What we noticed
From Նշիր's Mixpanel data, last 28 days. Each insight names the goal step it affects. Last run 19 Nov 2026, 18:52 UTC.
Text written by templates, not Claude · add an Anthropic API key to use Claude
From Նշիր's Mixpanel data, last 28 days. Each insight names the goal step it affects. Last run 19 Nov 2026, 18:52 UTC.
Text written by templates, not Claude · add an Anthropic API key to use Claude
In the last 28 days, 246 people reached "Open a modal". 74 of them went on to "Request an email code", so 172 stopped here.
Why it matters: This is one of the two biggest losses on the way to "More people create an account". Everyone who stops here is lost for the goal.
Possible reasons (not confirmed yet)
Measured numbers
Open a modal → Request an email code: 74 of 246 (30.1%) continue, 69.9% drop
How sure we are. Counted from reliable events over one period.
S1 · funnel_drop · sample 246 · 2026-10-22 to 2026-11-19
-- funnel_timelines.sql
-- params: ["product","2026-10-22T00:00:00.000Z","2026-11-19T00:00:00.000Z",[],"session_started",["session_started","element_clicked","modal_opened","email_code_requested","$mp_submit","account_created"],"2026-11-19T00:00:00.000Z",[]]
-- Per-user timelines for the funnel. Cohort: users whose first entry event falls in [$2, $3).
-- $1 product_id, $2 cohort start, $3 cohort end, $4 excluded distinct_ids,
-- $5 entry event, $6 funnel event names, $7 horizon end (events after it are ignored),
-- $8 property names that step conditions read (only those are returned, to keep rows small)
with cohort as (
select distinct_id,
min(time) as first_time,
(array_agg(props->>'platform' order by time))[1] as platform,
(array_agg(props->>'browser' order by time))[1] as browser
from events
where product_id = $1 and name = $5 and distinct_id <> all($4::text[])
group by distinct_id
having min(time) >= $2::timestamptz and min(time) < $3::timestamptz
)
select c.distinct_id, c.platform, c.browser, e.name, e.time,
case when cardinality($8::text[]) = 0 then null
else (select jsonb_object_agg(k, e.props->k) from unnest($8::text[]) k where e.props ? k) end as props
from cohort c
join events e on e.product_id = $1 and e.distinct_id = c.distinct_id
where e.name = any($6::text[]) and e.time >= c.first_time and e.time < $7::timestamptz
order by c.distinct_id, e.time
Run run_f545f7a97996484b · context v1 · prompt S3/v2 · model offline-template
In the last 28 days, 74 people reached "Request an email code". 27 of them went on to "Submit a form", so 47 stopped here.
Why it matters: This is one of the two biggest losses on the way to "More people create an account". Everyone who stops here is lost for the goal.
Possible reasons (not confirmed yet)
Measured numbers
Request an email code → Submit a form: 27 of 74 (36.5%) continue, 63.5% drop
How sure we are. Counted from reliable events over one period. Below the minimum sample of 100 users, so confidence stays Low.
S2 · funnel_drop · sample 74 · 2026-10-22 to 2026-11-19
-- funnel_timelines.sql
-- params: ["product","2026-10-22T00:00:00.000Z","2026-11-19T00:00:00.000Z",[],"session_started",["session_started","element_clicked","modal_opened","email_code_requested","$mp_submit","account_created"],"2026-11-19T00:00:00.000Z",[]]
-- Per-user timelines for the funnel. Cohort: users whose first entry event falls in [$2, $3).
-- $1 product_id, $2 cohort start, $3 cohort end, $4 excluded distinct_ids,
-- $5 entry event, $6 funnel event names, $7 horizon end (events after it are ignored),
-- $8 property names that step conditions read (only those are returned, to keep rows small)
with cohort as (
select distinct_id,
min(time) as first_time,
(array_agg(props->>'platform' order by time))[1] as platform,
(array_agg(props->>'browser' order by time))[1] as browser
from events
where product_id = $1 and name = $5 and distinct_id <> all($4::text[])
group by distinct_id
having min(time) >= $2::timestamptz and min(time) < $3::timestamptz
)
select c.distinct_id, c.platform, c.browser, e.name, e.time,
case when cardinality($8::text[]) = 0 then null
else (select jsonb_object_agg(k, e.props->k) from unnest($8::text[]) k where e.props ? k) end as props
from cohort c
join events e on e.product_id = $1 and e.distinct_id = c.distinct_id
where e.name = any($6::text[]) and e.time >= c.first_time and e.time < $7::timestamptz
order by c.distinct_id, e.time
Run run_f545f7a97996484b · context v1 · prompt S3/v2 · model offline-template
9% of Chrome users on desktop who reached "Start a session" went on to "Click an element" (5 of 56), compared with 50% of everyone else. This is based on 56 people, fewer than the 100 we need to be sure.
