
Which marketing numbers should change a decision?
Once a month you pull the numbers. Sessions and engagement from GA4, clicks and conversions from Google Ads, reach and cost per result from Meta, opens and clicks from your email tool. They go into a slide deck or a shared doc, often as 14 screenshots that get a polite nod in the meeting and are never opened again. The report is complete. What it doesn’t tell you is whether to raise the ad budget, fix the booking page, or stop posting on Instagram three times a week.
Most lists of marketing metrics make this worse. They hand you 40 KPIs, each with a definition, and leave the choosing to you. A formula for cost per lead tells you how to calculate it. It doesn’t tell you what to do when the number doubles.
This guide works the other way around, starting from the outcome the business needs, the same order Google recommends for a measurement plan that runs from business objectives down to KPIs and targets. Metrics are grouped by the business question each one answers. Every metric gets a formula, a one-line example and the decision it should drive. The guide also flags the numbers you can safely ignore and ends with a one-page monthly dashboard of five to eight metrics.
The running example is Riverside Physical Therapy, a fictional two-location clinic that runs Google Ads, SEO, email and social. It needs leads, but it gets paid for booked patients, and those are two different numbers.

Before picking any of them, you need a way to tell which numbers belong on that page. What separates a metric from a KPI, and either of those from noise?
Metrics, KPIs and vanity metrics, defined
A metric is any number you can measure: sessions, clicks, followers, calls, refunds. Your tools hand you hundreds of them for free, which is part of the problem.
A KPI is one of the few metrics tied to a business goal, and each has a target attached. “New patients” is a metric. “60 new patients a month at a CAC under $100” is a KPI, because it states what success looks like and someone can be held to it. Most small businesses need five to eight of these. More than that and you’re back to reading screenshots.
A vanity metric goes up and to the right and changes no decision. The test is blunt. If this number moved 20% next month, what would you do differently? If the answer is “nothing,” it’s vanity, or at best context. Eric Ries made the vanity versus actionable distinction popular in The Lean Startup (2011), and it has aged well.
Vanity metrics aren’t useless, though. Impressions and page views help you diagnose a problem, as in “did the ad stop showing, or did people stop clicking?” They work as diagnostic or awareness measures and become KPIs only when they’re tied to a goal and a decision. They belong in an appendix. They shouldn’t sit on the front page of your report.
Each vanity number has an actionable counterpart:
| Usually vanity | More actionable |
|---|---|
| Impressions, followers | Cost per lead |
| Total sessions | Sessions to target pages that convert |
| Email open rate (inflated by privacy features) | Click rate, revenue per email |
| Likes | Engagement rate on posts that drive profile or site visits |
| Page views | Conversion rate by landing page |
Riverside is a good example. Its Instagram followers rose 30% over the quarter, and the team felt great about it. Booked evaluations stayed flat. Follower growth tells you the audience got bigger. It doesn’t tell you whether that audience books appointments, so it can’t decide where next month’s budget goes.

With those definitions in place, the next step is the list of metrics worth considering, each with its formula and the decision it drives.
The core marketing metrics, with formulas
Every example uses Riverside’s fictional month, so the numbers carry into the dashboard at the end: $3,000 on Google Ads, $5,400 on marketing overall, 250 leads (120 from ads) and 60 new patients, each worth $1,100 in revenue and $440 in gross profit.
Are the right people finding us?
| Metric | Formula | What it tells you | Decision it drives | Riverside example |
|---|---|---|---|---|
| Sessions (and organic search sessions) | GA4 visits, filtered by channel | Where traffic comes from | Which channels to grow or fix | 2,100 organic sessions |
| CTR | Clicks / impressions × 100 | Whether the ad earns the click | Rewrite ads or titles | 1,500 / 25,000 = 6% |
| Cost per lead (CPL) | Spend / leads | What a lead costs | Shift budget between campaigns | $3,000 / 120 = $25 |
| Customer acquisition cost (CAC) | Total marketing cost / new customers | What a paying customer costs | Overall budget level | $5,400 / 60 = $90 |
A lead called or filled in a form. A customer paid. If CPL falls while CAC rises, you’re buying cheaper leads who never book.
Are visitors becoming leads and customers?
| Metric | Formula | What it tells you | Decision it drives | Riverside example |
|---|---|---|---|---|
| Conversion rate | Conversions / sessions or clicks × 100 | How well a page turns visits into leads | Fix pages before buying traffic | 120 / 1,500 = 8% |
| Average order (or customer) value | Revenue / orders or customers | Revenue per sale | Pricing, packages, upsells | $66,000 / 60 = $1,100 |
Is marketing paying for itself?
| Metric | Formula | What it tells you | Decision it drives | Riverside example |
|---|---|---|---|---|
| ROAS | Ad revenue / ad spend | Revenue per ad dollar | Scale or cut a campaign | 30 × $1,100 / $3,000 = 11 |
| Marketing ROI | (Gross profit − total marketing cost) / total marketing cost × 100 | Profit on everything you spend | Whether marketing keeps its budget | ($26,400 − $5,400) / $5,400 = 389% |
| LTV | Revenue per customer × gross margin, over a set period | A customer’s profit value | Max you can pay to acquire one | $1,100 × 40% = $440 |
| LTV:CAC | LTV / CAC | Whether acquisition is profitable | Push harder or fix costs | $440 / $90 = 4.9 |
ROAS is revenue over ad spend. ROI is profit over all marketing costs, so an 11 ROAS can sit beside an ugly ROI when margins are thin. The famous 3:1 LTV:CAC target is a SaaS rule of thumb from David Skok, so watch your own ratio’s trend instead, and compare CAC payback against current B2B marketing benchmarks.
Are customers coming back and remembering us?
| Metric | Formula | What it tells you | Decision it drives | Riverside example |
|---|---|---|---|---|
| Retention rate (churn = 100 − retention) | (End customers − new) / start customers × 100 | Who stays | Invest in follow-up | (215 − 60) / 200 = 77.5% |
| Repeat purchase rate | Returning / total past customers × 100 | Who comes back | Rebooking campaigns | 18 / 150 = 12% |
| Email open and click rate | Unique opens or clicks / delivered × 100 | Clicks beat privacy-inflated opens | Subject lines, offers | Opens 30%; clicks 124 / 4,000 = 3.1% |
| Engagement rate | GA4: engaged / total sessions; social: engagements / reach | Whether people pay attention | Content and page fixes | GA4 60%; social 3% |
| Branded search or share of voice | Search Console branded clicks; your mentions / category mentions | Whether people seek you by name | Brand spend, PR | 640 branded clicks |
Branded search is a directional proxy for awareness, easily nudged by promotions and seasonality. The social engagement formulas cover more variants.

