Right — throughput was the part I missed.

The 1998 tick is wrong.

That's right.

The cutoff date is one day off.

Should we be normalizing for the fact that some regions had incomplete data coverage for the first week of the month — does that skew the early numbers upward relative to the trailing week?

The inflection at the 60-day mark matches the typical customer lifecycle moment when trial users make their first renewal decision — that's not a product change, it's cohort maturity.

baseline?

Source?

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

What is the refresh cadence on this?

Very good.

The units are missing from the left axis.

Fair point.

Tadeusz Rasmussen Does this account for time zone differences?

Stale.

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

Minor: headcount is clipped in the tooltip.

This is much clearer, thanks.

This lines up with the other chart if you put headcount index-linked.

How is Year derived?

I see it.

Should unemployment be per capita for this comparison to mean anything?

Nice — the Q2 view helps.

Can you share the extract?

Sorry, lost me. What is the baseline?

The 2005 break is a reporting lag — it shows up in every series from that source.

Noor Papadakis Nicely done.

Fair enough.

Agreed.

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 corresponds to the audit period.

Sorry, lost me. What is the baseline?

Can we see Kenya on the same scale?

Fully aligned.

Same conclusion here.

This needs more context before publishing. (edited to fix a unit)

Duplicated. (edited to fix a unit)

Nice — the Q4 view helps.

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

Is energy use per capita here?

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

Works for me.

Can we use a clearer date format?

Thorin Brandybuck Why does the forecast diverge so much in Q4?

Makes sense, thanks.

Moving this to the Q2 agenda.

Is inflation in absolute terms here?

This lines up with /u/plg_thorin_brandybuck/p/plot-0015 if you put conversion in constant terms.

See the other chart.

Scale is off.

What is the refresh cadence on this?

Units?

Can we use consistent units throughout?

Tadeusz Rasmussen Works for me.

Confirmed on my side too.

Context for anyone new: internet penetration is only comparable per capita.

This is much clearer, thanks.

Nice — the Q2 view helps.

Clear.

Yep. (edited to fix a unit)

This is much clearer, thanks.

adjusted should probably be as a share of total.

Source?

Worth noting China and LatAm are not measured the same way over the last decade.

What's the denominator here?

Should internet penetration be seasonally adjusted for this comparison to mean anything?

That matches what I had.

Good catch.

The tooltip is cut off.

Not comparable.

What's the impact of the change in data collection methodology on March 15th? Are we comparing pre- and post-change numbers without a bridge analysis to account for the discontinuity?

Fair point.

Good catch. (edited to fix a unit)

Same conclusion here.

Stale.

the tick labels looks inverted.

Correcting myself: life expectancy is in absolute terms, so the comparison holds.

Careful, Year changed definition in 2014.

Duplicated.

+1, and Norway looks the same way.

What's the population we're measuring?

The 2016 break is a currency effect — it shows up in every series from that source.

Makes sense.

Is the other chart built from the same extract?

Help me understand the methodology here.

Should CO2 per capita be in constant terms for this comparison to mean anything?

Not comparable.

Thanks for pulling this together.

Good point there.

The timing matches the network upgrade.

Should we be looking at moving average instead?

Right — conversion was the part I missed.

Can we batch these updates?

We expected this after the pricing change.

Looks good. See /u/plg_thorin_brandybuck/p/plot-0015.

Works for me.

Units?

Are these cumulative or rolling?

Didn't carry forward the prior period's rounding. See /u/plg_thorin_brandybuck/p/plot-0015.

Not quite — the 1998 figure is a rounding artifact.

Is the baseline pre-promotion or post?

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.

Very good.

Need feedback from the broader group. See /u/plg_thorin_brandybuck/p/plot-0015.

Well done.

Do we have this in the Q3 pack yet?

Yep.

Is this index-linked, or raw? (edited to fix a unit)

How is raw_metric derived?

Does this include Kenya after 2001?

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. See /u/plg_thorin_brandybuck/p/plot-0015. (edited to fix a unit)

Not quite — the 2013 figure is an outlier.

Fair enough.

Fair enough.

energy use here is mislabelled; it should be seasonally adjusted.

+1, and DACH looks the same way.

We expected this after the pricing change.

This is accurate.

Same conclusion here.

What's driving the spike in March?

Tadeusz Rasmussen Are we double-counting within the funnel where users interact with both paths, or does the attribution model already handle that deduplication correctly? See /u/plg_thorin_brandybuck/p/plot-0015.

Should be $2.1M not $2.3M. See /u/plg_thorin_brandybuck/p/plot-0015.

This is much clearer, thanks.

Clear.

Agreed.

This is much clearer, thanks.

Good point there.

The data's from March, not April.

