- Home
- Resources
- Blog
- Strategy & Data
- Marketing KPIs: which ones to track channel by channel (and which to ignore)
Marketing KPIs: which ones to track channel by channel (and which to ignore)
Alessandro Papa18 min read

IN SHORT
A marketing KPI is a number that, if it changes, makes you take a decision. One or two per channel are enough: non-brand clicks and leads from organic search for SEO, cost per qualified conversion for Google Ads, cost per result for Meta, cost per qualified lead for LinkedIn, clicks and unsubscribes for email, margin and returning customers for e-commerce. Followers, impressions and open rate remain signals. It all fits in a one-page monthly dashboard.
Topic of our service Analytics & Data · industry Running and events
The end-of-month meeting. The report has thirty-two slides: 1.2 million impressions on Meta, LinkedIn followers up 9%, a newsletter open rate of 48%, an improved average position on Google. The owner listens and at the end asks the only question that matters: “So, what do we do with the budget next month?” Nobody can answer, because none of those numbers had been chosen to answer that question.
The short answer: a marketing KPI is a number that, if it moves, makes you take a decision. One or two per channel are enough, tied to leads, sales or margin, plus a few control metrics that explain why the KPI moved. Followers, impressions, open rate and average position aren’t KPIs: they’re signals to be read alongside a result. The rest fits in a one-page dashboard, read once a month on a closed period, with a written decision for every row.
We’re not talking here about how to calculate the overall return on investment: this is a map of the numbers to watch, channel by channel. On how to read the platforms’ ROAS there’s Gaetano’s guide, why ROAS can mislead you.
What’s the difference between a KPI and a metric?
A metric is any number a platform shows you. A KPI (key performance indicator) is a metric chosen because it measures the result you care about and because, if it changes, you know what to do. We work on three levels:
| Level | Question | Examples | Who looks at it |
|---|---|---|---|
| Business KPIs | Is marketing bringing in customers and margin? | New customers, cost per customer acquired, margin after ad spend | Owner, management |
| Channel KPIs | Is this channel doing its part? | Cost per qualified lead on LinkedIn, leads from organic search | Marketing manager |
| Diagnostic metrics | Why did the KPI move? | CTR, CPM, frequency, impression share | Whoever runs the channel |
The most common mistake is promoting a diagnostic metric to KPI. An ad with a sky-high CTR that brings in the wrong leads is a problem, not a success.
How do you choose a good marketing KPI?
A good KPI passes four checks:
- It leads to a decision. Write down what you’d do if it went up by 20% and what you’d do if it went down by 20%. If the answer is “nothing” in both cases, it isn’t a KPI.
- It’s close to revenue. A click is far away, a qualified lead is close, a contract is the result.
- It always comes from the same reliable source. If you take cost per lead from Meta one month and from the CRM the next, you’re comparing two different things. If the platforms and the CRM don’t add up, sort out the data first: the article on how to check data from click to sale explains how.
- It’s read over a period that suits the sales cycle. For an e-commerce store a month is enough. In B2B with six-month negotiations, cost per contract is read over the quarter, and within the month you look at the step before: the qualified lead.
Watch the vocabulary. In Google Analytics 4 the important actions are called key events, while “conversions” are the ones used by Google Ads: according to Google’s guide to key events and conversions, a conversion originates from a key event and in Analytics only appears in the Advertising section, not in the standard reports. In the dashboard it pays to use the right words.
Which KPIs should you track channel by channel?
The table we start from, to be adapted to the goal.
| Channel | Main KPI | Diagnostic metrics | Don’t use as a KPI | Where to read it |
|---|---|---|---|---|
| SEO | Non-brand organic clicks; key events from organic search | Impressions, CTR by query, pages with the most clicks | Site average position, total traffic | Search Console, GA4 |
| Google Ads | Cost per qualified conversion (or conv. value/cost) | Impression share lost to budget and to rank, search terms | Clicks, optimisation score | Google Ads, CRM |
| Meta ads | Cost per result, if the result is a lead or a purchase | Frequency, CPM, link CTR | Reach, likes, engagement | Ads Manager, CRM |
| Cost per qualified lead; target accounts reached | CTR, form completion rate, audience penetration | Followers, impressions | Campaign Manager, CRM | |
| Unique clicks and conversions from email; unsubscribes | Bounces, spam complaints | Open rate | Email platform, GA4 | |
| E-commerce | Margin after ad spend; new customers; repeat customers | Conversion rate, average order value, checkout abandonment | Sessions, page views | GA4, management system |
Let’s go through the rows one by one, using the names in use on the platforms as of October 2026.