Why it matters: Chrome users on desktop are 10% of people at this step. At everyone else's rate, about 23 more of them would reach "Click an element".
Possible reasons (not confirmed yet)
Measured numbers
Start a session → Click an element: platform=desktop, browser=chrome 8.9% (5 of 56) vs 50.2% for everyone else
How sure we are. Fewer than 100 people in this group. Below the minimum sample of 100 users, so confidence stays Low.
S3 · segment_gap · sample 56 · 2026-10-22 to 2026-11-19
-- segment_step_gap.sql
-- params: ["product","2026-10-22T00:00:00.000Z","2026-11-19T00:00:00.000Z",[],"session_started","element_clicked",0]
-- Conversion from $5 to $6 per platform and browser. Users whose first $5 is in [$2, $3);
-- converted if $6 happens after it, at least $7 calendar days later, before $3.
-- $1 product_id, $4 excluded distinct_ids
with f as (
select distinct_id, min(time) as t,
(array_agg(props->>'platform' order by time))[1] as platform,
(array_agg(props->>'browser' order by time))[1] as browser
from events
where product_id = $1 and name = $5 and distinct_id <> all($4::text[])
group by distinct_id
having min(time) >= $2::timestamptz and min(time) < $3::timestamptz
),
c as (
select f.distinct_id
from f
join events e on e.product_id = $1 and e.distinct_id = f.distinct_id and e.name = $6
and e.time >= f.t and e.time < $3::timestamptz
and date_trunc('day', e.time) >= date_trunc('day', f.t) + make_interval(days => $7::int)
group by f.distinct_id
)
select coalesce(f.platform, 'unknown') as platform, coalesce(f.browser, 'unknown') as browser,
count(*)::int as n, count(c.distinct_id)::int as converted
from f left join c using (distinct_id)
group by 1, 2
order by 1, 2
Run run_f545f7a97996484b · context v1 · prompt S3/v2 · model offline-template
70% of iPhone users on facebook_mobile who reached "Start a session" went on to "Click an element" (38 of 54), compared with 44% of everyone else. This is based on 54 people, fewer than the 100 we need to be sure.
Why it matters: If we learn what helps them, it may work for everyone. Treat this as a lead to check, not a fact.
Possible reasons (not confirmed yet)
Measured numbers
Start a session → Click an element: platform=ios, browser=facebook_mobile 70.4% (38 of 54) vs 43.6% for everyone else
How sure we are. Fewer than 100 people in this group. Below the minimum sample of 100 users, so confidence stays Low.
S4 · segment_gap · sample 54 · 2026-10-22 to 2026-11-19
-- segment_step_gap.sql
-- params: ["product","2026-10-22T00:00:00.000Z","2026-11-19T00:00:00.000Z",[],"session_started","element_clicked",0]
-- Conversion from $5 to $6 per platform and browser. Users whose first $5 is in [$2, $3);
-- converted if $6 happens after it, at least $7 calendar days later, before $3.
-- $1 product_id, $4 excluded distinct_ids
with f as (
select distinct_id, min(time) as t,
(array_agg(props->>'platform' order by time))[1] as platform,
(array_agg(props->>'browser' order by time))[1] as browser
from events
where product_id = $1 and name = $5 and distinct_id <> all($4::text[])
group by distinct_id
having min(time) >= $2::timestamptz and min(time) < $3::timestamptz
),
c as (
select f.distinct_id
from f
join events e on e.product_id = $1 and e.distinct_id = f.distinct_id and e.name = $6
and e.time >= f.t and e.time < $3::timestamptz
and date_trunc('day', e.time) >= date_trunc('day', f.t) + make_interval(days => $7::int)
group by f.distinct_id
)
select coalesce(f.platform, 'unknown') as platform, coalesce(f.browser, 'unknown') as browser,
count(*)::int as n, count(c.distinct_id)::int as converted
from f left join c using (distinct_id)
group by 1, 2
order by 1, 2
Run run_f545f7a97996484b · context v1 · prompt S3/v2 · model offline-template
24% of iPhone users who reached "Submit a form" went on to "Create an account" (11 of 46), compared with 41% of everyone else. This is based on 46 people, fewer than the 100 we need to be sure.
