The overall average hides 4 different numbers

plan splits saas-subscriptions into 4 groups. Here is what mrr_usd looks like inside each.

• enterprise — mean 4,282, median 4,361 (150 rows)
• business — mean 693, median 655 (442 rows)
• team — mean 166, median 160 (800 rows)
• starter — mean 31, median 27 (1,108 rows)

Top to bottom that is 4,282 against 31, a spread of 13,612.2%. The pooled average is 447.

select plan,
  count(*)                                             as rows,
  round(avg(mrr_usd)::numeric, 2)                      as mean,
  percentile_cont(0.5) within group (order by mrr_usd) as median
from saas_subscriptions
group by 1
order by mean desc;

A spread that wide means the pooled number is not a summary, it is an artefact of the mix. Change the proportion of enterprise rows and the overall average moves without any individual group changing at all — which is how a metric goes up while every segment goes down.

Notice the mean and median columns disagree only slightly here. Always compute both in the group-by. The comparison between them per segment is free and tells you whether you are looking at a level difference or a tail difference.

This is the setup for Simpson's paradox — the case where every segment moves one way and the total moves the other.