3 drafts ready. Nothing sent.
Shopify email · approval first
The email tool that gets paid to send less.
Joon connects to Shopify, learns your brand and sends fewer, better emails. You approve every campaign, and random controls help prove what changed.
Request an invite connect Shopify in one click
Joon charges only 5% of revenue attributed to its emails. If they drive no revenue, you pay nothing. Sending and the email creator are included.
Run your numbers- example
- customers
- 4,820
- campaign candidates
- 187
- left alone today
- 612
Illustrative product walkthrough—not merchant results.
joon ›Priya bought a linen tunic in March
✓joon noticed: linen, not wool
✓won't pitch her wool in October
✓kept on the list · left alone
- 05:00, Win-back: Sent for approval. 3 drafts ready. Nothing sent.
- 09:04, Priya: Silent. Bought linen yesterday. Left alone.
- 16:30, Reema: Sent. Reorder nudge timed to her cycle.
- 19:00, 28 of 187: Silent. Held back to measure the campaign.
- 23:58, Karan: Sent. Delivered when he usually reads.
- 00:12, Maya: Silent. Already at the fatigue limit.
Decisions illustrative
The loop gets sharper without getting louder.
Joon turns store history into customer state, customer state into fewer decisions, and every outcome into better evidence for the next one.
- 01Connect
Orders, customers, products and consent
→ - 02Learn
Buying rhythm, state and product relationships
→ - 03Decide
Who needs contact and who should be left alone
→ - 04Send
Approved campaigns and journeys at a useful time
→ - 05Measure
Attributed orders and pooled control evidence
a changing customer, a changing decision
Left alone today. Ready for a nudge tomorrow.
Joon does not put people in a permanent bucket. It checks their buying rhythm, recent behaviour and the campaign in front of it, then explains when the decision changes.
She normally buys at full price
→Day 45 · MayaWatchNo order yet
Her usual buying window has just opened
→Day 60 · MayaCandidate15 days later than usual
A reminder may now be useful
→Today · MayaSendViewed her repeat product
Reorder email, no discount
Every quiet decision has a reasonPriya · bought 2h ago left alone · recently purchasedDev · 3 emails this week left alone · fatigue capNisha · 22:40 local deferred · sends 07:00 her time
Illustrative state progression. Joon re-evaluates when orders, browsing or time change the customer’s state.
from joon · overnightMon, 23 Jun · 5:54
Quiet night. I scanned all 4,820 of your customers. Here’s what I lined up. Nothing’s sent until you say so.
187 lapsed buyers drafted for win-back · expected recovery ₹1.2L.Last spring’s cohort · an estimate until the control proves it.
Reema’s reorder nudge, timed to her cycle, not a Tuesday blast.Her Triphala runs low this week. Queued for your sign-off.
The 9am blast, left alone. I held back 28 as a control, so the lift is proven, not claimed.Holdouts are one-way. You can’t measure this after the fact.
control · held
That’s the whole morning. Approve the lot, or tap into any line.· joon
drafts before sunrise · approvals over coffee
the morning brief
Drafts before sunrise. Approvals over coffee.
Ask for a campaign or activate a journey and joon prepares the audience, message and holdout. Read the reasoning, approve the exact version, finish your coffee.
// nothing ships over your head · every line waits for your nod
every customer, a different email decision
4,820 customers. Not one undifferentiated blast.
A human marketer can only truly know a handful of customers, so everyone else gets the same blast. joon knows each one, and decides whether an email is useful from the evidence the store already has:
- who they are
- what they want
- when they want to hear from you
- which brand-approved message variant they received
// lapsed · loyal · at risk · new · every one a person, not a row
4,820the same people, sorted into who they actually are
- Lapsed187
- Loyal1,240
- At risk612
- New904
Attention means knowing when to stay quiet, too.
Blast your whole list and your best customers learn to mute, archive and unsubscribe, until the audience you paid to build quietly stops opening. joon sends email when it can help, and holds back when it doesn’t, so everyone feels looked after and no one feels spammed.
More revenue, from fewer, better messages.
cohort · vana-spring-lapsed
Sent · 159 customers
₹0/ customer
Control · 28 customers
₹0/ customer
Illustrative pooled result
₹1,15,593
The sent group spent ₹865 per customer and the control group spent ₹138. Across 159 sent customers, that is ₹1,15,593 above the control baseline. This evidence teaches Joon which campaigns to repeat; it does not calculate the invoice.
Illustrative example. Live results appear only after enough control evidence accumulates.
more from less
More revenue from fewer, better messages. And you can prove it.
Reaching the right customer at the right moment grows revenue without adding sends. On measurable campaigns Joon first decides who the message may help, then holds a small random sample of those candidates back. Results pool over time, so proof is reported only when the evidence supports it.
Holdouts are not a setting. Holdouts are one-way: you can’t run a control on history. Every campaign that runs without one loses that proof forever.
Left alone is a decision. A regular full-price buyer may not need this discount. If her state changes, she becomes a candidate again.