Why it matters: IPhone users are 31% of people at this step. At everyone else's rate, about 8 more of them would reach "Create an account".
Possible reasons (not confirmed yet)
Measured numbers
Submit a form → Create an account: platform=ios 23.9% (11 of 46) vs 40.6% for everyone else
How sure we are. Fewer than 100 people in this group. Below the minimum sample of 100 users, so confidence stays Low.
S5 · segment_gap · sample 46 · 2026-10-22 to 2026-11-19
-- segment_step_gap.sql
-- params: ["product","2026-10-22T00:00:00.000Z","2026-11-19T00:00:00.000Z",[],"$mp_submit","account_created",0]
-- Conversion from $5 to $6 per platform and browser. Users whose first $5 is in [$2, $3);
-- converted if $6 happens after it, at least $7 calendar days later, before $3.
-- $1 product_id, $4 excluded distinct_ids
with f as (
select distinct_id, min(time) as t,
(array_agg(props->>'platform' order by time))[1] as platform,
(array_agg(props->>'browser' order by time))[1] as browser
from events
where product_id = $1 and name = $5 and distinct_id <> all($4::text[])
group by distinct_id
having min(time) >= $2::timestamptz and min(time) < $3::timestamptz
),
c as (
select f.distinct_id
from f
join events e on e.product_id = $1 and e.distinct_id = f.distinct_id and e.name = $6
and e.time >= f.t and e.time < $3::timestamptz
and date_trunc('day', e.time) >= date_trunc('day', f.t) + make_interval(days => $7::int)
group by f.distinct_id
)
select coalesce(f.platform, 'unknown') as platform, coalesce(f.browser, 'unknown') as browser,
count(*)::int as n, count(c.distinct_id)::int as converted
from f left join c using (distinct_id)
group by 1, 2
order by 1, 2
Run run_f545f7a97996484b · context v1 · prompt S3/v2 · model offline-template
40% of Android users on Chrome who reached "Submit a form" went on to "Create an account" (39 of 97), compared with 26% of everyone else. This is based on 97 people, fewer than the 100 we need to be sure.
Why it matters: If we learn what helps them, it may work for everyone. Treat this as a lead to check, not a fact.
Possible reasons (not confirmed yet)
Measured numbers
Submit a form → Create an account: platform=android, browser=chrome 40.2% (39 of 97) vs 26% for everyone else
How sure we are. Fewer than 100 people in this group. Below the minimum sample of 100 users, so confidence stays Low.
S6 · segment_gap · sample 97 · 2026-10-22 to 2026-11-19
-- segment_step_gap.sql
-- params: ["product","2026-10-22T00:00:00.000Z","2026-11-19T00:00:00.000Z",[],"$mp_submit","account_created",0]
-- Conversion from $5 to $6 per platform and browser. Users whose first $5 is in [$2, $3);
-- converted if $6 happens after it, at least $7 calendar days later, before $3.
-- $1 product_id, $4 excluded distinct_ids
with f as (
select distinct_id, min(time) as t,
(array_agg(props->>'platform' order by time))[1] as platform,
(array_agg(props->>'browser' order by time))[1] as browser
from events
where product_id = $1 and name = $5 and distinct_id <> all($4::text[])
group by distinct_id
having min(time) >= $2::timestamptz and min(time) < $3::timestamptz
),
c as (
select f.distinct_id
from f
join events e on e.product_id = $1 and e.distinct_id = f.distinct_id and e.name = $6
and e.time >= f.t and e.time < $3::timestamptz
and date_trunc('day', e.time) >= date_trunc('day', f.t) + make_interval(days => $7::int)
group by f.distinct_id
)
select coalesce(f.platform, 'unknown') as platform, coalesce(f.browser, 'unknown') as browser,
count(*)::int as n, count(c.distinct_id)::int as converted
from f left join c using (distinct_id)
group by 1, 2
order by 1, 2
Run run_f545f7a97996484b · context v1 · prompt S3/v2 · model offline-template