The definitions this guide uses for ROI, CAC, LTV and ROAS
Three of these formulas come in two common versions, and switching between them month to month wrecks your trend lines.
ROI here runs on gross profit. Revenue-based ROI flatters low-margin businesses: in one worked example, a 320% ROI drops to 68% once a 40% gross margin is applied. Riverside’s 389% already uses $440 of gross profit per patient.
CAC here is fully loaded: ad spend plus tools, agency or freelancer fees, and staff time spent on marketing. LTV is gross margin per customer over 12 months, so Riverside’s $440 covers a patient’s first year.
ROAS stays revenue divided by ad spend, because every ad platform reports it that way. That’s how ROAS can look great while ROI is quietly underwater. At a 40% margin, breakeven ROAS is 2.5, and that covers only cost of goods sold. Tools, fees and staff time come out of what’s left.
Riverside needs both versions of CAC. Ad-only CAC is $3,000 / 30 ad-sourced patients = $100, which tells you whether Google Ads is priced well against other channels. Fully loaded blended CAC is $5,400 / 60 new patients = $90, which tells you what growth costs the whole clinic and whether it pays.
Write your definitions at the top of your dashboard, so nobody swaps formulas in March and makes the numbers look better (or worse) than they are.
Which digital marketing metrics to track for each channel
Each channel gets three or four metrics. Lead with the first column, keep the last as context, and pull every formula from the master table above.
| Channel | Lead with | Also track | Context only | Riverside note |
|---|---|---|---|---|
| SEO | Conversions from organic landing pages | Organic sessions; Search Console clicks (CTR = clicks / impressions) | Impressions, branded vs non-branded clicks | Google Business Profile calls and direction requests count here |
| Paid search and social | CPL or CPA | Conversion rate, ROAS, CTR | CPC | Google Ads CPL of $25 sits well under the 2025 Google Ads average of $70.11 |
| Conversions or revenue per email | Click rate, list growth vs unsubscribes | Open rate | Count rebooking clicks from past patients. See email marketing metrics | |
| Social | Leads or sessions from social | Profile and link clicks | Engagement rate, follower count | 20 leads a month, so judge social on brand support |
| Website | Conversion rate by landing page | Key events, engaged sessions | Engagement rate | Compare each location’s booking page |
GA4 defines engagement rate as engaged sessions / sessions. A session counts as engaged when it lasts longer than 10 seconds or meets GA4’s other criteria. Bounce rate is the share that didn’t. Search Console’s average position is a blended number, so clicks and impressions show what changed more clearly.
Use marketing benchmarks as rough comparisons, and marketing statistics for market context. Across 16,000+ U.S. campaigns, the average Google Ads CTR is 6.66% and conversion rate is 7.52% (WordStream 2025). Physicians & Surgeons, the closest category to a clinic, converts at 11.62%. For email, Mailchimp puts click rate around 2.66%, typically 1 to 5% by industry. The next section covers why open rate sits in the context column.
Why Google Ads, Meta and email all claim the same patient
Every ad and email platform grades its own homework. Each one credits a booking to itself under its own rules, so their totals overlap and can add up to more than you actually sold. Meta’s attribution settings commonly default to seven-day click and one-day view, which means a patient who merely saw an Instagram ad yesterday and then booked through Google can count for both.
Privacy changes blur the picture further. When visitors decline cookies, GA4 may fill gaps with behavioral modeling, and Google Ads can put modeled conversions in the Conversions column. Those are estimates, not bookings. Apple’s App Tracking Transparency cuts off the ad identifier for iPhone users who say no, which weakens paid social tracking. Mail Privacy Protection preloads tracking pixels, so emails get marked “opened” that nobody read.
Riverside’s month shows how this plays out. Google Ads reports 38 conversions, Meta 14 and email 9, for 61 claimed. The booking system shows 60 new patients, 30 of them from ads. Ads’ extra 8 are rebookings by existing patients plus modeled conversions. Meta’s 14 include view-through credit for people who also clicked a Google ad. Most of email’s 9 are returning patients rebooking. The 61 and the 60 sit close by coincidence, and they count different people.
The booking system or CRM wins every argument about totals. Use platform numbers to compare campaigns inside that platform, and never add them together.
Which metrics fit ecommerce, B2B, local service and SaaS
The right five to eight metrics depend on how money reaches your bank account. Here are four common models.
Ecommerce. North Star: revenue. Core metrics: conversion rate, AOV, ROAS, repeat purchase rate. The trap is a revenue-based ROAS that looks healthy on thin margins. A 4x ROAS on products with 20% gross margin loses money once shipping and returns land.
B2B and lead gen. North Star: marketing-sourced pipeline, meaning the value of opportunities where marketing generated the initial lead. Core metrics: leads, lead-to-opportunity rate, CPL next to CAC, and marketing pipeline contribution (marketing-sourced pipeline / total pipeline × 100). For example, $300,000 sourced / $1,000,000 total = 30%. Agree on the attribution rule with sales first, because different rules give different answers. The trap is celebrating a falling CPL while fewer leads ever become opportunities.
Local service. North Star: new customers booked. Core metrics: calls and form leads, lead-to-booking rate, CAC, Google Business Profile actions, reviews and rebooking. Riverside is the worked case in the final section.
SaaS and subscription. North Star: net new recurring revenue. Core metrics: trial or demo conversion, CAC, monthly churn, LTV:CAC and CAC payback period, which is CAC / monthly gross profit per customer. For example, $600 CAC / ($100 × 80% margin) = 7.5 months. The trap is an LTV built on revenue. LTV belongs on margin, as defined earlier.