Which KPIs should you track for SEO?
The main KPI is organic clicks on non-brand searches, that is searches without the company name, alongside the key events brought by organic search in GA4. Brand clicks tell you how well known you are; non-brand clicks tell you how much SEO gets you found by people who don’t know you.
Search Console’s Performance report has four metrics: clicks, impressions, CTR and average position, as Google points out in its deep dive into performance data. The guide to Search Console metrics specifies that position is an average calculated across all impressions. It follows that a site that starts appearing on page two for lots of new searches sees its average position get worse just as it grows. That’s why the site’s average position isn’t a KPI, while that of a few important queries, read over time, is a good diagnostic metric.
According to Google’s guide to AI features in Search, clicks and impressions from AI Overviews and AI Mode are counted in the Performance report together with web search. If impressions rise and clicks don’t, part of the explanation may lie there.
For diagnosis you need CTR by page and the queries with lots of impressions and few clicks: they’re the first list of titles and descriptions to rewrite. To connect searches with what happens next on the site, Search Console can be linked to GA4: we explain how in the guide to the GA4 reports an SME needs.
Which KPIs should you track for Google Ads?
The main KPI is cost per qualified conversion: how much you spend for an action that has real value, not for a phone click or a visit to the contact page. For an e-commerce store it becomes conv. value/cost, ROAS in Google Ads language.
“Qualified” is the word that matters. If the primary conversion is the submission of any form, cost per conversion falls even when job applications and supplier pitches come in. The next step is to feed CRM statuses, such as qualified lead or opportunity, back into Google Ads: the article from lead to contract explains how.
The most useful diagnostic metrics:
- Search impression share lost to budget and lost to rank. It tells you whether you’re losing searches because the budget runs out or because the ads are losing the auction. The solutions are opposite: more budget in the first case, better ads, pages and bids in the second.
- Search terms report, with the real searches that triggered the ads.
- CTR by ad, useful for comparing messages, not for judging the campaign.
Don’t use as KPIs: clicks, which grow when you broaden keywords and lower the quality of traffic, and the optimisation score, which measures how closely the account follows the platform’s recommendations, not how well it performs.
Which KPIs should you track for Meta ads?
The main KPI is cost per result, provided the result is a lead, a purchase or a sign-up. In Ads Manager the “result” changes with the campaign objective: the cost per result of a traffic campaign and that of a leads campaign can’t be compared.
The diagnostic metrics:
- Frequency: how many times on average the same person has seen the ad (impressions divided by reach). If it rises while cost per result gets worse, the audience is often tired of the creatives. There’s no threshold that applies to everyone.
- CPM, cost per thousand impressions: it explains cost changes that don’t depend on the ads.
- Link CTR and performance by creative: on Meta it’s the ads that find the audience, so “which video works?” matters more than “which audience?”.
There’s a 2026 change to bear in mind when comparing with the previous year. Since the end of March 2026, with a gradual rollout, for campaigns optimised for conversions on the website or in store Meta only counts a click on the link as a click. Conversions that arrive after shares, saves or likes move to “engage-through” attribution, which has a default window of one day instead of seven, as explained in Meta’s announcement on click attribution. A drop in attributed results may depend on the yardstick, not on the campaigns.
Reach, likes and engagement aren’t KPIs: they grow easily and are almost never linked to sales.
Which KPIs should you track for LinkedIn?
The KPI for campaigns is cost per qualified lead, calculated on the leads the salesperson recognises as good, not on those counted by Campaign Manager. If you run account-based marketing, the second KPI is target account coverage: how many companies on your list have seen or clicked on the ads.
In Campaign Manager, for diagnosis we look at:
- Average CTR: according to LinkedIn’s guide to engagement metrics, it’s paid clicks divided by impressions. Total engagements can exceed clicks, because the same person can interact several times.
- Form completion rate in campaigns with Lead Gen Forms: if it’s low, the form asks too much or the offer isn’t convincing.
- Audience penetration: LinkedIn’s delivery metrics describe it as a rough estimate only, visible with audiences of at least 300 members, which in some cases, for example in campaigns on the LinkedIn Audience Network, can exceed 100%.
- Conversions: the conversion tracking metrics count actions taken after seeing or clicking an ad, and some may be estimated with models for privacy reasons. The final count is done in the CRM.
