KPIkeyperformanceindicators.com home

Independent guide. Not affiliated with EOS Worldwide.

Last reviewed August 2026

Article

Marketing KPIs for a 10 to 250 person company

A marketing team in a company this size is accountable for two things a leadership team can see: how much qualified work reaches sales, and what it cost to get there. Everything else is craft. Twelve numbers cover it, five read weekly and seven reviewed monthly, and every one of them has an obvious way to look better without the work being better.

By the Key Performance Indicators editorial teamPublished Updated 10 min read

01

What is marketing accountable for?

Two outcomes, and neither of them is traffic. Marketing is accountable for the volume of qualified work handed to sales and for the cost of producing it. A team that doubles sessions while accepted leads stay flat has done work that did not land. A team that halves cost per accepted lead has done the job even if the site looks quiet.

That framing settles most arguments a marketing manager has with a leadership team, because it names a numerator and a denominator instead of a mood. It also puts the awkward number on the board early: if leads are cheap and sales will not take them, the qualification rule is wrong.

It also separates work from results. Publishing and testing happen on your calendar. Accepted leads and sourced revenue arrive on the market's calendar, weeks later, so a weekly board made only of the second kind gives a marketing manager nothing to change on Monday.

Marketing among the departments on a weekly Scorecard: controllable inputs first, money later

02

Which five marketing numbers belong on the weekly board?

The ones a marketing manager can still change before Friday. Accepted leads, cost per accepted lead, published pieces, sends to the named list and landing page conversion rate cover the week without turning the meeting into a channel report. Each has an owner who can act on a bad reading rather than explain it.

Read them as a group. Published pieces at target with accepted leads at zero is a targeting problem. Accepted leads at target with cost per lead climbing is a bidding problem. Neither reading comes from a single row, which is the argument for keeping all five.

01

Qualified leads accepted by sales

Leads sales agreed to work, counted after acceptance rather than at capture. No formula. Leading, weekly, owned by the marketing lead. Set the band from your own trailing 13 weeks. Gamed by loosening the qualification rule until everything qualifies, which shows up two months later as a falling win rate.

02

Cost per accepted lead

All marketing spend in the period divided by accepted leads. Include salaries, contractors and tools, not just ad spend. Lagging, weekly, owned by the marketing lead. Band from your own trailing 13 weeks. Gamed by quietly redefining spend, which is why the definition should be written down once.

03

Pieces published

Finished work that went live: an article, a case study, a video, a landing page. Leading, weekly, owned by the content lead. Take the band off your own trailing 13 weeks. Gamed by chopping one piece into three, so count finished assets and agree what finished means before the first red week.

04

Sends to the named account list

Emails delivered to people you deliberately chose, not to everyone who ever downloaded a file. Leading, weekly, owned by the marketing lead. Band from your trailing 13 weeks. Gamed by padding the list to hit the send count, so read it beside the unsubscribe rate.

05

Landing page conversion rate

Conversions divided by sessions on the same page, one page at a time. Leading, weekly, owned by the marketing lead. Set the band at the median of your trailing 13 weeks on that page. Gamed by narrowing the traffic mix, since sending only branded search to a page lifts the rate and produces no new customers.

03

What belongs in the monthly marketing review?

The money and the cohorts. Seven numbers need a full month to mean anything: acquisition cost, the ratio of lifetime value to it, sourced revenue, non-brand organic leads, lead to customer conversion by cohort, unsubscribe rate and pipeline created. Reading any of them weekly produces noise and a meeting spent explaining a sample size.

Give the monthly review a different shape from the weekly meeting. Weekly is a read of rows against targets and a decision on the red ones. Monthly is one hour with a cohort table open, asking whether the leads you bought in June became customers by September.

01

Customer acquisition cost

Sales and marketing spend in a period divided by new customers won in it. Lagging, monthly, owned by the marketing lead with finance agreeing the inputs. Derive the band from your own trailing 13 weeks. Gamed by leaving sales salaries out, which halves the number and changes no decision anybody makes.

02

Lifetime value to acquisition cost

Gross margin expected from an average customer divided by what it cost to win them. Lagging, monthly, owned by the marketing lead. Band from your own trailing 13 weeks rather than a published multiple. Gamed by stretching the assumed customer lifetime, so state the lifetime you used every time you show the ratio.

