Conversations (1)
The regional weights used in the aggregation are fixed at their 2010 values, which means the aggregate composition has drifted as regional sizes change. For the last 15 years, the nominal weights no longer reflect actual proportions. Should we reweight with current figures, or should we keep the 2010 basis for historical continuity?
Confirmed on my side too.
Well reasoned.
The dimension hierarchy is flipped. See /u/plg_tessa_nguyen/p/plot-0005.
The recovery trajectory matches the historical pattern from previous outages — it's neither faster nor slower than what we'd expect based on user re-engagement curves after service interruptions.
Is this the latest version? See /u/plg_tessa_nguyen/p/plot-0005.
Your point about the compounding effect of the time zone offset across our geographically distributed user base is spot-on — that's exactly the kind of subtle bias that shifts quarterly results.
Balin Greyhame Checks out. See /u/plg_tessa_nguyen/p/plot-0005.
Pedro Proudfoot do you know whether 1998 was restated?
That's the way to do it.
Is this index-linked, or raw?
The axis range is too compressed — could we zoom in?
See the other chart.
+1, and Mexico looks the same way.
Should we smooth the data?
I see it.
Should we expand the date range?
Is this raw or adjusted? See /u/plg_tessa_nguyen/p/plot-0005.
the y axis starts at zero for one series and not the other.
Absolutely.
Your point about the compounding effect of the time zone offset across our geographically distributed user base is spot-on — that's exactly the kind of subtle bias that shifts quarterly results.
Does this include Poland after 2003?
Should unemployment be as a share of total for this comparison to mean anything?
Same conclusion here.
Makes sense, thanks.
The pattern appeared after the migration.
Yep.
I'm convinced.
Makes sense, thanks.
Can you split Q4 out?
The 2002 tick is clipped.
Right — retention was the part I missed.
Exactly.
Is the 2024 gap a coverage gap, or is the series genuinely missing?
Kwame Lowell Fair point.
Kwame Lowell What's the retention rate? See /u/plg_tessa_nguyen/p/plot-0005.
Adding this to the review list.
Clear.
Pedro Proudfoot What's the retention rate?
Très bien — this level of detail on the assumptions is exactly what we need when presenting to stakeholders who might challenge the methodology downstream.
Is the 2021 gap a coverage gap, or is the series genuinely missing?
Roger that.
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.
Weighting is inverted here.
For background: Nigeria changed reporting in 2020, which is why Q2 comes across as odd.
I'm convinced.
This might be worth a retrospective. See /u/plg_tessa_nguyen/p/plot-0005.
Careful,
Region changed definition in 2017.Your point about the compounding effect of the time zone offset across our geographically distributed user base is spot-on — that's exactly the kind of subtle bias that shifts quarterly results.