Window
Mar 1 – Apr 15
46 days
Ground truth
Manual daily reconciliation from orders DB
Last updated
Apr 16 20:15 UTC
Regenerate by re-running the sales-coverage script
GA4 coverage (all 46d)
88.1%
Transactions captured vs sheet
PostHog coverage (Apr only)
91.4%
/checkout/completed pageviews vs sheet
Latest sheet day
50
orders on Apr 15
Latest sheet revenue
$10,264
Apr 15

How reliable is each tool?

Coverage tells you the average. Reliability tells you whether a single day's number is trustworthy.
Metric GA4 PostHog What it means
Days of data 45 15 PostHog installed Apr 1
Overall coverage 88.1% 91.4% Total tool orders ÷ total sheet orders
Median daily coverage 88% 83% Typical day's capture rate
Daily swing (std dev) ±19% ±20% Lower = steadier. ±10% or less is trustable.
Worst / best day 18% / 129% 54% / 126% Range of daily capture rates
Days within ±10% of truth 31% 40% How often you can read the day's number as-is
Days within ±20% of truth 73% 53% Ballpark accuracy
Trend correlation vs sheet 0.81 0.92 1.0 = perfect trend match. 0.90+ = good. Below 0.70 = noisy.
Mean absolute error (MAPE) 17% 18% Average % off per day, regardless of direction
Verdict. Both tools land around 88-91% of real orders on average, but daily swings are ±19-20%, so a single day's figure can be off by a lot in either direction (GA4's range on this window is 18% to 129% of actual orders, PostHog's 54% to 126%). For day-to-day reporting, neither is trustworthy without the sheet to check against. PostHog's trend correlation (0.92) is stronger than GA4's (0.81), so if you want to read the direction of sales from a dashboard, PostHog is the better signal today. The reverse-proxy + event-based trigger fixes should tighten both towers.

Period summary

Totals per period with coverage %
Period Sheet orders GA4 tx GA4 coverage PostHog PH coverage
Mar 1-15 (pre-drop) 618 469 75.9%
Mar 16-31 (post-migration) 526 478 90.9%
Apr 1-15 (current) 559 515 92.1% 511 91.4%

Daily breakdown

Newest first · 46 rows
Date Sheet orders Sheet revenue GA4 tx GA4 coverage PostHog PH coverage
Apr 15 Wed 50 $10,264 49 98.0% 52 104.0%
Apr 14 Tue 49 $10,190 31 63.3% 38 77.6%
Apr 13 Mon 43 $9,512 50 116.3% 54 125.6%
Apr 12 Sun 28 $6,432 29 103.6% 23 82.1%
Apr 11 Sat 23 $4,950 20 87.0% 19 82.6%
Apr 10 Fri 41 $9,354 36 87.8% 42 102.4%
Apr 9 Thu 44 $9,088 46 104.5% 47 106.8%
Apr 8 Wed 47 $9,724 42 89.4% 46 97.9%
Apr 7 Tue 49 $10,049 48 98.0% 50 102.0%
Apr 6 Mon 38 $8,800 49 128.9% 39 102.6%
Apr 5 Sun 19 $4,086 9 47.4% 11 57.9%
Apr 4 Sat 25 $6,393 17 68.0% 16 64.0%
Apr 3 Fri 42 $9,140 35 83.3% 33 78.6%
Apr 2 Thu 33 $6,286 23 69.7% 26 78.8%
Apr 1 Wed 28 $5,369 31 110.7% 15 53.6%
Mar 31 Tue 32 $7,567 27 84.4%
Mar 30 Mon 37 $7,301 32 86.5%
Mar 29 Sun 18 $3,821 13 72.2%
Mar 28 Sat 22 $4,302 19 86.4%
Mar 27 Fri 30 $5,811 25 83.3%
Mar 26 Thu 37 $8,695 37 100.0%
Mar 25 Wed 32 $6,600 29 90.6%
Mar 24 Tue 51 $9,721 46 90.2%
Mar 23 Mon 31 $5,947 34 109.7%
Mar 22 Sun 24 $4,663 16 66.7%
Mar 21 Sat 25 $5,278 22 88.0%
Mar 20 Fri 39 $8,472 38 97.4%
Mar 19 Thu 24 $5,400 26 108.3%
Mar 18 Wed 38 $7,364 34 89.5%
Mar 17 Tue 48 $9,833 46 95.8%
Mar 16 Mon 38 $7,852 34 89.5%
Mar 15 Sun 39 $9,324 31 79.5%
Mar 14 Sat 32 $7,665 26 81.3%
Mar 13 Fri 27 $5,576 30 111.1%
Mar 12 Thu 34 $6,062 28 82.4%
Mar 11 Wed 49 $10,726 47 95.9%
Mar 10 Tue 55 $11,404 32 58.2%
Mar 9 Mon 43 $8,393
Mar 8 Sun 38 $7,832 7 18.4%
Mar 7 Sat 31 $6,188 20 64.5%
Mar 6 Fri 33 $7,787 27 81.8%
Mar 5 Thu 47 $10,839 37 78.7%
Mar 4 Wed 53 $13,225 43 81.1%
Mar 3 Tue 49 $11,927 46 93.9%
Mar 2 Mon 60 $11,800 68 113.3%
Mar 1 Sun 28 $5,535 27 96.4%
How to read this. The sheet is ground truth — it's reconciled manually from the orders database. GA4's purchase trigger is DOM-based ("order message does not contain needs payment") and misses purchases when that DOM isn't set as expected. PostHog uses a pageview of /checkout/completed as a purchase proxy because native e-commerce events aren't wired yet. PostHog was installed on April 1, so March rows are empty. Coverage < 80% is flagged red.