03

Marketing-sourced closed-won revenue

Revenue from deals whose first touch was a marketing channel. Lagging, monthly, owned by the marketing lead. Band from your trailing 13 weeks. Gamed by the attribution rule itself, so write the rule down, date it, and change it at a quarter boundary rather than in the month it makes you look bad.

04

Non-brand organic leads

Leads from search visits that did not include your company name. No formula beyond the filter. Lagging, monthly, owned by the content lead. The band comes from your own trailing 13 weeks. Gamed by leaving brand searches in, which turns a sales team's own prospecting into a content result.

05

Lead to customer conversion by cohort

Of the leads accepted in one month, the share that had become customers by a fixed later date. Lagging, monthly, owned by the marketing lead. Use the median of your trailing 13 weeks. Gamed by measuring recent cohorts too early, since a young cohort always looks worse than a mature one.

06

Unsubscribe rate per send

Unsubscribes divided by delivered emails for one send. Lagging, monthly, owned by the marketing lead. Band from your trailing 13 weeks. Gamed by sending less often, which fixes the rate and starves the pipeline, so it only means anything read next to sends.

07

Pipeline created from marketing

Value of opportunities opened this month whose source was marketing. Lagging, monthly, owned jointly by the marketing and sales leads. Set the band from your trailing 13 weeks. Gamed by opening opportunities early at optimistic values, which inflates the row and the sales forecast at the same time.

04

Where do marketing targets come from?

Your own last thirteen weeks, in almost every case. Published benchmarks come from companies with different prices, channels and sales cycles, and a target lifted from one is a number nobody in the room can defend. Take the median of your last thirteen weeks, set the goal a little above it, and revisit at the quarter.

There are two honest exceptions. Some rows have a structural target: follow-ups past due should be zero, and a page with no conversion tracking has no rate at all. Some have an arithmetic one, like the leads a channel must produce to cover its own cost at your current win rate. Everything else is your own history.

Set a band rather than a line. A single number makes every ordinary week a win or a failure, and teams respond by managing the number. A band says what normal looks like and what is worth talking about.

05

Which marketing numbers should you drop?

Anything cumulative, anything nobody would act on, and anything a platform reports about itself. Total followers, impressions, cumulative downloads and email open rates all pass the test of being true and fail the test of being useful. None of them tells a marketing manager what to do differently next week.

Open rates deserve a specific mention. Apple's Mail Privacy Protection, offered as a prompt since iOS 15 in September 2021, loads tracking images for anyone who turned it on whether or not they read the email, still true when we checked in August 2026. Clicks and replies survive that; opens do not. Keep opens as a rough comparison between two of your own sends, never as a target.

Dropping a row is not deleting the data. Leave it in the reporting tool and take back the weekly slot. That framing removes most of the resistance, since nobody is told their work does not count.

06

What does the accepted-leads row look like written out?

Take accepted leads. The row reads: name, owner, definition, band, source. Written out it is one line in a spreadsheet and about four lines in the document behind it, and the document is the part that stops the weekly meeting turning into an argument about the formula every time the number goes red.

Name: qualified leads accepted by sales. Owner: marketing lead, because they control targeting and spend. Definition: a lead the sales lead has marked accepted in the CRM by Friday noon, counted once, with a resubmitted lead counting on its original date. Band: 18 to 24, the median of our last thirteen weeks plus a little. Source: the CRM accepted-lead report, pulled by the marketing lead each Friday.

In EOS® this row lives on the weekly Scorecard and gets read out in about ten seconds, with anything outside the band dropped into the issues list rather than debated on the spot. You do not need the vocabulary to run it that way. You need one owner, one written definition and one place the number comes from.

Start with a real question

Not sure your marketing board is measuring the right things?

Send us the numbers your marketing team reads every week. We will tell you which rows are inputs, which are results wearing an input's clothes, and which of your outcomes has no early number underneath it at all. One reply from a person, inside a business day.

A person reads every one.

Talk to us

Nothing is sent yet. The full form asks for your email, a phone number and a little context.