A control is random. A small, rotating sample of campaign candidates misses one email so Joon can learn what the campaign changed. Controls are proof, never the billing base.
a day, on the clock
A day of decisions. Each one at its hour.
One customer at a time, each decision posting at the real hour it happens. Scroll the day forward.
- 05:00pre-dawn
Drafts written while you sleep.
A merchant-requested campaign is segmented and written in your voice, queued, not sent.
- 08:00morning
Approvals over coffee.
One page of what’s queued. Approve, hold, or edit. Nothing ships over your head.
- 09:00restraint
Priya: linen, not wool. Left alone.
She bought a linen tunic in March. joon won’t pitch her October’s woollens, kept on the list, the 9am blast left alone.
- 16:30afternoon
Reema: Triphala running low.
A gentle reorder nudge, timed to her cycle, not a Tuesday blast. Queued for your sign-off.
- 19:00dusk
The Diwali win-back goes out.
187 lapsed campaign candidates, with 28 held back as a control, so results can be pooled into honest evidence.
- 00:00midnight
Karan reads at midnight.
Not at 9am. joon left the blast alone and sends the emailwhen he’s actually reading.
more campaigns, sharper decisions
A small control stays random. Deliberate restraint grows.
Joon can start with order history: who buys at full price, their usual reorder rhythm and what they buy. As fresh campaigns add evidence, more customers can be left alone for offers they do not need—and brought back the moment their state changes.
Suppression changes with the customer. The control remains a small random sample of current campaign candidates.
General rule
1
Rohanrepeat customer · high value
A recent buyer is unlikely to need a discount.
Your store's evidence
2
Rohanrepeat customer · high value
Rohan returns without being asked, but uses a code when offered one.
Measured against control
3
Rohanrepeat customer · high value
Across comparable campaigns, the random control shows whether the nudge helped.
Illustrative decision progression. Estimates remain labelled until enough control observations exist.
restraint at the front door
Grow the list. Don’t pay twice for the same customer.
Popups, flyouts, bars and hosted pages, with consent wording set per market. If someone is already on your list or bought last week, they do not get another discount code.
email and phone consent stay separate
10% off your first order
Join for considered edits and occasional offers.
□ Yes, email me offers and updates.Already subscribed or bought recently. No new discount issued.
Illustrative form and incentive decision.
what you’d pay
List-based tools charge whether emails work. Joon charges only when its emails drive revenue.
After early access, Joon charges only 5% of non-cancelled order revenue attributed to an email it actually sent within seven days. If that revenue is ₹0, the fee is ₹0. Sending and the email creator are included.
free during early access · email only · merchant approval required
State-based decisions: regular full-price buyers, recent purchasers, fatigue and poor offer fit.
A rotating sample of campaign candidates, used for proof—not billing.
Shown for context. Click-through below is measured from delivered emails.
2,00,000 blast-tool deliveries→40,000 deliberately left alone→24,000 random control→1,36,000 sent by Joon (68%)
every month, based on active profiles—whether the emails work or not
on ₹4,68,000 attributed revenue
₹37,800 less in this modelCampaigns: 1,36,000 delivered × 4% clicked × 2% ordered = 109 orders = ₹2,18,000
Abandoned cart: 50,000 sessions × 5% carts × 5% recovered = 125 orders = ₹2,50,000
Joon charges only 5% of non-cancelled order revenue attributed to an email it actually sent within seven days. Deliberately-left-alone and control customers are never billed. Sending and the email creator are included.
Illustrative model, not a forecast. Opens are diagnostic; click-through is calculated from delivered emails. The journey example includes abandoned-cart recovery only. Random controls are pooled over time for proof and learning, never used to calculate the invoice. The list-based figure is an illustrative comparison, not a quote.
Email today.
Campaigns, journeys, customer state and approval—one channel, made accountable.
connect · learn · attend
It does the job, not “a tool to do the job.”
You type a goal in plain language. joon works the way a careful marketer would, out loud, before a single message sends.
$ joon connect
It reads your store.
One click into Shopify and your email sender. In this illustrative walkthrough, joon scans 4,820 customers and finds the 187 lapsed buyers worth reaching, about ₹4.2L in past revenue.
$ joon learn
It holds out a control first.
From the 187 candidates, joon randomly assigns 28 to control, balanced on past spend. They miss this campaign; the rest are assigned treatment.
$ joon attend
It writes, then waits for your nod.
joon drafts 3 win-back variants in your voice, explains the evidence before you approve, and keeps 28 customers silent as a control.
What joon shows before you approveEstimate · not yet measured
Expected upside₹1.2L recovereddirectional estimate vs 28 control customers
Named downside~0.6% unsubscribethe risk joon will not hide from you
ConfidenceModeratefirms up as control data lands
joon ❯ deploy on my store
Try joon with your store.
Connect Shopify in one click. joon holds out a control when the campaign is big enough to measure and writes your first campaign before you finish your coffee.