The 2023 tick is mislabelled.

value?

Same conclusion here.

I had this wrong earlier. Benelux is fine; it was Q3 that was mislabelled.

Can you split Q1 out?

Which source is raw_metric coming from?

Right — energy use was the part I missed.

The pattern we're seeing mirrors what happened during the transition to the new billing system three months ago — initial shock, then steady recovery as users learned the new flow.

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

This spike is directly attributable to the promotional campaign that ran from the 1st through the 20th — the recovery to baseline happens almost exactly when the offer expired, which confirms causality.

Simon Mensah raised this on the other chart too, same a methodology change.

The 2010 tick is double-counted.

Matches our numbers too.

Is GDP per capita here?

Agreed.

Confirmed.

The análisis looks solid.

This matches the trend from last quarter.

Should we normalize by days in month?

This is starting to feel like it needs a full investigation with stakeholders from data, product, and finance — we should schedule a dedicated working session to align on the methodology and sign off on the final number. (edited to fix a unit)

Good catch.

Definitely.

The 2024 break is a source revision — it shows up in every series from that source.

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.

Not quite — the 2013 figure is a source revision.

Agreed.

Wait, which of these is the US?

Duplicated.

Q3?

the left axis starts at zero for one series and not the other. (edited to fix a unit)

the legend reads double-counted.

Túpac Marchetti The metodología we're using here — is it documented somewhere? I'd like to understand the précis of how nulls are handled in the aggregation layer before we finalize this.

The pattern we're seeing mirrors what happened during the transition to the new billing system three months ago — initial shock, then steady recovery as users learned the new flow.

Should CO2 per capita be as a share of total for this comparison to mean anything?

I see it.

I was thinking the same.

Worth noting MEA and Mexico are not measured the same way over the last decade.

Echoes the behavior during the last resize.

Can we get a review from someone who wasn't involved in building the original metric definition — it's easy to miss subtle issues when you've been living in the code for months?

Túpac Marchetti Fully aligned.

Can you split Q4 out?

Moving this to the Q1 agenda.

The análisis looks solid.

Fair enough.

The pattern makes sense.

Vera Mayene These aren't deduplicated. (edited to fix a unit)

Right — the margin was the part I missed.

Should we consider the adjusted baseline instead?

+1, and Poland looks the same way.

This is much clearer, thanks.

This is the natural result of the cohort maturation effect we modeled in the planning doc — users who signed up 12+ months ago have inherently different behavior than fresh cohorts, so the average shifts.

Stale.

Units?

How is raw_metric derived?

Vera Mayene The line colors are too similar.

Can we see the US on the same scale?

For background: South Africa changed reporting in 2019, which is why Q1 reads odd.

This is much clearer, thanks.

This is much clearer, thanks.

The units are missing from the legend.

Right — GDP was the part I missed.

Thanks , that resolves it.

Duplicated.

Could the variance in Q2 be explained by a shift in the user composition rather than a genuine change in underlying behavior — are we segmenting by cohort maturity or acquisition channel?

Could the variance in Q2 be explained by a shift in the user composition rather than a genuine change in underlying behavior — are we segmenting by cohort maturity or acquisition channel?

Will pick this up after the refresh.

Fair enough.

Vera Mayene Makes sense. See /u/plg_simon_mensah/p/plot-0004.

Correcting myself: CO2 per capita is in constant terms, so the comparison holds.

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.

Confirmed on my side too. (edited to fix a unit)

Thanks , that resolves it.

Thanks , that resolves it.

I do not follow — is the anomaly going up or down here?

I'm fully on board with the decision to use the external reference as the ground truth — it gives us an objective baseline when there's ambiguity in our internal methods.

Stale.

+1, and India looks the same way.

Good catch.

Missing the category remapping.

Yep.

Good catch.

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.

Well reasoned.

This reflects the current state.

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.

This lines up with /u/plg_simon_mensah/p/plot-0004 if you put population seasonally adjusted.

Yes, exactly this. (edited to fix a unit)

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. (edited to fix a unit)

The category names are truncated. See /u/plg_simon_mensah/p/plot-0004.

Agreed.

Makes sense.

Yep.

That holds up.

What happened to MEA around 2022?

Good catch.

Makes sense, thanks.

raised this on the other chart too, same a source revision.

Source?

The bars need a bit more padding.

The trend looks right.

Title says Q3 but the data runs longer.

the legend starts at zero for one series and not the other.

This reflects the current state.

Sorry, lost me. What is the baseline?

This is much clearer, thanks.

The 2019 tick is inverted.

Which vintage?

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.

Can we label the tick labels? Hard to read otherwise.

Yep.

Is this index-linked, or raw?

Wait, which of these is Nigeria?

Can we see MEA on the same scale?

That holds up.

Units?

Exactly.

That's last week's snapshot.

Do we have this in the Q1 pack yet?

the secondary axis seems clipped.

Should we use a darker background? See /u/plg_simon_mensah/p/plot-0004.

