Carepatron growth

Referral program performance

PostHog production data, 3 July 2025 to 6 July 2026. Published 2 July 2026, updated 6 July 2026. All amounts in US dollars.

Referred customers are our highest quality signups, but the channel is small. Referred signups convert to paid at 26%, about 10 times the 2.6% trial baseline. Most stay: 83 to 92% of referred converts still pay today. The channel adds about 4 paying customers a month, around 1.4% of new paid customers.

We raised the US reward from $300 to $1,000 on 1 July 2026. Early signs are positive: invites doubled in the first week, and half the senders are first time referrers. Referred signups have not moved yet, and 6 days is too little to call it.

26%

of referred signups convert to paid

53 of 203 in the past year

83 to 92%

of referred converts still pay

44 to 49 of 53, method dependent

1 in 5

paid workspaces sent an invite

777 in the past year

4 a month

referred customers become paying

about 1.4% of new paid customers

Invites doubled in the first week of the $1,000 reward

Workspaces sent 79 invites in the first 6 days of July, against 51 in the same June window. Daily invites are running at about twice the June rate. The reward changed on 1 July, and that day saw 27 invites, the most of any day in the past month.

The lift is in sending, not visits. Opens are flat, so people who open the referral page now send at about twice the June rate. New referrers are arriving too: 29 of the 56 workspaces that sent in July (52%) had not sent an invite in the previous year.

June, $300 reward July, $1,000 reward

Invites sent each day, 8 June to 6 July 2026. 6 July is a part day.

First 6 days of each month compared

Measure	1 to 6 June	1 to 6 July	Change
Referral page opens	84	79	-6%
Invites sent	51	79	+55%
Workspaces sending	37	56	+51%
Referred signups	4	3	flat

Treat this as an early signal, not a result:

it covers 6 days and 56 workspaces
referred signups have not moved, with 3 so far in July. Signups lag invites, and paid conversion lags signup by a median of 67 days
the drip email carrying the new amount was itself updated on 1 July, so fresh email content may explain part of the lift
we could not confirm whether the announcement emails to about 5,900 US admins and staff have gone out. They were drafts on 1 July

The instrumentation gap still stands. Referral events do not record country, so comparing the US ($1,000) with other regions ($300) needs a join to signup location data. We suggest a full checkpoint at the end of July on 3 measures: workspaces sending invites, invites per referrer, and referred signups.

The funnel over the past 90 days

Counts are events, not unique people, except workspaces sending. The joined and activated steps only exist since 12 and 25 May 2026, so they undercount the 90 day window.

Invite volume barely moved in the past year

Workspaces sent 136 to 263 invites a month all year. Opens spiked to 1,099 in October 2025, but signups only reached 42 that month. Invites rose to 245 in May 2026 around the relaunch, then fell back to 194 in June. Referred signups did not follow: 24 in May, 13 in June.

Referral page opens Invites sent

Referred signups by month

Monthly detail

Month	Opens	Invites sent	Referred signups

The charts show full months only. July appears in the table and in the daily chart above.

Referrals are an owner behavior

Admins account for 95% of paid workspaces that referred. Only 35 staff members sent an invite all year. Staff are the largest untapped audience. Whether staff can even open the referrals page is an open question with engineering.

Referrer role	Paid workspaces referring	Share	Invites sent
Admin or owner	753	95%	1,800
Staff	35	4%	101
Contact	1	0%	1

A further 110 non-paying workspaces sent 226 invites. The adoption denominator is approximate: 777 referring paid workspaces against a paid base of 3,600 to 4,800, so 16 to 21%.

Half of referrers send exactly one invite

The median referrer sends one invite. The mean is 2.5. A small tail of 26 workspaces sent 10 or more. Getting one-time referrers to send a second invite is a bigger pool than finding new referrers.

857 workspaces sent at least one invite in the past year, including non-paying workspaces.

Referred signups convert at 10 times the baseline

203 people signed up through a referral link in the past year. 53 of them started paying, all after their referral signup. That is 26%, against a 2.6% trial to paid baseline. The median time from signup to first subscription is 67 days. 2 of the 53 converted in the first days of July.

Retention is strong. 9 of the 53 have a cancellation event, which puts retention at 83%. The current subscription flag says 49 of 53 still pay, which puts it at 92%. The truth sits between the two, because some cancellations were trial cancellations or were followed by a resubscribe.

One caveat: this is correlation, not proof the reward causes the quality. Referred signups choose us because someone they trust recommended us. Proving the reward adds customers we would not otherwise get needs a holdout test.

What PostHog tracks

Five events cover the referral funnel. The two server side events are new, so conversion telemetry only starts in May 2026.

Event	Count, past year	Tracked since	Useful properties
Referral opened	4,878	Before July 2025	role, providerId
Referral sent	2,162	Before July 2025	role, providerId, isSubscribed, code
Referral signup	204	Before July 2025	code, providerId, isReferred
Referral joined	35	12 May 2026	referrerProviderId, referredProviderId, referredEmail
Referral activated	10	25 May 2026	referrerProviderId, referredProviderId, activatedAt

Notes for anyone querying this data

use providerId to identify workspaces. It is never null. personId is null on 43% of sent events
the isSubscribed flag is reliable on Referral sent only. The person level flag is polluted and returns 95,590
Referral activated means the referee became active in the product, at a median of 6 days. It is not the 3 month payout gate
PostHog has no reward or payout event. Reward records live in the Customer.io Referral Rewards custom object
referral events do not record country, so US and rest of world need a join to signup location

How we verified these numbers

temporal ordering: every one of the 53 converters subscribed after their referral signup. None had a prior subscription, so person merging did not inflate the rate
test accounts: zero test users on all 5 referral events
client against server: 32 client side signups against 33 server side joins from 12 May to 2 July 2026, near perfect agreement
retention measured 2 ways: cancellation events give 83% and the current subscription flag gives 92%
the adoption rate is a range because there is no clean count of paying workspaces in PostHog. We bounded it with 12 month and 18 month subscription activity and the May 2026 segmentation of about 4,150

Prepared by growth. Source: PostHog production, events from 3 July 2025 to 6 July 2026. Reward records: Customer.io workspace 199150.