ConvoHut
Ecommerce use case

Recommend the next thing at the moment it makes sense.

Every store knows which products go together. Almost none of them act on it at the right moment — the recommendation arrives at checkout, when the customer has already decided, or in a batch email, when it has nothing to do with what they bought.

Free PDF. No email required.

At a glance

TriggerPurchase of a specific product or category
GoalIncrease customer value
Primary KPICross-sell revenue
Messages1–2
Estimated setup timeUnder 25 minutes

What is automated cross-sell on WhatsApp?

Automated cross-sell waits a defined period after a purchase, checks what the customer already owns, and recommends a genuinely complementary product in a conversation they can reply to. Timing is drawn from the product's usage cycle rather than a fixed marketing calendar.

Key takeaways

  • Recommend after the product has been used, not on the day it arrives.
  • Suppress anything the customer already owns — the fastest way to look automated.
  • One recommendation converts better than a list of five.
  • A reply asking 'will this fit mine' is the conversion, not an objection.

Why this leaks revenue today

01

The recommendation arrives too early

An accessory pitched the day the parcel lands reads as a sales push. The customer has not used the product yet.

02

It recommends things they already own

Nothing signals 'automated and careless' faster than being sold the thing sitting in your last order.

03

There is no way to ask a question

Most cross-sell hesitation is compatibility — will it fit, will it work with mine. A one-way email cannot answer that.

The signal ConvoHut acts on

The journey only starts when all of these are true. Getting the entry conditions right matters more than the message.

Purchase of a specific productProduct used for 14+ daysComplementary item not ownedPositive satisfaction signal

What has to be connected

Order historyEverything the customer already owns, so you never re-recommend it
Product relationshipsWhich items genuinely complement which — curated, not just co-purchased
Usage cycleHow long before the complementary product becomes relevant
Compatibility rulesWhich variants, sizes and models actually work together

The journey

Built once, then it runs unattended. Every branch has a defined outcome, and the journey stops the moment the goal is met.

Trigger

Order placed

Delay

Wait 14 days

Condition

Owns product X, not Y

Message

Recommend one item

Then it branches on what the customer does

Asks about compatibility

AI answers from product data

Adds to order

Prefilled checkout link

Not interested

Suppress this recommendation for 90 days

Goal: Second order placed — journey stops

A journey without a stop condition is a journey that gets your number blocked.

The messages, and why each one exists

Starting points, not scripts. You rewrite them in your own voice and approve every message before it goes live.

Day 14

Waits until the product is in use, and frames the recommendation as what other owners do rather than what you want to sell.

Now you've had the {product} a couple of weeks — most people pair it with {complement}. Want the details?

On question

Answers the compatibility question directly using the customer's own order data. That is the conversion.

Yes, it fits the {model} you have. Here it is, and it ships with the same delivery.

What it looks like to the customer

This is the part a broadcast tool cannot do. The customer replies, the question gets answered from real data, and the journey continues — in one thread.

If the AI reaches the edge of what it has been approved to answer, it escalates to your team with the full history attached. The customer sees one continuous conversation.

Now you've had the grinder a couple of weeks — most people pair it with a tamper and mat set. Want the details?
Will it fit the 58mm basket?
Yes — the set is 58.5mm, sized for exactly that basket. It's the one we ship with your machine model.
Add it
Done. Here's a checkout link with it added, same delivery address.

Illustrative conversation. Not a customer transcript.

What to measure

Pick one primary metric before launch and agree how it is calculated. Everything else is secondary.

Cross-sell revenue

Revenue from second orders attributed to the journey.

Attachment rate

Customers who add the recommended item.

Average order value change

Whether customer value actually moved, not just whether orders happened.

Opt-out rate

The guardrail. Cross-sell is where over-messaging shows up first.

How we keep this honest. The attribution window is agreed before launch. Direct and influenced conversions are reported separately. Where volume allows we hold out a control group. We do not claim every order that happens after a message.

Set it up

  1. 1

    1. Map genuine product relationships

    Curate the pairs. Co-purchase data alone will recommend nonsense.

  2. 2

    2. Set the delay from usage, not sales

    Fourteen days is a starting point. Match it to when the complement becomes relevant.

  3. 3

    3. Build the ownership suppression

    Never recommend something in the customer's order history.

  4. 4

    4. Write one recommendation

    One item, one reason, one question. Lists convert worse.

  5. 5

    5. Cap the frequency

    One cross-sell message per customer per quarter is a sensible ceiling.

Then tune it

Move the delay before you change the product

Underperforming cross-sell is usually mistimed rather than mismatched.

Test one item against three

One recommendation with a reason typically beats a carousel of options.

Watch opt-outs, not just revenue

A cross-sell journey that lifts revenue while burning your contactable list is losing money.

Questions this raises

When is the right moment to cross-sell?

After the customer has used the product long enough to understand it. For most categories that is one to three weeks, not the day of delivery.

How is this different from a recommendation email?

The customer can ask 'will this fit mine' and get an accurate answer from their own order data, in the same thread, immediately. That question is where most cross-sell dies.

Won't this annoy customers?

It will if it is frequent, generic or early. One well-timed, well-matched recommendation per quarter is a very different experience from a weekly product email.

How do I stop recommending things they already own?

Ownership suppression against order history is set up before launch. It is the single most important rule in this journey.

Does upsell work for one-off purchases?

Less well. This journey earns its place in catalogues with genuine accessories, consumables or complementary ranges.

Should I discount the cross-sell?

Usually not. If the recommendation is right and the timing is right, a discount adds cost without adding conversion.

Related use cases

Free download

The Upsell Playbook, as a PDF.

Everything on this page in a document you can send to your team or your client: entry conditions, journey logic, the messages with the reasoning behind each one, the metrics and the setup steps.

No email required. Take it and build the journey somewhere else if you want to — the logic is the useful part.

Upsell Playbook

PDF · 4 pages · Version 1.0

Download the playbook See all resources

Your customers are already showing intent.

Turn those moments into automated conversations that move customers toward conversion. We work with a small group of early access customers and set the first journey up with you.

Join the waitlist