Northwind · Weekly campaign analysis

May 5–11, 2026 · Prepared by Clarai · Every number verified against source

Client:
Northwind (fictional)
Stack:
Meta Ads · Google Ads · GA4

Executive summary

Sessions dropped 34% week-over-week on the back of a softening Meta campaign, but blended ROAS held above target because Google Search picked up the slack. Two paused Meta ad sets need attention before the weekend; Google Shopping is the channel to put more into.

Recommended actions

The most important part: what to actually do this week, in priority order.

  1. This week

    Re-enable the two paused Meta prospecting ad sets and refresh their creative: the existing assets have been live for 28 days, well past Northwind's fatigue threshold.

  2. This week

    Shift €600 of weekly budget from Meta retargeting (over-served) to Google Shopping (limited by feed coverage, not by budget cap).

  3. Next 2 weeks

    Brief the design team for two new Meta creative concepts targeting the new-customer segment, since AOV is being propped up by returning traffic.

  4. Watch

    Track new-customer share of revenue daily. If it stays below 42% next week, treat Meta prospecting as a deeper problem, not just a temporary dip.

Key metrics

Verified data
Sessions

How we know this

Sessions

GA4 · sessions, May 5–11 vs the prior week. Counted from the raw event stream.

Verified against source

5,400
−34.1%· prev 8,200
Conversions

How we know this

Conversions

GA4 · key-event conversions, May 5–11 vs the prior week.

Verified against source

182
−26.3%· prev 247
Revenue

How we know this

Revenue

GA4 · purchase revenue, May 5–11 vs the prior week.

Verified against source

€41,820
−14.0%· prev €48,600
Blended ROAS

How we know this

Blended ROAS

Total revenue ÷ total ad spend across Meta and Google.

Verified against source

3.42
−3.7%· prev 3.55
Avg. order value

How we know this

Avg. order value

Revenue ÷ conversions.

Verified against source

€229.78
+16.8%· prev €196.76
Meta ROAS

How we know this

Meta ROAS

Meta Ads account · purchase ROAS, pulled directly from the ad platform.

Verified against source

2.10
−44.7%· prev 3.80
Google Ads ROAS

How we know this

Google Ads ROAS

Google Ads account · conversion value ÷ cost.

Verified against source

4.85
+42.6%· prev 3.40
Cost per conversion

How we know this

Cost per conversion

Total ad spend ÷ conversions.

Verified against source

€67.20
+44.4%· prev €46.55

Channel breakdown

ChannelSpendRevenueROAS
Meta Ads

How we know this

Meta Ads

Meta Ads account · spend and revenue for the week. ROAS = revenue ÷ spend.

Verified against source

Spend held flat, ROAS dropped from 3.80. Two ad sets paused mid-week without replacement creative.

€7,420€15,5802.10
Google Search

How we know this

Google Search

Google Ads account · Search campaigns. ROAS = conversion value ÷ cost.

Verified against source

Strong week. Branded queries up 18%, generic prospecting holding steady.

€3,100€15,0304.85
Google Shopping

How we know this

Google Shopping

Google Ads account · Shopping campaigns. ROAS = conversion value ÷ cost.

Verified against source

Best-performing channel of the week. Limited by inventory feed, not budget.

€2,180€11,2105.14

What changed

Verified data

Meta is the entire story of the down week. ROAS fell from 3.80 to 2.10 and revenue from the channel was down 41%, dragging the overall numbers down even though Google made up most of the gap on a much smaller spend.

Two prospecting ad sets in Meta were paused on Tuesday and never switched back on, which removed about 38% of Meta's weekly impression volume.

Google Search filled the gap. Conversions on Search were up 14% week-over-week, and Google Shopping had its strongest week since launch.

Why it changed

AI interpretation

Clarai's reading of the verified numbers above. Treat these as informed hypotheses, not as additional facts.

This most likely comes down to reach, not quality. The creative and audience are performing as before, and half the prospecting set has been switched off for three days.

AOV likely climbed because the remaining traffic was mostly returning customers rather than new ones. That lifts revenue per visit, but it can hide that new-customer acquisition is down.

Google's tailwind appears partly seasonal (week 19 historically out-performs week 18 for Northwind) and partly the result of the new Shopping feed shipped on April 28.

How we verified this

Every figure in this analysis was computed in code from the raw platform data, then re-checked against the source after generation. The sessions, conversions, and revenue numbers reconcile to GA4. The spend and platform-ROAS figures reconcile to the native ad-platform reports. No AI hallucinations. No rounded-up wins.

Northwind is a fictional client built for illustration. No real campaigns, customers, or revenue figures are represented here.

Want one of these for your own campaigns?

Book a short call and we'll walk you through exactly how Clarai would build this for your clients.

← Back to home