Upload your customer data. See who will churnin 30 seconds.
Drop in a CSV or Excel file, point out two columns — customer ID and the churn label — and we do the rest. Your uploaded file is destroyed right after processing.
Original destroyed after processingResult CSV auto-deleted within 30 minNever used for trainingNever shared across accounts
●Destroyed after processingWe process in-memory and never write your upload to disk. The original is destroyed right after diagnosis.
●Not used for trainingYour data never trains any model and is never shared across accounts.
●PII auto-removedEmails, phones, names are auto-detected and excluded; IDs that look personal are pseudonymized.
●Honest metricsOnly held-out (out-of-fold) performance is shown — no inflated training scores.
How it works
From CSV to a ranked churn list
No setup, no connectors, no account. Four steps, about 30 seconds.
1
Upload
Drag & drop a CSV or Excel file. Processed in memory, then destroyed.
2
Pick two columns
Just the customer ID and the churn label. Every other column becomes a feature — you don't need to know what they mean.
AUC, high-risk count and top factors on screen — free. Per-customer probability CSV by email.
1
Upload
2
Columns
3
Diagnose
4
Result
Step 1 · Upload
Drop your file here
Your file is processed in memory only, is never stored, and is destroyed after processing. Uploading is taken as consent to the privacy & auto-destroy policy.
You don't need to include personal-data columns (recommended not to). Auto PII detection is not guaranteed to catch everything, so please remove email / phone / name columns before uploading. (If included, detected ones are auto-excluded.)
↑
Drag & drop a CSV/XLSX here, or click
Up to 50 MB · CSV / XLSX
Download a sample CSV — an example of the expected shape that you can also upload as-is to try a real diagnosis (a few thousand rows). The sample includes one email column to demonstrate auto-exclusion; you don't need personal-data columns in your own upload.
Step 2 · Map columns
Pick just two columns
Every other column is used automatically as a feature. You don't need to know what they mean.
The label must be churn status after the cutoff (0/1 or churned/retained). Features must be from before the cutoff. Columns created after churn (cancel date, refund flag…) are auto-excluded as leakage.
0other column(s) will be used automatically as features
Preview (top 5 rows)
← Scroll sideways to see more columns
Step 3 · Diagnosing
Validating your data
Usually done within 30 seconds (up to a minute for large files).
1PII screening
2Label-leakage audit
3Model training (gradient boosting)
4Held-out evaluation + calibration
This is taking a little longer for a larger file…
Diagnosis · Free summary
This is sample data.Upload your real CSV to get the churn list for your customers.
Predictive power
0.00
AUC
Baseline — 0.5 random · 0.7+ usable in practice
High-risk customers
0customers
of0 · base churn rate0%
(the segment clearly above the base rate)
Headline · Top 3 factors pushing churn
% is relative importance among the three factors, not "explains X% of churn." Column names are shown exactly as in your uploaded data.
This is correlation, not causation.These are columns that moved together with churn — not what to change to actually reduce it. Causes and the actions to take are covered in the free 30-minute diagnosis.
Get the high-risk list
Get every customer's ID · churn probability · risk tier · top factor as a CSV.
Sample results aren't downloadable. Upload your file to get the full per-customer probability CSV.
Optional
The result CSV excludes columns detected as personal data; personal-looking IDs are stored as pseudonyms (hashes) only. Collected: email. Purpose: result delivery (required), marketing (opt-in). Retention: 12 months. Uploaded data is not stored. Deletion requests: pbh95@nlift.co. Security note
Result CSV ready — if the download didn't start automatically, use the button/link below.
Your free 30-min diagnosis request is in.We'll reach out to your email within one business day.
If the download didn't start, use the button below. You can re-download in this session for up to 30 minutes, after which it is auto-destroyed on the server.
This free screen is the summary. The managed service goes further.
Nlift is a managed data service — we own the result and the actions, not a dashboard you run alone. What you just saw is one slice (churn correlation). With your team's real data we go deeper, tailored to your business.
Customer segmentation
Group customers by behavior and value, so retention and offers are targeted — not one-size-fits-all.
Ongoing churn prediction
A refreshed, calibrated risk list on a cadence — not a one-off snapshot.
Next best action (NBA)
For each high-risk customer, a suggested action to run — matched to your levers and channels.
Customer 360
One unified view per customer across your sources, so decisions use the whole picture.
Underperformer diagnosis
Why a store, branch, or cohort is lagging — and which lever moves it — for multi-location or multi-segment businesses.
Built around your goal
We start from your north-star metric and design the model and the operating loop around it.
The on-screen summary is free. The detailed list, the per-customer CSV, and everything above are delivered as a managed service — start with a free 30-minute diagnosis.
Just like this free tool never kept your original file, the managed engagement needs no raw-data export — we work inside your environment.
Auto-excluded columns
Ready to turn this into causes, actions, and a running loop?
Book the free 30-minute diagnosis. A person walks you through what the numbers mean and what to do next — no data export, no commitment.