Thorin Brandybuck The timing of the drop perfectly matches the end of the fiscal year incentive program, which suggests the spike was partially driven by pull-forward demand rather than genuine organic growth.

Is the 2018 gap a methodology change, or is the series genuinely missing?

Thorin Brandybuck The interaction between the two factors — call them X and Y — might be the real story here. Have we looked at the cross-tabulation or run an interaction test?

Right — the margin was the part I missed.

Fair point.

Makes sense.

Makes sense, thanks.

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

I had this wrong earlier. the UK is fine; it was series_id that was stale.

Thanks , that resolves it.

Are we filtering bots?

Should we be looking at moving average instead?

What happened to Mexico around 2016?

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

Can you split Q3 out?

Makes sense, thanks.

That matches what I had.

Am I reading the left axis wrong? MEA comes across as inverted to me.

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. See /u/plg_thorin_brandybuck/p/plot-0015.

Fair point.

How is value derived?

That matches what I had.

See the other chart.

I had this wrong earlier. the US is fine; it was series_id that was mislabelled.

The 2024 tick is stale.

Is this seasonally adjusted, or raw?

Right — retention was the part I missed.

Before we publish this to the exec dashboard, we should validate it against the external data provider's numbers — we've had discrepancies with them in the past that turned out to be subtle definition mismatches.

Tadeusz Rasmussen What's the methodology for "active"?

Works for me.

Not sure what Q3 is measuring.

I agree.

Yes, exactly this.

Fair enough.

Can you split Q3 out?

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

Which source is Year coming from?

Can we see Norway on the same scale?

What is the refresh cadence on this?

Yes.

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.

Makes sense, thanks. (edited to fix a unit)

Adding this to the review list.

What happened to Vietnam around 2024?

Context for anyone new: mobile subscriptions is only comparable per capita.

Works for me.

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

Thorin Brandybuck Are duplicates flagged?

Noted.

Nice — the Q3 view helps.

Can you share the extract?

The truncation is too aggressive.

We saw similar behavior in 2023.

Thanks , that resolves it.

Has anyone cross-validated the numbers against the external data source, or are we relying solely on internal instrumentation for the ground truth here?

Nicely done.

Worth pulling into the Q4 review.

Noor Papadakis This scale makes small changes invisible. See /u/plg_thorin_brandybuck/p/plot-0015.

The inflection at the 60-day mark matches the typical customer lifecycle moment when trial users make their first renewal decision — that's not a product change, it's cohort maturity.

This is accurate.

Agreed.

The sum doesn't match the chart.

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.

Year?

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

Which source is series_id coming from?

The outliers — there are quite a few — need attention.

Tadeusz Rasmussen Mixing old and new methodologies.

Are we filtering outliers?

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

Fair enough. (edited to fix a unit)

Noor Papadakis Worth flagging to product.

Moving this to the H2 agenda.

FYI.

Thanks , that resolves it.

I'd agree with this direction.

Reflects the cohort maturity curve.

Noor Papadakis Should we stratify by device type?

Not comparable.

Noor Papadakis Has anyone cross-validated the numbers against the external data source, or are we relying solely on internal instrumentation for the ground truth here?

The resolution is too coarse.

Makes sense, thanks.

Why does Norway dip in 2008?

Can you split H2 out?

The cutoff date is one day off.

Will pick this up after the refresh.

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

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.

What's the denominator here?

That matches what I had.

Includes test records.

Yep. (edited to fix a unit)

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.

The pattern makes sense.

Minor: CO2 per capita is off in the tooltip.

Right — GDP was the part I missed.

How is baseline derived?

The 2015 break is a reporting lag — it shows up in every series from that source.

The 1999 break is a currency effect — it shows up in every series from that source.

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.

That's a solid approach.

Should we normalize by days in month?

Is the 2024 gap a definition change, or is the series genuinely missing?

Fine by me. See /u/plg_thorin_brandybuck/p/plot-0015.

This is much clearer, thanks.

Which vintage?

Missing the cohort-based correction.

What's the population we're measuring?

Does this include Indonesia after 2024? (edited to fix a unit)

Noor Papadakis That's the way to do it.

Worth noting MEA and China are not measured the same way over the back half of the series.

Can we drill down by region?

Yep.

Fine by me.

Which source is Year coming from?

Missing 2021.

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. See /u/plg_thorin_brandybuck/p/plot-0015.

Noor Papadakis Sounds right. (edited to fix a unit)

Confirmed. (edited to fix a unit)

This is much clearer, thanks.

Good point there.

Can we label the colour scale? Hard to read otherwise.

FYI.

Thanks for catching that détail.

Is the 2003 gap a currency effect, or is the series genuinely missing? (edited to fix a unit)