For the company page without advertising, the number of followers isn’t a KPI; their make-up by role, industry and company size is, as we explain in the article on how to make your LinkedIn company page work. Add to that clicks through to the website, read in GA4 with UTM parameters, and the enquiries that come in from there.
In our experience, decision-makers often don’t click: they read, remember the name and months later search for the company on Google. That’s why, in quarters with active campaigns, we also look at brand searches in Search Console. It isn’t proof, it’s a clue worth noting.
Which KPIs should you track for email marketing?
The main KPI is unique clicks, together with what people do next on the site (key events or purchases in GA4, with consistent UTMs). Alongside it goes the unsubscribe rate, which tells you whether you’re sending too much or the wrong thing.
Open rate is no longer reliable. With Apple’s Mail Privacy Protection, Apple Mail downloads email images in advance, including the pixel that counts opens, so many emails show as opened even if nobody has read them. Mailchimp writes that opens from Apple Mail are inflated and that clicks and purchases are more solid signals. Brevo explains that since 6 February 2025 its open rate has included opens detected by the system, including Apple’s, and since 8 July 2025 bot activity too, which inflates both opens and clicks. The same send can therefore show different rates depending on the platform and the date, and comparisons with past years don’t hold up.
For diagnosis you need bounces, which tell you when the list needs cleaning, and spam complaints: in its sender guidelines Google asks senders to keep them below 0.1% in Postmaster Tools and never to reach 0.3%.
Which KPIs should you track for an e-commerce store?
The KPI that counts most is margin after ad spend: revenue minus cost of goods sold, shipping, fees and advertising. Revenue on its own can be pushed up with discounts that wipe out the profit. Alongside it go new customers and repeat customers: a shop that lives only on new customers pays for acquisition all over again every month.
For an overall reading of spend there’s MER, the ratio between total revenue and total ad spend, calculated with data from the management system and independent of platform attribution. It’s explained in the same article on ROAS mentioned at the start.
The diagnostic metrics are conversion rate by device and channel (how to read it is in the guide to e-commerce conversion rate), average order value, checkout abandonment and the E-commerce purchases report in GA4’s Monetisation section. Sessions and page views aren’t KPIs: more traffic with the same orders just means more paid visits.
Which vanity metrics should you ignore?
Vanity metrics grow easily, look good in the report and don’t change any decision. They should be moved out of the KPI slot and read, if necessary, as diagnostics.
| Vanity metric | Why it misleads | What to look at instead |
|---|---|---|
| Followers | Grows with competitions, invitations and off-target audiences | Follower make-up, clicks to the website |
| Impressions and reach | Can be bought: more budget, more impressions | Cost per result, brand searches |
| Likes and engagement | Reward light content, not content that sells | Comments from people in the target audience, link clicks |
| Email open rate | Inflated by Apple Mail privacy and bots | Unique clicks, conversions from email |
| Site average position | Gets worse when the site appears for more searches | Non-brand clicks |
| Sessions and page views | Also grow with traffic that doesn’t buy | Key events by channel |
| Sum of platform conversions | Each platform claims the same sale | Leads and sales in the CRM or management system |
A quick test: if a metric goes up and nobody in the company has more customers, more good leads or more margin, it was vanity.
How do you build an essential monthly dashboard?
An essential dashboard fits on one page and has three blocks: business KPIs at the top, one row per channel in the middle, decisions at the bottom. Ten to twelve numbers in all. The rules we follow:
- Every number has two comparisons: the same month the year before, to remove seasonality, and the average of the last three months, to see the trend.
- Every number has a written source, for example “qualified leads: CRM, status as of the 15th of the following month”.
- You read a month that closed at least two weeks ago, because the platforms complete conversion data in the following days and the CRM needs time to qualify leads.
- Each channel has one KPI and at most two diagnostic metrics. The rest is only opened if the KPI has moved.
- At the bottom there are three decisions: what we change, who does it, how we measure it. Even “we’re not changing anything” counts, if it’s written down.