Agreed.

I've been concerned about this for a while — your approach of separating the analysis by cohort maturity sidesteps the aggregation bias that was confounding the earlier dashboard.

Absolutely.

conversion here is stale; it should be in absolute terms.

+1, and APAC looks the same way.

The series names could be shorter.

That's fair. See /u/plg_simon_mensah/p/plot-0004.

Definitely.

Confused by the 2008 value — is that a definition change?

Can we see LatAm on the same scale?

Thanks , that resolves it.

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.

Wait, which of these is the UK?

Fine by me.

Which source is value coming from?

+1, and MEA looks the same way.

Not quite — the 1998 figure is an outlier.

Right — energy use was the part I missed.

Good catch.

Will pick this up after the refresh.

Are duplicates flagged?

Units?

Excellent presentation.

Thanks , that resolves it.

Which source is adjusted coming from?

Is this in absolute terms, or raw?

Why does South Africa dip in 2021?

Sorry, lost me. What is the baseline?

Minor: energy use is wrong in the tooltip.

Same as before.

the y axis starts at zero for one series and not the other. (edited to fix a unit)

The annotation font is hard to read.

Is there a lag between when an event happens and when it's reflected in the data — could the timing shift be a data pipeline issue rather than a user behavior change? See /u/plg_simon_mensah/p/plot-0004. (edited to fix a unit)

Is this the same chart as the other chart? It seems different.

See the other chart.

Colour order does not match the legend order.

+1, and Poland looks the same way.

Agreed.

This is much clearer, thanks.

Do we have this in the Q1 pack yet?

Does this include MEA after 2007?

The tooltip is cut off. See /u/plg_simon_mensah/p/plot-0004.

Is this the same chart as the other chart? It comes across as different.

The data's from March, not April.

Not sure what observed is measuring.

Good catch.

What is the refresh cadence on this?

Can you split Q4 out?

Are duplicates flagged?

Can we compare to last year? (edited to fix a unit)

Is the 1999 gap a reporting lag, or is the series genuinely missing?

I'm on board.

Is the 1999 gap a coverage gap, or is the series genuinely missing?

See the other chart.

Scale is off. (edited to fix a unit)

Nice — the H2 view helps.

That's last week's snapshot.

This is mislabelled — Chile was restated in 2003.

Très bien — this level of detail on the assumptions is exactly what we need when presenting to stakeholders who might challenge the methodology downstream.

Correcting myself: throughput is seasonally adjusted, so the comparison holds.

What happened to APAC around 2024?

The axis truncation hides the context of where these numbers sit in the range of possible values — what if we extended it to show the historical min/max band behind the current data?

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.

Agreed. (edited to fix a unit)

Nice — the H2 view helps.

Fine by me.

Right — retention was the part I missed.

the secondary axis reads clipped.

Makes sense, thanks.

Good work.

That is a currency effect, not a real move.

What is the refresh cadence on this?

The pattern makes sense.

Same as before.

Same conclusion here.

Túpac Marchetti That's last week's snapshot.

Context for anyone new: conversion is only comparable in constant terms.

Thanks , that resolves it.

Can we break this down by tier?

Consistent with the published SLA.

Should we use a stacked view? See /u/plg_simon_mensah/p/plot-0004.

Túpac Marchetti Checks out. (edited to fix a unit)

Checks out. (edited to fix a unit)

Wait, which of these is Kenya? (edited to fix a unit)

Let's sync on this tomorrow.

Good catch.

Should the margin be as a share of total for this comparison to mean anything?

I had this wrong earlier. DACH is fine; it was value that was stale.

Same conclusion here.

Fair enough.

Confused by the 2007 value — is that a source revision?

Is the margin in absolute terms here?

The metodología we're using here — is it documented somewhere? I'd like to understand the précis of how nulls are handled in the aggregation layer before we finalize this.

Works for me.

Makes sense.

How is Year derived?

Fully aligned. (edited to fix a unit)

Vera Mayene This is accurate. See /u/plg_simon_mensah/p/plot-0004. (edited to fix a unit)

Roger that.

Sorry, lost me. What is the baseline?

Exactly.

Looks good.

Scale is off.

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

This would be clearer as a table.

The 2007 break is a definition change — it shows up in every series from that source.

For background: China changed reporting in 2001, which is why Q1 comes across as odd.

Does this include the Nordics after 2014?

Can someone share the methodology doc?

Nice — the Q1 view helps.

Decimals are inconsistent between the US and Indonesia.

What's the methodology for "active"? See /u/plg_simon_mensah/p/plot-0004.

Minor: retention is wrong in the tooltip.

do you know whether 1999 was restated?

What is the refresh cadence on this?

Is the other chart built from the same extract?

Need to align on the handoff point.

Same conclusion here.

That matches what I had.

Right — urban share was the part I missed.

Nice — the Q3 view helps.

I agree.

Is the other chart built from the same extract?

Is the other chart built from the same extract?

The filter applied twice by accident.

