Seen enough to have a conversation?
No sales deck. Just a direct conversation about your problem.
Talk to an engineerRetail & E-commerce · Intelligent Automation
Recovering 34% of abandoned revenue through multi-signal conversion automation
A US-based direct-to-consumer brand generating $40M+ in annual online revenue was recovering less than 6% of abandoned cart value — relying on a single generic email sent 24 hours after abandonment. We built a multi-signal conversion automation system across exit intent, browse abandonment, and cart abandonment that recovered 34% of previously lost revenue within 90 days of go-live, without replacing their existing Klaviyo stack.
The Problem
They had the traffic.
They were losing the revenue at the door.
The brand had built a strong acquisition engine — paid search, social, and influencer channels driving consistent top-of-funnel volume. Conversion rate on first visit sat at 2.8%, which is respectable for their category. The problem was everything that happened after a visitor left without buying. Roughly 78% of sessions ended without a purchase. Of those, the brand was recovering less than 6% through a single Klaviyo cart abandonment email sent the following day. The rest — representing tens of millions in potential annual revenue — was walking out the door permanently.
The root cause was not the email tool — Klaviyo was capable of far more than how it was being used. The problem was the absence of a signal layer. Every abandonment was treated identically: a visitor who spent 40 minutes comparing products and added three items to cart received the same message as someone who bounced after viewing a single product page. There was no behavioural segmentation, no timing logic, no channel coordination between email and SMS, and no mechanism to suppress already-converted customers from receiving recovery messages.
At their traffic volume, the difference between a 6% recovery rate and a 30%+ recovery rate was not a creative problem. It was an engineering and data problem. The sequences needed to be driven by real-time behavioural signals, not a 24-hour batch job.
The scale of the problem
- 78%
- sessions ended without purchase
- <6%
- abandoned revenue recovered
- $14M+
- estimated annual revenue leakage
Industry average: 70–75% for DTC apparel and home goods
One generic email, 24-hour delay, no segmentation
Based on average order value and abandonment volume
Scope of Work
What we were asked to build
Behavioural signal ingestion layer
Real-time event pipeline capturing exit intent, product page dwell time, add-to-cart, checkout initiation, and partial form completion — normalised into a unified session profile per visitor.
Multi-signal segmentation engine
Segmentation logic classifying each abandonment by signal type, intent score, product category, and customer lifetime value tier — determining which sequence, channel, and timing applies to each session.
Klaviyo sequence orchestration
Custom Klaviyo flows built per segment — browse abandonment (3-touch), cart abandonment (4-touch with dynamic product rendering), and checkout abandonment (2-touch with urgency logic). SMS integrated via Klaviyo for high-intent segments.
Suppression and conversion attribution
Real-time purchase event listener suppressing in-flight sequences the moment a conversion occurs. Multi-touch attribution model tracking which signal and which touch in the sequence drove the recovery.
Constraints we worked within
- Klaviyo could not be replaced — all sequence delivery had to run through their existing instance
- No changes to the storefront codebase — signal capture via tag manager only
- SMS required opt-in compliance review before activation — delayed that workstream by 3 weeks
- Attribution model had to reconcile with their existing GA4 and Northbeam reporting
Explicitly not in scope
- Paid retargeting or social ad integration
- On-site personalisation or product recommendation engine
- Loyalty programme or post-purchase retention sequences
- Changes to pricing, promotions, or discount strategy
How We Worked
5 months. 3 phases. One revenue metric that mattered.
Audit & Architecture
Full audit of existing Klaviyo setup, GA4 event taxonomy, and session data. Mapped all abandonment touchpoints. Defined segment logic and sequence architecture. Identified the SMS compliance gap early — scoped around it.
Signal Layer & Segmentation Build
Built the behavioural event pipeline via Google Tag Manager. Developed the intent scoring model using 90 days of historical session data. Klaviyo flows built and QA'd in staging. Suppression logic tested against live order stream.
Controlled Rollout
Launched to 30% of traffic with holdout group for clean measurement. Email sequences live first. SMS activated in week 3 after compliance sign-off. Iterated on send timing based on early open and click data.
Full Rollout & Handoff
Expanded to 100% of traffic. Attribution dashboard handed to their internal marketing team. Documented segment logic and flow architecture for ongoing management. 90-day post-launch monitoring period included.
Working rhythm
- CadenceTwo-week sprints, bi-weekly revenue reviews
- Decision ownerVP of E-commerce on client side
- Primary metricAbandoned revenue recovered (weekly)
- Escalation SLA24 hours with written recommendation
Results
Measured at 90 days post full rollout.
of abandoned cart value recovered within the sequence window
Measured as revenue attributed to recovery sequences divided by total abandoned cart value over the same period. Cart abandonment sequences drove 61% of recovered revenue; checkout abandonment sequences drove 29%; browse abandonment drove the remaining 10%.
increase in revenue per recovery email sent
Behavioural segmentation meant high-intent segments received fewer, better-timed messages. Revenue per send improved because suppression logic eliminated noise — converted customers no longer received recovery emails that diluted the metric.
suppression accuracy — converted customers removed from sequences in real time
Prior to the build, roughly 12% of recovery emails were sent to customers who had already purchased — damaging brand perception and inflating unsubscribe rates. Real-time purchase event suppression eliminated this entirely.
reduction in email list churn over the 90-day measurement period
Fewer irrelevant messages sent to the wrong segments at the wrong time. Behavioural relevance improved engagement scores across the board, which also improved deliverability and inbox placement rates.
Is This Your Situation?
The system we built here is not specific to this brand.
It applies to any e-commerce operation where abandonment recovery is treated as a creative problem rather than an engineering one.
- Your abandonment sequences are batch-triggered, not event-driven — every visitor gets the same message on the same delay
- You have no real-time suppression — converted customers are still receiving recovery emails
- Your recovery rate has plateaued and changing the email copy has not moved it
This engagement was scoped as an additive layer on top of an existing Klaviyo instance — no platform migration, no replatforming, no disruption to live campaigns. Five months from kickoff to full rollout.
Seen enough to have a conversation?
We scope every engagement before we quote. No sales deck. Just a direct conversation about your problem.
More Work
Cutting new CX rep time-to-proficiency from 8 weeks to 3 using AI customer service training
Read case studyAutomating carrier rate selection for 2,200 monthly shipments — cutting cost per booking by 31%
Read case study