You now have your model’s candidate metrics. How do you cut that list down to one page?
Six steps to a shortlist, and what goes in the monthly report
- Write the quarter’s one goal at the top of the report, for example “increase booked new patients without increasing acquisition cost.”
- Pick one outcome metric tied to money. For Riverside, that’s new patients, with marketing ROI on gross profit beside it.
- Add two or three efficiency metrics that explain the outcome: fully loaded CAC, Google Ads CPL and lead-to-patient rate.
- Add one or two leading indicators per key channel that you can act on this month. For Riverside, that’s total leads by channel.
- Add one retention or brand metric, such as email click rate plus rebookings, or branded search clicks.
- Give each metric a target, an owner and a written definition, plus the decision it triggers when it misses. “CAC over target two months running: cut the weakest campaign.”
That gives Riverside seven metrics. Run each one through the 20% test from earlier.
The monthly report holds those KPIs with month-over-month and year-over-year change, a short “what we changed and why” note, and two or three decisions for next month. Raw sessions, impressions, follower counts and platform-claimed conversion totals go in an appendix, if they appear at all. A report that can’t produce a decision is a screenshot collection.
Riverside’s one-page monthly dashboard
Copy the column headers as your blank template. Riverside’s numbers are fictional. Status is On if a metric hits target, Watch if it’s within 15%, and Act if it’s worse.
| Metric | Definition used | This month | Last month | Target | Status | Decision if off-target |
|---|---|---|---|---|---|---|
| Leads by channel | Calls + forms | 250 (Ads 120, organic/GBP 85, email 25, social 20) | 238 | 240 | On | Move budget to cheapest converting channel |
| Google Ads CPL | Ad spend / Ads leads | $25 | $27 | $28 | On | Pause weakest ad groups |
| Lead-to-patient rate | New patients / leads | 24% | 25% | 27% | Watch | Audit call handling |
| New patients | First evaluations attended | 60 | 59 | 65 | Watch | See lead-to-patient rate |
| CAC | Fully loaded cost / new patients | $90 | $92 | $95 | On | Cut weakest campaign |
| Marketing ROI | (Gross profit − cost) / cost | 389% | 371% | 300% | On | Recheck margins |
| Email clicks + rebookings | Click rate; returning bookings | 3.1%; 42 | 2.9%; 39 | 3%; 40 | On | Rework reactivation email |

Ads CPL is fine, but the second location converts 20% of its 125 leads (25 patients) while the first converts 28% (35). Riverside fixes missed calls and slow callbacks there before raising the ad budget. Social brought only 20 leads, so it stays as brand support with no added spend.
Next month, print the same page, fill in the new column, and act on what it shows.