Simon Mensah FYI forwarding to the team.

the secondary axis comes across as wrong.

Part of the monthly rebalancing.

Is this as a share of total, or raw?

This is exactly the kind of methodological rigor we've been lacking — breaking it down by user acquisition source and controlling for platform differences is the right call.

I've been concerned about this for a while — your approach of separating the analysis by cohort maturity sidesteps the aggregation bias that was confounding the earlier dashboard.

Scale is off.

+1, and the US looks the same way.

Túpac Marchetti Is this raw or adjusted?

That matches what I had.

That matches what I had.

How is value derived? (edited to fix a unit)

Not quite — the 1996 figure is a coverage gap.

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

Definitely. See /u/plg_simon_mensah/p/plot-0004.

Same conclusion here.

Can we see the numerator separately?

Too many decimals on the labels.

For background: Mexico changed reporting in 2023, which is why Q3 looks odd.

We hit the same thing last Q3. The fix was to hold value constant.

Fair enough.

Is this seasonality or a structural change?

How is series_id derived?

Good catch.

That matches what I had.

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.

Yep.

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

Context for anyone new: the anomaly is only comparable in absolute terms.

Context for anyone new: retention is only comparable as a share of total. (edited to fix a unit)

I agree.

Yep.

Works for me.

Can you split H1 out?

That matches what I had.

Yes, exactly this.

The zero line should stand out more.

Works for me.

Should we validate this externally?

Checks out.

+1, and Indonesia looks the same way.

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.

Simon Mensah The pattern echoes what we observed during the platform migration window last year — there was an initial drop, then a gradual recovery as users got accustomed to the new interface and workflows.

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.

Is this index-linked, or raw?

Same conclusion here.

Am I reading the tick labels wrong? Vietnam seems off to me.

Help me understand the methodology here. See /u/plg_simon_mensah/p/plot-0004.

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

These aren't deduplicated.

The "Null" category is distracting.

Am I reading the gridlines wrong? Japan reads clipped to me.

Simon Mensah Looks good.

Which source is Q3 coming from?

Should we version control this?

Not quite — the 1998 figure is a coverage gap.

Très bien — this level of detail on the assumptions is exactly what we need when presenting to stakeholders who might challenge the methodology downstream.

When we segment by région, does the trend hold for all five regions equally, or is the spike concentrated in one geographic area?

Works for me.

Can we see the numerator separately? See /u/plg_simon_mensah/p/plot-0004. (edited to fix a unit)

Wait, which of these is MEA?

Worth noting Chile and Brazil are not measured the same way over the 2010s.

Looks good.

Is the 2009 gap a currency effect, or is the series genuinely missing?

Works for me.

The 1995 break is a methodology change — it shows up in every series from that source.

How is series_id derived?

The color convention isn't obvious.

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.

Duplicated.

For background: Japan changed reporting in 2013, which is why Q3 looks odd.

Simon Mensah The análisis looks solid.

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.

+1, and Japan looks the same way.

raised this on the other chart too, same a definition change.

That's a solid approach.

Vera Mayene I'm fully on board with the decision to use the external reference as the ground truth — it gives us an objective baseline when there's ambiguity in our internal methods.

this contradicts the other chart — one of the two is stale.

When was the data last refreshed?

Careful, Q3 changed definition in 2022.

Yep.

Noted.

I've been concerned about this for a while — your approach of separating the analysis by cohort maturity sidesteps the aggregation bias that was confounding the earlier dashboard.

Is the 2017 gap a rounding artifact, or is the series genuinely missing?

No.

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.

Does this include Iberia after 2004?

Absolutely.

Colour order does not match the legend order.

I do not follow — is the margin going up or down here?

The 2023 break is a currency effect — it shows up in every series from that source.

Fair enough.

Yes.

Fair enough.

Vera Mayene The timing matches the network upgrade.

Can you split Q4 out?

Confused by the 2023 value — is that a rounding artifact?

Agreed.

The series names could be shorter.

We hit the same thing last H2. The fix was to hold Q3 constant.

Which source is raw_metric coming from? (edited to fix a unit)

Is /u/plg_simon_mensah/p/plot-0004 built from the same extract?

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. See /u/plg_simon_mensah/p/plot-0004.

Could the variance in Q2 be explained by a shift in the user composition rather than a genuine change in underlying behavior — are we segmenting by cohort maturity or acquisition channel?

Yep.

That's right.

Is the 2013 gap a definition change, or is the series genuinely missing?

Duplicated.

Missing the month-end adjustment.

Should we version control this?

What happened to APAC around 2001?

The legend is ordered by total magnitude rather than by the natural categorical hierarchy — reorganizing it to match the domain structure would make the chart more scannable and reduce cognitive load.

Which source is adjusted coming from?

Do we have this in the H2 pack yet?

Have we accounted for the holiday effect?

Confused by the 2002 value — is that an outlier?

That's the way to do it.

Worth scheduling a knowledge transfer.

Careful, Q3 changed definition in 1998.