How to measure and analyze marketing campaign performance

1. Name the campaign's job

Pick one job: awareness, lead generation, pipeline acceleration, reactivation or a message test. The job decides the KPI, and therefore how you measure campaign success; a campaign judged against five goals can always claim success on one. For lead generation, start with building a simple lead-generation campaign.

Example: "create qualified leads for the free assessment," not "raise awareness and drive leads."

2. Choose one primary KPI and two supporting metrics

The primary KPI decides success; the two supporting metrics explain why it moved. Everything else is context.

3. Record a baseline, a target and a review window

Record what the last comparable campaign produced, set the target to beat, and fix the review date before launch. Pipeline and revenue lag behind leads, so their review window must cover your sales cycle.

Illustrative example: baseline 12 MQLs, target 15, leads judged at 4 weeks, pipeline at 90 days.

4. Set up tracking before launch

Tag every campaign link that points to your site with UTMs, make the lead action a GA4 key event and confirm the CRM stores source and campaign. Tracking added mid-campaign leaves gaps you cannot fill; see the tracking setup section below.

Example: one test lead per channel, checked in GA4 and the CRM.

5. Analyze results by segment and funnel stage

Compare each channel, audience and offer with the baseline and find the funnel stage that fell short; the analysis section below shows how.

Example (from the hypothetical scorecard below): paid social's cost per MQL is double the blended figure.

6. Decide: keep, change or stop

A report is finished when it says what to keep, change or stop and what to test next. Agree these rules before launch, apply them to the whole campaign and to each channel, and write the decision next to the numbers.

ROAS vs marketing ROI: which one to report

ROAS covers media only and suits comparing ad sets or platforms. Marketing ROI adds creative, tools and people; it is what leadership asks about. Report either only when revenue is attributable; until then, use cost per qualified lead and pipeline created.

Hypothetical example: $8,000 of attributed revenue from $2,000 of ads is a ROAS of 4.0. Add $3,000 of other costs, for a total of $5,000, and marketing ROI is ($8,000 − $5,000) ÷ $5,000 × 100 = 60% on revenue.

Both formulas use revenue, not profit, so a positive ROI can hide a loss. At a 50% gross margin, ROI on the $4,000 of gross profit is ($4,000 − $5,000) ÷ $5,000 × 100 = −20%. If you know your margin, use gross profit.

Compare against your baseline and target

Month-over-month comparisons on their own mislead, because seasonality and budgets shift. Compare with the pre-launch target and the last comparable campaign; beating last month while missing the target is still a miss.

Segment by channel, audience and offer

A blended average hides the winning segment. Split by channel, then by audience or offer where volume allows. To track cross-channel performance, use one utm_campaign value everywhere and compare channels on the same CRM outcome, not on each platform's conversion count.

Find where the funnel drops off

Follow the path from impressions to clicks, sessions, leads, MQLs, SQLs and opportunities, and calculate the conversion rate between each stage. Every funnel narrows, so look for the stage furthest below its own baseline. To optimize campaign performance while it runs, fix that stage first rather than changing everything at once.

Reconcile ad-platform numbers with CRM records

Ad platforms and the CRM count differently: attribution windows vary, and the CRM may already know the person. Pick one source of truth per metric: spend from the ad platform, sessions from GA4, leads and pipeline from the CRM. For ROAS and ROI, also name the attribution rule, such as first touch, last touch or multi-touch; the same deals give different revenue under each. I cover why ad-platform and CRM numbers disagree in a separate article.

Reading the scorecard

The campaign beat its target with 18 MQLs against 15, but email produced half of them at no media cost, so paid social's own cost per MQL ($333.33) is double the blended $166.67.

Decision: by the step 6 rules, keep the offer and the email follow-up, change the paid audience, and judge ROAS after one more sales cycle. To reuse the table, fill the Target column before launch and the Actual column on the review date.

A UTM naming convention

Use lowercase, hyphens instead of spaces and one campaign name on every channel, recorded in a shared naming sheet. Never tag internal links; it overwrites the visit's original source. Do tag external links in SMS messages, QR codes and influencer or partner posts, with a separate utm_content value for each influencer or partner.

To track email campaign performance with UTM parameters, reuse the ads' utm_campaign, set utm_medium to email and give each link its own utm_content.

GA4 key events

Mark the form submission or booking confirmation as a GA4 key event. Then send a test lead and confirm it shows in GA4's realtime report and in the CRM.

CRM source, campaign and lifecycle fields

GA4 stops at the lead; the CRM carries the rest. If source fields are patchy or lifecycle stages inconsistent, audit your CRM data first. The CRM needs:

• Source and campaign captured from UTMs through hidden form fields or source properties.
• Campaign association on contacts and deals, so pipeline can be reported per campaign.
• Lifecycle-stage dates, so stage conversion rates and time between stages can be calculated.

In HubSpot or your CRM

If your HubSpot subscription includes the campaigns tool, associate the campaign's emails, forms, landing pages and ads with a HubSpot campaign. If not, save utm_campaign to a contact property through a hidden form field and filter reports by it. Either way, put source, lifecycle-stage and deal reports on a dashboard that updates as records change. Salesforce and other CRMs work the same way with campaign records.

In Power BI or Tableau

Connect each source directly or through a staging table, schedule the refresh, and keep one KPI definitions table that every chart reads from, so cost per lead means the same thing everywhere.

Connecting sources with Make or Zapier

Where no native connector exists, a scheduled Make scenario or Zapier Zap can pull daily spend and results into Google Sheets, Airtable or a database, and send an alert when a day shows spend but zero leads.

This is the work I do in marketing reporting automation: agreeing the numbers, connecting the sources and setting up a scorecard that refreshes itself.