
LTV:CAC: is there headroom
LTV:CAC shows how many times a customer’s value over the period exceeds acquisition cost. The often-quoted guideline is 3:1: David Skok writes that the best SaaS businesses exceed 3 and treats CAC recovery beyond 12 months as weak economics. It is a subscription guideline without a published methodology, not a norm.
Formula: LTV / CAC, for the same group and period.
Example: 12-month contribution $1,600 / $800 = 2.0; revenue $4,000 / $800 = 5.0; revenue over ad spend per customer $4,000 / $500 = 8.0.
Mistake: ignoring time. 3:1 over five years and over six months are different businesses, so always show CAC payback next to LTV:CAC.
How to agree on shared definitions in a week
The result of the week: one page of definitions signed off by marketing, sales and finance, and the chain recalculated on a real lead group. No agency is needed.
Limits. All figures are illustrative, not benchmarks. Platform definitions were checked on 31 October 2026 and change, so confirm them in your accounts. The 3:1 and 12-month guidelines come from subscription businesses and lack a published methodology. The publisher is connected to WEBNOSTRA, which sells advertising and analytics services; the glossary works without hiring anyone.
- Write down how CPL, CPA, CAC, ROAS and ROMI are calculated today: numerator, denominator, period, data source.
- Agree what goes into full CAC and how shared costs are split.
- With sales, define lead, MQL, SQL and customer.
- Recalculate the chain from CPM to LTV:CAC on one month’s leads.
- Next to every metric, state the attribution model and window.
- Name an owner for the glossary and the next review date.
Key decisions
- sort metrics into four levels, attention, inquiries, customers and money, and never judge profit from the top levels
- ask every number for its numerator, denominator, period and attribution rule, because identical labels often hide different calculations
- set budget limits on full CAC, contribution margin and payback, and use CPL, CPA and ROAS to manage channels
What to discuss with your team
- Every metric has its numerator, denominator, period and data source written down
- Lead, MQL, SQL and customer mean the same in ad platforms, CRM and reports
- Ad spend per customer is labelled apart from full CAC
- ROAS, ROMI and LTV are on contribution or labelled revenue-based
- Platform totals match the CRM; each channel shows its attribution model and window
- LTV:CAC appears with CAC payback and the observation period
Terms in plain language
- CAC
- All acquisition costs divided by new customers in the same group, with the included costs stated
- Contribution margin
- Selling price minus the variable costs of the sale; it pays for acquisition and fixed costs
- LTV
- Money from a customer over the relationship; for acquisition decisions, measure it on contribution
- MER
- All revenue divided by all marketing spend for the same period; a business-wide check, not a channel metric
- MQL and SQL
- Leads accepted by marketing (MQL) and then by sales (SQL) under agreed criteria
Sources
- Google Ads Help: Cost-per-click (CPC): Definitionsupport.google.com · checked 31/10/2026
- Google Ads Help: Average cost-per-click (Avg. CPC): Definitionsupport.google.com · checked 31/10/2026
- Google Ads Help: Clickthrough rate (CTR): Definitionsupport.google.com · checked 31/10/2026
- Google Ads Help: Cost-per-thousand impressions (CPM): Definitionsupport.google.com · checked 31/10/2026
- Google Ads Help: Conversion rate: Definitionsupport.google.com · checked 31/10/2026
- Google Ads Help: Average CPA: Definitionsupport.google.com · checked 31/10/2026
- Google Ads Help: About Target CPA biddingsupport.google.com · checked 31/10/2026
- Google Ads Help: About Target ROAS biddingsupport.google.com · checked 31/10/2026
- Google Ads Help: Return on investment (ROI): Definitionsupport.google.com · checked 31/10/2026
- Google Ads Help: About conversion valuessupport.google.com · checked 31/10/2026
- Google Analytics Help: Get started with attributionsupport.google.com · checked 31/10/2026
- Google Analytics Help: Cohort explorationsupport.google.com · checked 31/10/2026
- Google Analytics Help: Dimensions and metrics (purchase revenue, LTV, first user source)support.google.com · checked 31/10/2026
- Meta Business Help Center: CPM (cost per 1,000 impressions)facebook.com · checked 31/10/2026
- Meta Business Help Center: CTR (link click-through rate)facebook.com · checked 31/10/2026
- Meta Business Help Center: Cost per leadfacebook.com · checked 31/10/2026
- Meta Business Help Center: Purchase ROAS (return on ad spend)facebook.com · checked 31/10/2026
- Yandex Direct: Glossaryyandex.com · checked 31/10/2026
- Yandex Direct: Revenue and conversion valueyandex.com · checked 31/10/2026
- Yandex Metrica: Glossaryyandex.com · checked 31/10/2026
- Yandex Metrica: Attribution modelsyandex.com · checked 31/10/2026
- HubSpot Knowledge Base: Use lifecycle stagesknowledge.hubspot.com · checked 31/10/2026
- HubSpot Blog: Marketing efficiency ratio (vendor blog)blog.hubspot.com · checked 31/10/2026
- Shopify: Customer acquisition cost (vendor blog, definition and cost scope only)shopify.com · checked 31/10/2026
- Shopify: Average order value (vendor blog)shopify.com · checked 31/10/2026
- Shopify: Customer lifetime value, CLV formulas (vendor blog)shopify.com · checked 31/10/2026
- OpenStax: Principles of Accounting, Volume 2, contribution marginopenstax.org · checked 31/10/2026
- OpenStax: Principles of Accounting, Volume 2, return on investment, residual income and EVAopenstax.org · checked 31/10/2026
- Salesforce: How to calculate customer churn and revenue churnsalesforce.com · checked 31/10/2026
- Mixpanel Docs: Retentiondocs.mixpanel.com · checked 31/10/2026
- David Skok, forEntrepreneurs: SaaS Metrics 2.0, LTV:CAC and CAC recovery guidelines (subscription companies)forentrepreneurs.com · checked 31/10/2026
- Wall Street Prep: CAC payback period (vendor training material, definition and formula only)wallstreetprep.com · checked 31/10/2026