Upsell Recommender
Explained upsells when customers truly need more.
Upsell Recommender
Customer Support & Success
- RuleDone:
Each night, or after a Stripe or Shopify event, a customer is checked.
- RuleDone:
It skips customers who are past due, in dispute, complaining, at high churn risk or just bought.
- RuleDone:
It finds compatible products and scores them by usage signals.
- AIDone:
It checks the live price; AI writes the reason and an optional message.
- PersonWaiting for approval:Waiting for you
The CSM gets a HubSpot task; store messages wait for the daily batch approval.
- RuleNot started:
After 90 days it compares revenue with the holdout group.
For example: A project-tool SaaS on Stripe has a customer using 47 of 50 seats for three weeks.
Waiting for you. Nothing has changed yet.
Review for important actions. Customer-facing and record-changing actions are reviewed first.
What is Upsell Recommender?
In shortUpsell Recommender is an AI automation for Customer Support & Success teams that works with Gmail, HubSpot, Outlook, Shopify, Stripe and WhatsApp. Spots real upgrade needs from usage, purchases and renewals, gives account owners an explained suggestion, and proves extra revenue against a holdout group. It is built for head of Customer Success, account managers and CSMs and E-commerce owner or marketer (approves message batches).
Last reviewed
Key facts
- Department
- Customer Support & Success
- Starts when
- Every night at 02:30, and within minutes of a Stripe subscription/invoice event or a new Shopify order
- Works with
- Gmail, HubSpot, Outlook, Shopify, Stripe, WhatsApp
- Run modes offered
- Shadow, Draft, Assisted and Auto
- Human control
- Review for important actions
- Status
- Available now
The problem today
Upsell is ad hoc. CSMs notice "they are at 95% of seats" only when the customer complains. Stores send the same "you may also like" to everyone, even people who just bought it, and nobody measures whether upsells really add revenue.
What it does
Every night it checks each customer's plan, usage, purchases and renewal date. Rules skip customers who should not be pitched, pick only compatible products from your catalog and score them with clear signal points. The account owner gets a HubSpot task with the reason and a talk track; for a store, consented customers can get an email after the marketing owner approves the day's batch. A holdout group is kept back, so the real extra revenue can be measured.
How does Upsell Recommender work?
How it works- Rule you set
- AI step
- A person decides
- Step 1Rule
Each night, or after a Stripe or Shopify event, a customer is checked.
- Step 2Rule
It skips customers who are past due, in dispute, complaining, at high churn risk or just bought.
- Step 3Rule
It finds compatible products and scores them by usage signals.
- Step 4AI
It checks the live price; AI writes the reason and an optional message.
- Step 5Person
The CSM gets a HubSpot task; store messages wait for the daily batch approval.
- Step 6Rule
After 90 days it compares revenue with the holdout group.
What you get
Key features- Clear scores, for example: seats ≥ 90% for 14 days +40, limit hit 3+ times in 30 days +35, complement +25, renewal within 60 days +10, usage decline −30; minimum 50, top 2 per customer.
- Compatibility rules: requires, excludes, region, minimum plan, one tier up only, and in stock.
- Holdout group of 10% by default (0–30%), stable per customer.
- Weekly lift report with a confidence interval, per product and overall.
- Cooldowns: 30 days after a purchase, 45 days between upsell messages.
- The live price must match the catalog exactly, or the suggestion is dropped and you are told.
How much does it do on its own?
Run modesPick a mode to see what Upsell Recommender would do at each step.
Every automation starts in Shadow. Move it up when you are ready, and back down at any time. How run modes work
Upsell Recommender in Shadow
Runs on real data and shows what it would have done. No app is changed and no message is sent.
- Rule
Each night, or after a Stripe or Shopify event, a customer is checked.
Watches for it
- Rule
It skips customers who are past due, in dispute, complaining, at high churn risk or just bought.
Runs on real data, read only
- Rule
It finds compatible products and scores them by usage signals.
Runs on real data, read only
- AI
It checks the live price; AI writes the reason and an optional message.
Runs on real data, read only
- Person
The CSM gets a HubSpot task; store messages wait for the daily batch approval.
Shown to you as a suggestion
- Rule
After 90 days it compares revenue with the holdout group.
