An AI copilot lifted average order value by 24%
A chain of service centers. We gave reps an AI copilot that suggests upsells live in the conversation — and the average order value climbed.
Where we started
Reps across the service center chain took orders and closed deals, but rarely offered add-ons or related services. Customers left with what they came for, even when something else fit their order.
Top performers kept upsell scripts in their heads and pitched the right thing at the right time. Others forgot, or didn't know what to offer with a given order. The gap between reps was visible across the chain.
The chain had tried to fix this with training for years, but training never scaled to the whole network.
An upsell only works during the conversation: once the customer has paid, it's too late. So the task was to suggest the best next move to the rep right in the moment of the customer conversation.
The pain points
- Upsells were almost never offered
- Scripts lived only in top reps' heads
- Uneven rep performance across the chain
- Training that wouldn't scale
- Average order value left on the table
What we built
Collected the best upsell scripts from the chain's top performers
Trained the copilot on deal history and product pairings
The copilot suggests the add-on and the pitch right inside the deal card
The rep decides what to offer — the human stays in control
Analytics show which suggestions actually move the needle
How the copilot works in a conversation
The copilot is built into the Kommo deal card the rep already works in.
The rep talks to the customer and works in the deal card, where the order is visible.
The copilot checks the order against the chain's deal history and product pairings and shows in the card which add-on to offer and with what argument.
The rep decides whether to offer it. The AI never writes or speaks to the customer on its own.
Deals in Kommo show which suggestions lead to an add-on purchase.
Analytics show which suggestions actually work, and that tells us which to keep and which to rewrite.
The result in money
The copilot pulled weaker reps up to the level of the best ones — no lengthy training required.
A similar situation? We review your funnel in an hour and tell you what from this case you can repeat.
I want the sameTimeline and how we work
First results in this project came after 3 weeks. This is how we work with every client:
Free. We look at your CRM and deals and show where money is lost.
What we build, what data we need, which metrics go into the contract.
We test the solution on your data before the main contract.
We move the team to the new way of working and train reps and their manager.
Included in the project. If we miss the metrics in the contract, we keep working for free or refund your money.
"We spent years trying to teach the whole chain to upsell. The copilot did it in a month — it nudges exactly when it matters."
What you can repeat in your company
A similar result is realistic where an upsell depends on whether the rep remembers to offer it. We don't promise numbers in advance: they go into the contract after the audit.
If your first line also needs relief, look at AI agents in messengers and CRM: they answer customers 24/7 and hand the deal to a rep.
When it works
- You have repeat sales and clear pairings: a service or product often needs an add-on.
- Deals live in a CRM with history — that's what the copilot learns from.
- Your team has strong sellers whose moves can be turned into suggestions.
- A manager is ready to look at the analytics and drop suggestions that don't work.
- Not sure whether reps offer add-ons today? Start with AI call quality control: it checks every call against your checklist.
Questions about this case
Tell us where leads get lost — Viktor runs the review personally
Describe your task in a couple of sentences. During business hours we reply within 20 minutes.
Book a call