EdgeKit AI PaywallComing soon
Meter the curious. Sell to the ready.
Your paywall rules read your own scores and RFV beside the meter count today. AI Paywall will add propensity to subscribe, churn risk and price sensitivity, learned from your own readers.
First match wins
- 01Likely to subscribe is HighSkip the meter · Annual offerMatched
- 02Paid · Churn risk is HighWin-back offer
- 03OtherwiseMeter · 5 free reads / 30 days
Chosen: annual offer, meter skipped
Today and next
Three ways to run your paywall
- Today
Rules defined by you and your AI assistant
Your commercial team writes paywall rules in the rule builder, or has an AI assistant connected to EdgeKit draft them. A published change goes live in 60 seconds.
- Today
Send us your scores and use them in the paywall
Send propensity or churn scores from your own model, and use RFV — recency, frequency and volume of reading — from your live event stream. A rule reads both beside the meter count.
- Coming soon
Fully automatic
The model runs the rules and the A/B tests on them, and optimises for yield.
The problem
A meter counts reads. It cannot tell who is about to pay.
Two readers can sit on the same meter count and be nothing alike. One found you through a search link this morning. The other reads you most days and came in from your newsletter. A rule built on reads alone treats them the same, so you either hold back the offer from the reader who would take it, or push it at the one who was only passing through.
Reader A
- Visits this month
- 1
- Sections read
- 1
- Arrived from
- Search
- Rule on reads alone
- 1 free read left
- Rule with a score
- Likely to subscribe · LowKeep metering · 1 free read left
Reader B
- Visits this month
- 12
- Sections read
- 6
- Arrived from
- Newsletter
- Rule on reads alone
- 1 free read left
- Rule with a score
- Likely to subscribe · HighAnnual offer, now
What it will do
Two ways to use it. You pick.
Keep writing the rules and give them a score to read. Or hand the rules over and let the model run them.
Scored rules
A score is one more fact your rule reads
Today a paywall rule reads the meter, the reader’s membership, where they signed up and whether the device is new, its RFV and any score you send. AI Paywall will add three learned scores to that list. Your commercial team will write the condition in the same rule builder, and a published change will go live in 60 seconds, as it does today.
- Propensity to subscribe: stop metering the reader who is ready, and show the offer.
- Churn risk: reach a drifting subscriber with a win-back before they cancel.
- Price sensitivity: offer the trial to the reader who needs it, and the full price to the reader who does not.
- Test it first: run a scored rule against your current rule as an A/B test before it takes all your traffic.
- Metered reads, 30 days
- 4
- Membership
- Registered
- Signed up from
- Newsletter
- New device
- No
- RFV
- 5 · 4 · 3
- Your score
- 0.82
AI Paywall
- Likely to subscribe
- High
- Coming soon
- Churn risk
- —
- Coming soon
- Price sensitivity
- Low
- Coming soon
Fully automatic
Or hand it the rules and the tests
If you would rather not write rules at all, the model will run them, and the A/B tests on them, and optimise for yield. You choose which mode runs.
- Rules you writeEvery decision is a rule your team wrote.Live
- Scored rulesYour rules, reading your scores and RFV.Live
- Fully automaticThe model runs the rules and the A/B tests, for yield.Coming soon
Your data
Learned from your readers, and nobody else’s
One stream per organization, never pooled
The model will learn from the first-party stream EdgeKit Analytics already collects on your own domain: what each reader read, how often, how recently and where they came from. Every organization’s stream is collected and stored separately, so a score will never be learned from another media group’s readers.
- Collected first-party, on your own domain.
- Kept per organization. Never pooled, never resold.
- A copy of every valid event loads on a schedule into Snowflake, BigQuery or MotherDuck in your cloud account, or into a Snowflake database of your own when EdgeKit hosts you.
- Reads, last 30 days
- 23
- Days active
- 12
- Sections
- Politics · Business · Culture
- Arrived from
- Newsletter
- Device
- Mobile · iOS
Available today
Use your own scores today
If your data team already scores readers, send the score and your paywall rules read it today, beside RFV.
Bring your own score
Your warehouse score, on the reader record
You name the field. EdgeKit’s team declares it on the reader record and sets up a secured endpoint for it, and a job on your side or ours sends each reader’s score from your warehouse. No reader can set their own score, and a late or replayed row never overwrites a fresher one. On the record, a paywall rule reads it like the meter, and your profile rules can turn the number into a band.
- Set only through the endpoint EdgeKit sets up, never by the reader.
- Newest score wins. Stale rows are counted and refused.
- Erased with the reader when they ask to be forgotten.
- Your titlesReading on your domain
- EdgeKit AnalyticsFirst-party collection
- Your warehouseA copy of every event, on a schedule
- Your modelYour data science team · sends each score
- Reader recordpropensity_band: High
- Paywall ruleReads it
Today, and with AI Paywall
- Rules your team writes, or your AI assistant drafts, live in 60 seconds
- A/B tests between rules
- Your own scores, sent from your warehouse and read by paywall rules
- RFV: recency, frequency and volume of reading, read by paywall rules
- Scores EdgeKit learns from your stream: propensity, churn risk, price sensitivity
- Rules and tests run by the model, for yield
Adoption
Start before it ships
Nothing here is a big-bang deploy. Each step works on its own.
White-glove onboarding: we set it up with you, start to finish.
- 01
Put the paywall on rules
EdgeKit Paywall: meters, offers and tests your commercial team writes and publishes itself.
- 02
Collect first-party
EdgeKit Analytics on your own domain: the stream AI Paywall will learn from.
- 03
Send the score you have
If your data team already scores readers, EdgeKit sets up an endpoint and the score lands on the reader record, where paywall rules read it.
- 04
Switch on AI Paywall
When it ships, learned scores land on the same record, for the same rules.
Questions
Before you book the demo
Is AI Paywall available today?
In part. Paywall rules read your own scores and RFV today. Learned scores and fully automatic are coming soon, and we do not publish a ship date: book a demo and we will tell you where they are.
We already have a propensity model. Can we use it?
Yes, today. EdgeKit sets up an endpoint, each reader’s score lands on their record, where no reader can change it, and your paywall rules read it there.
Do we hand the paywall to a model?
Only if you choose to. With scored rules, every decision will still be a rule your team wrote, with the score as one more condition. Fully automatic will hand the rules and the A/B tests to the model, which optimises for yield.
What will the model learn from?
Your own readers: the first-party stream EdgeKit Analytics collects on your domain. Every organization’s data is collected and stored separately and never pooled, so a score will never be learned from another media group’s readers.
What else do we need to adopt?
AI Paywall will work through EdgeKit Paywall’s rules and learn from EdgeKit Analytics, so you adopt it alongside those two. Your CMS, email platform and payment processor stay where they are: the CMS keeps rendering and the processor keeps billing.
How fast does a rule change go live?
A published rule change goes live in 60 seconds today. Scored rules publish the same way.