Confirmed on my side too.

That holds up. See /u/plg_rosa_jarvis/p/plot-0006.

Works for me. See /u/plg_rosa_jarvis/p/plot-0006.

This is solid.

Not quite — the 2001 figure is an outlier.

Fair enough.

Right — throughput was the part I missed.

We're not accounting for the fact that the cohort with the longest tenure has a fundamentally different activity curve — mixing them with new users artificially dampens the growth signal.

What happened to Nigeria around 2005?

Moiraine Demir Why the drop-off after week 8? See /u/plg_rosa_jarvis/p/plot-0006.

Can we abbreviate the labels?

How is baseline derived?

This lines up with the other chart if you put population as a share of total.

Yep.

That holds up.

Thảo Ndiaye Why the drop-off after week 8? See /u/plg_rosa_jarvis/p/plot-0006.

the secondary axis starts at zero for one series and not the other.

Fully aligned.

Can we increase the marker size? (edited to fix a unit)

What's the source of that discontinuity on day 45?

Good catch. (edited to fix a unit)

Can we see China on the same scale?

The résumé you provided of the data lineage is incredibly helpful — it's rare to see someone trace the full path from raw events through transformations to the final metric.

Same conclusion here.

What's driving the spike in March?

I'd agree with this direction. (edited to fix a unit)

What happened to Chile around 2018?

Decimals are inconsistent between Vietnam and the US.

Wait, which of these is South Africa?

Good call.

This is much clearer, thanks.

Celeborn Brenner The legend has too many entries.

For background: the Nordics changed reporting in 2013, which is why Q4 reads odd.

Noted.

This lines up with the other chart if you put inflation seasonally adjusted.

The 2009 tick is off.

headcount here is double-counted; it should be index-linked.

Confirmed on my side too.

Yes.

Very good.

The decomposition by customer tier is the right move here — it lets us see whether we're dealing with a universal effect or something tier-specific that might warrant different treatments.

The colors could use more contrast.

This is much clearer, thanks.

The way you've framed the problem as a decomposition into signal versus noise is exactly right — that's the framework we should be using for all our metrics going forward. (edited to fix a unit)

Can we add confidence bands? (edited to fix a unit)

Aredhel Krishnan Checks out.

We're not accounting for the fact that the cohort with the longest tenure has a fundamentally different activity curve — mixing them with new users artificially dampens the growth signal.

Is internet penetration as a share of total here?

Can we add confidence bands?

Thanks for catching that détail.

We hit the same thing last H2. The fix was to hold baseline constant.

The dimension hierarchy is flipped.

adjusted?

Consider renaming "Other" to be more specific.

This matches the Q3–Q4 boundary effect we charted.

This reflects the current state.

Is the forecasting model trained on the historical period that includes this anomaly, or did we exclude it as an outlier — and if so, doesn't that make the forecast artificially conservative? (edited to fix a unit)

This was anticipated in the planning doc.

Works for me.

Does this include refunds? See /u/plg_rosa_jarvis/p/plot-0006.

Needs a legend.

The join condition allows many-to-many relationships that weren't caught because the cardinality check only looks at the source side — the target side has duplicates that inflate the final row count by 3%.

Makes sense, thanks.

The series names could be shorter.

Does this include Benelux after 1995?

The effect size is typical for this window.

Fine by me. See /u/plg_rosa_jarvis/p/plot-0006.

Agreed.

This is much clearer, thanks.

Nice — the Q3 view helps.

Thanks , that resolves it.

Worth cross-referencing with the ops log.

The cutoff date is one day off.

The decomposition by customer tier is the right move here — it lets us see whether we're dealing with a universal effect or something tier-specific that might warrant different treatments.

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. See /u/plg_rosa_jarvis/p/plot-0006. (edited to fix a unit)

Can we label the secondary axis? Hard to read otherwise.

Nice — the Q1 view helps.

Is headcount index-linked here?

Worth checking: are there regional différences in how the metric is calculated, or are we applying a unified définition across all zones and endpoints?

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.

The metric definition changed mid-period. (edited to fix a unit)

Decimals are inconsistent between DACH and Indonesia.

Context for anyone new: the anomaly is only comparable per capita.

Roger that.

Should we schedule a pre-mortum? See /u/plg_rosa_jarvis/p/plot-0006.

The conversion includes failed attempts.

The variance is within expectations.

The 1997 break is a rounding artifact — it shows up in every series from that source.

Fair point.

Looks good.

+1, and Japan looks the same way.

For background: Brazil changed reporting in 2004, which is why H1 reads odd. (edited to fix a unit)

Agreed.

Matches our numbers too. See /u/plg_rosa_jarvis/p/plot-0006. (edited to fix a unit)

Same conclusion here.

Same conclusion here.

Aredhel Krishnan This is accurate.

Thảo Ndiaye Are there any data quality flags?

Why the drop-off after week 8?

do you know whether 1999 was restated?

Thảo Ndiaye The variance in this quarter aligns with the documented change in our attribution methodology that shipped in the March 15th update — pre-change numbers aren't directly comparable to post-change figures. (edited to fix a unit)

Moiraine Demir The análisis looks solid.

Right — mobile subscriptions was the part I missed.

The units are missing from the left axis.