An example monthly dashboard
A B2B services company with a website, Google Ads, LinkedIn Ads and a newsletter. The figures are made up for illustration.
| Row | October | October last year | 3-month average | Note |
|---|---|---|---|---|
| Qualified leads (CRM) | 22 | 15 | 19 | Growth mainly from organic search |
| Opportunities opened (CRM) | 7 | 5 | 6 | |
| Cost per qualified lead, all channels | €182 | €240 | €205 | Total spend divided by qualified leads |
| SEO: non-brand clicks | 1,840 | 1,310 | 1,700 | Two new articles among the pages with the most clicks |
| Google Ads: cost per qualified conversion | €160 | €210 | €175 | Impression share lost to budget at 35% |
| LinkedIn Ads: cost per qualified lead | €410 | n/a | €455 | Form completion rate falling |
| LinkedIn: target accounts reached | 118 of 200 | n/a | 104 | |
| Email: unique clicks | 186 | 160 | 175 | Unsubscribes stable |
The month’s decisions: increase the budget for search campaigns on the main services, given that they lose a third of their impressions to budget; remove two fields from the LinkedIn form; update the two pages with the most impressions and a low CTR.
Notice what’s missing: impressions, followers, open rate, sessions. They wouldn’t change any of the three decisions. If cost per lead on LinkedIn got worse, that’s when you’d open CTR, frequency and completion rate to understand why.
Where to build it
For the first few months, a spreadsheet filled in by hand is enough. With three or four channels it’s worth having a dashboard that pulls in the data automatically, for example in Looker Studio. For the clients whose campaigns we manage we use RunAlytics, our dashboard that puts Meta, Google Ads, TikTok and LinkedIn in the same table, updated every four hours, down to the individual ad. Business KPIs, however, come from the CRM and the management system: no advertising platform knows how many contracts you’ve signed.
The rhythm: every week you check spend and anomalies, every month you read the dashboard, every quarter you review how the budget is split between channels.
Mistakes to avoid
Choosing KPIs after seeing the data. You end up highlighting the numbers that went up. KPIs are written down beforehand, together with the goal.
Using the same cost per lead for every channel. A lead from Google Ads and one from LinkedIn can be worth very different things. The comparison is made on the qualified lead and, where possible, on the contract.
Adding up the platforms’ conversions. Google, Meta and LinkedIn can all claim the same sale. The CRM gives you the true total.
Comparing periods that can’t be compared. October with September, or one month before and one after a change in attribution. Use the same month the year before and note down every change in the rules.
Reacting to every fluctuation. With small volumes, three leads more or fewer is noise. Before acting, look at the average of the last few months.
Checklist for your dashboard
- Each channel has one main KPI and at most two diagnostic metrics.
- Each KPI has a written source and reading date.
- Qualified lead and customer have a definition agreed with sales.
- Business KPIs come from the CRM or management system.
- Email open rate doesn’t appear among the KPIs.
- At the bottom there are three decisions, each with an owner.
Want a dashboard with the right KPIs for your business?
We start from the decisions you make every month and choose with you the numbers you need, channel by channel. Then we fix the tracking where needed, bring campaigns, website and CRM together with the same definitions and build a one-page dashboard, which we walk you through once a month. It’s the work of our analytics and data service.
You can book a call with Federico or write to us via the contact form.
Frequently asked questions
What are the most important marketing KPIs?
The ones closest to revenue: new customers, cost per customer acquired, qualified leads and, for an e-commerce store, margin after ad spend. Each channel needs one or two KPIs chosen according to the goal; the rest are metrics that explain why the KPIs move.
How many KPIs should a marketing report have?
Few: a monthly dashboard works with ten to twelve numbers, three or four for the business and one or two per channel. If a number never leads to a decision, take it out of the report and consult it only when needed.
Is email open rate still a KPI?
No. Apple’s Mail Privacy Protection makes many emails that were never read show as opened, and some platforms also count opens by bots. It can be used to compare similar sends, but the KPIs for email are clicks, conversions and unsubscribes.
What’s the difference between KPIs and OKRs?
A KPI continuously measures how an activity is going. An OKR (Objectives and Key Results) is a goal for a defined period, with a few key results that tell you whether it’s been reached. Key results are often KPIs with a value to hit by a certain date.
How often should marketing KPIs be reviewed?
The values are read every month; the choice of KPIs is reviewed when goals or channels change, usually once a year. Changing them more often makes it impossible to compare periods.
FREE GUIDE · 28-PAGE PDFSold-out running races: positioning, calendar, campaigns and sponsors in 20 weeksThe method we used in 2026 to take ten running races to a sell-out: from setting entry fees to the five-month calendar, from the registration page to the campaigns, through to sponsors and the story told the day after. With six appendices to fill in.Read it or download the PDF