Shown to you as a suggestion
Shadow vs Auto for Upsell Recommender
Auto is the most it can do on its own. Here is each step side by side.
| Step | Shadow | Auto |
|---|---|---|
| In short | Runs on real data and shows what it would have done. No app is changed and no message is sent. | Runs without approval, with limits, alerts and a kill switch. Unlocks after a trial period in Assisted. |
| Step 1Each night, or after a Stripe or Shopify event, a customer is checked. | Watches for it | Runs on its own within your limits |
| Step 2It skips customers who are past due, in dispute, complaining, at high churn risk or just bought. | Runs on real data, read only | Runs on its own within your limits |
| Step 3It finds compatible products and scores them by usage signals. | Runs on real data, read only | Runs on its own within your limits |
| Step 4It checks the live price; AI writes the reason and an optional message. | Runs on real data, read only | Runs on its own within your limits |
| Step 5The CSM gets a HubSpot task; store messages wait for the daily batch approval. | Shown to you as a suggestion | Within your limits; anything outside them goes to a person |
| Step 6After 90 days it compares revenue with the holdout group. | Shown to you as a suggestion | Runs on its own within your limits |
| True in every mode | Never changes prices, applies coupons, charges cards or changes subscriptions; Stripe and Shopify are read only. | |
What it will never do on its own
Safety & human control- Never changes prices, applies coupons, charges cards or changes subscriptions; Stripe and Shopify are read only.
- No fake urgency, guarantees, discounts, competitor or other-customer mentions; every number in the text is checked.
- Customer messages need marketing consent and a batch approval, with a footer (postal address + one-click unsubscribe).
- Enterprise accounts (MRR above $2,000 by default), price jumps over 50% and regulated products go to the CSM only.
- It re-checks at send time: a customer who already bought or is no longer eligible gets nothing.
- You approve before it acts
- Your data stays yours
- It knows its limits
- Stop anytime
What will you see on your dashboard?
Results you can trackTargets are quality goals the automation is built to reach. Estimates are calculated from your own volumes.
- Target
Incremental revenue vs holdout (target: ≥ +5% over 90 days, p < 0.1)
A goal it is built to reach
- Tracked
Revenue influenced (shown separately)
Measured on your own data
- Tracked
Acceptance rate by product
Measured on your own data
- Tracked
CSM "not relevant" rate
Measured on your own data
- Target
Complaint / unsubscribe rate (target: ≤ baseline + 0.2 points)
A goal it is built to reach
- Tracked
Customers evaluated, blocked by reason, messages approved and sent
Measured on your own data
- Target
Unsuitable recommendations (target: 0)
A goal it is built to reach
- Target
Number fidelity (target: 100%)
A goal it is built to reach
- Target
Recommendation accuracy (target: ≥ 95%)
A goal it is built to reach
- Target
Explanation quality (target: ≥ 4.0 / 5)
A goal it is built to reach
A project-tool SaaS on Stripe has a customer using 47 of 50 seats for three weeks. The account owner gets a HubSpot task for the seat add-on with the seat numbers and a short talk track. A similar customer in the holdout group gets nothing, so the lift report can show the real difference later.
Who is Upsell Recommender for?
- Head of Customer Success
- Account managers and CSMs
- E-commerce owner or marketer (approves message batches)
Industries: SaaS & tech, E-commerce & retail, Professional services
Also useful for: Sales & CRM, Marketing & Growth
Connects to the apps you already use
Works withApp names are trademarks of their owners. Charviam connects to them; it is not a partner or reseller.
- Gmail
- HubSpot
- Outlook
- Shopify
- Stripe
- WhatsApp(optional)
Better together
Automations that hand work to Upsell Recommender, or pick up where it stops.
What do you need to set it up?
Setup- Choose SaaS (Stripe + HubSpot) or store (Shopify, HubSpot optional); give a Stripe restricted read-only key.
- Import the catalog and review each product's compatibility (launch is blocked until done).
- Set signal points, minimum score, cooldowns, holdout %, enterprise MRR and blocked deal stages.
- Connect the store mailbox and set the email footer (postal address + unsubscribe).
- SaaS: make sure usage signals reach the metric layer.
Typical setup
- 01Connect your apps
- 02Configure the rules
- 03Run a watch-only trial
- 04Go live
Not included
- Price changes, discounts, coupons, subscription changes or charging cards.
- Churn scoring (that is blueprint 26).
- Bulk marketing to contacts without consent.
Questions about Upsell Recommender
FAQSee Upsell Recommender on your own data
WalkthroughIt starts in Shadow, where it only watches and suggests, so nothing changes in your apps until you say so.