The text mentions inflation adjustment, but I don't see it applied. Should we verify the methodology?
Húrin Harnesh That's last week's snapshot.
How are we handling the missing weeks?
Same as before.
Layla Greenleaf Worth documenting the assumptions.
The 2023 tick is wrong.
That's the way to do it.
I really appreciate how you've documented the edge cases here — the distinction between deleted records and deactivated ones matters more than most people realize in retention calculations.
Need to check the access permissions.
raised this on the other chart too, same a coverage gap.
Missing Benelux data after 2005
Following up on the previous investigation.
I see it.
raw_metric should probably be in constant terms.do you know whether 2000 was restated?
That is a coverage gap, not a real move.
The bars need a bit more padding.
The units are missing from the y axis.
Stale.
Nicely done.
Fixed.
The 1997 break in the anomaly
Scale is off.
Minor: life expectancy is stale in the tooltip.
The way you've structured the lookback window to exclude the migration period is clever — it gives us clean apples-to-apples comparisons without having to build a manual adjustment factor. See /u/vgt_layla_greenleaf/p/plot-0024.
We should loop in the data infrastructure team to understand whether the missing data from the 8th through the 12th was a known incident or an undiscovered gap — that could affect how we handle it in future work.
Title says Q3 but the data runs longer.
The units are missing from the legend.
Can we label the left axis? Hard to read otherwise. (edited to fix a unit)
Kai Gaidin Does this include test users? See /u/vgt_layla_greenleaf/p/plot-0024.
Checks out.
I'm convinced by your argument that the regional patterns are statistically significant rather than noise — that's a crucial insight for how we should interpret the trend.
I'm convinced by your argument that the regional patterns are statistically significant rather than noise — that's a crucial insight for how we should interpret the trend.
Not sure what
Region is measuring.The units are missing from the gridlines.
Which vintage?
We hit the same thing last Q3. The fix was to hold
Region constant.This is much clearer, thanks.
Forgot to remove the test environment data.
Confused by the 1996 value — is that a coverage gap?
the y axis looks inverted.
How is
raw_metric derived?Does this include Iberia after 2003?
This feels like the kind of transparent, defensible analysis that would hold up under scrutiny if we ever needed to justify the numbers to external auditors.
Nice — the Q2 view helps.
This was flagged in the postmortem.
That matches what I had.
Why does the Nordics dip in 1995?
Fixed.
For background: South Africa changed reporting in 2016, which is why H1 comes across as odd.
This is inverted — the Nordics was restated in 2005.
I'd agree with this direction.
Correcting myself: retention is per capita, so the comparison holds.
What's the methodology for "active"? See /u/vgt_layla_greenleaf/p/plot-0024.
How is
value derived?Consistent with published forecasts.
Sorry, lost me. What is the baseline?
Confirmed on my side too.
Agreed.
Can we use consistent units throughout?
Matches our numbers too. (edited to fix a unit)
Nicely done.
Nice — the Q3 view helps.
Yes.