The welcome series that stopped annoying people
Rebuilt a five-email onboarding flow around one honest test per send. Fewer messages, better timing, no dark patterns.
+38%second donations
Jacob · Data science for non-profits
You are datapoint #13.
From here on, this page shows you what it knows about your visit. The panel in the corner keeps a running total.
Section A · The analyst
Data for organizations that run on members, not ad budgets.
I'm Jacob. I build the numbers side of campaigns, running donation experiments, training churn models, and keeping clean the pipelines that hold a few million supporter records. Nine years in, my conclusion is simple. The organizations people trust with their money should be the best in the world at handling their data.
A/B testing
donor prediction
SQL and Python
dashboards people actually open
Placeholder bio. A fuller version is coming soon.
Section B · Evidence
Knowing who is about to leave
A gradient-boosted churn model that flags at-risk members a month before they lapse, in time for a human to reach out.
1,900members kept per year
Forty spreadsheets, one source of truth
Moved reporting from CSVs in inboxes to a tested warehouse. Same questions, one answer, checked nightly.
40 to 1reporting sources
Sample entries for now. Real case studies are coming soon.
Section C · Where you fit
12 people were counted before you.
Here are the past visitors, by time spent on this page. The red bar is you, and it moves while you read.
12 past visitors
you, live
Visitor reading times load here.
Right now you've read longer than 0% of them.
Counts began at launch and update as people visit.
Section D · The rules
What I collect. What I never will.
Collected, with consent
Scroll depth, right now—
Reading time—
Clicks—
Viewport width—
The clock pauses when you switch tabs. I only want real reading time.
Never collected
Name or email
IP address
Location
Device fingerprint
Cookies
If I wouldn't say it to your face, I don't collect it behind your back.
Section E · Your receipt
The whole dataset, drawn.
The line below is your scroll position over time, drawn right here in your browser from every sample this page took. When you leave, one anonymous summary (your time on page, deepest scroll, and screen width bucket) gets added to the running count. The per-second detail stays in this tab.
0:00
time on page
0%
deepest scroll
0
clicks
0
samples
See the raw data
Download saves this session as a JSON file on your device. Delete wipes this local view and stops the summary from being sent.
Consent-first data still wins campaigns. Ask the people counted before you.
Placeholder contact links. Real ones are coming soon.
No analytics on this page besides the ones you can watch being made. The only thing that leaves this tab is one anonymous count, with no cookies and no way to identify you.