5%

Rich Automated Behavioural Intelligence & Decisioning

The Marginal Gains
Problem.
And why only Conviva solves it.

Any mature business picked and ate its low-hanging fruit years ago. The gap between where you are and where you need to be is no longer closed by big decisions — it's closed by thousands of precise ones.

Scroll to explore

The Marginal Gains Problem

A marathon runner with a 3-hour PB doesn't shave an hour off their time with one obvious fix.

Usain Bolt doesn't wake up and realise there was a simple 1 second 100M adjustment sitting there all along.

An F1 team doesn't deploy 1 engineer with a screwdriver to shave 1 second off a lap time.

Elite performance is built on the accumulation of thousands of decisions — each one small, each one precise, each one informed by data most competitors can't even see.

Mature businesses are no different. The single decision that moves profitability by 20% doesn't exist. But a thousand well-informed decisions might compound to 5% — and that's the difference between winning and losing at scale.

The Resolution Problem

Traditional tools lack the precision
to operate at this level.

Funnels record whether an event happened at least once — a tally, not a story. Some tools go further, offering crude journey or path analysis that sequences events between two points you define. That's an improvement. But it still requires you to know what you're looking for before you look, it operates on sampled data at scale, and timing — how long a user spent on each step, and what that duration signals — is an afterthought, not a first-class dimension.

Traditional funnelsRecord whether an event happened at least once. No sequence. No timing. No context. A tally.
Journey toolsSequence-aware, but require you to define start and end points. Sampled at scale. Timing is not a first-class dimension.

The result: vast human teams burning hundreds of thousands of hours trying to stretch low-resolution insights into high-resolution decisions. And failing.

The Funnel's Fatal Simplification

A user can take a deeply non-linear journey — looping, researching, failing, returning — and the funnel reduces it to a clean linear progression, incrementing a tally each time a milestone is passed at least once.

The Real Journey — Campaign to Purchase

Step / 21 steps

Session 1 · Discovery

Arriving with intent

A summer sale campaign brings this customer in. They scroll to 40% of the home page and swipe through 5 products in the Big Sale Furniture carousel — engaged behaviour, not a bounce. Then they search directly. This is a considered shopper building a mental picture before committing to a query.

First Contact · Friction

Found it. Out of stock.

SKU-48821 is the product they want — the search was directed, not exploratory. They select their preferred variant: Grey/L. Out of stock. First friction point. The product decision is already forming; the journey has just hit its first obstacle.

Comparison

30 seconds on the alternative

A new search term, a different product. SKU-52104 holds their attention for 30 seconds — enough to evaluate, not enough to commit. They return to SKU-48821. The original product preference is confirmed.

Re-engagement · Intent Confirmed

Second visit — variant in stock

Returns to SKU-48821. This time Blue/L is available — 2 units. Selects the variant. Adds to cart. Purchase intent is now confirmed. The product decision is made. What happens next is a transaction problem, not a product problem.

Price Shock

The $50 calculation

Enters the delivery address. 7-day delivery — acceptable. But a $50 delivery fee is a significant add-on for a furniture purchase. The mental negotiation begins. They leave without completing. The funnel calls this abandonment. It's actually a pricing decision in progress.

⏸ 4 Hours · No Activity

Still in the funnel

Four hours of silence. Session-based funnels count this as abandonment. Patterns know better: the product decision is made, the delivery cost is the obstacle. This customer is thinking, not gone. The business has a narrow window to intervene.

Session 2 · Return

Searching for a solution

Returns with a specific search: "free delivery". They are not re-evaluating the product — they are trying to solve the delivery cost problem. High-intent, price-sensitive behaviour. This is the Research Oscillator in action.

Deep Evaluation · 9 Minutes

Final validation on SKU-48821

Third visit to the original product. 9 minutes. Opens reviews. Searches "firm" — quality check. Searches "door" — measuring whether the sofa fits through their doorway. A buyer doing due diligence, not a browser. The funnel reads it as failed PDP conversions.

Back in Cart · Conviction

Friction compounds at checkout

Back in cart with conviction. SAVE10 — invalid. SUMMER15 — expired. Each failed promo code increases cognitive friction. The business has 30 minutes of genuine engagement across two sessions — and is moments from losing the sale.

Finance Declined

The tipping point

Attempts financing — declined. Combined with the $50 delivery fee and two failed promo codes, the financial friction is now significant. The gap between purchase intent and transaction completion has never been wider.

⏸ 24 Hours · No Activity

The last chance window

Second abandonment. 24 hours pass. Without intervention, this sale is lost. The customer has high product intent, 9 minutes of review-reading, and a known email address. This is the window for targeted re-engagement. Every hour reduces the probability of return.

Session 3 · Re-engagement

A targeted email closes the gap

A WELCOME20 email re-engages the customer. The code works. The path from re-engagement to purchase is almost frictionless — the product decision was made 2 days ago. The email didn't create intent; it removed the last financial obstacle.

Purchase Complete

27 interactions. 7 checkmarks.

The funnel recorded: Home → Search → Product Page → Add to Cart → Cart → Checkout → Purchase. Patterns recorded everything else: the out-of-stock friction, the $50 shock, the 4-hour deliberation, the 9-minute review, the failed finance, the 24-hour gap, the email re-engagement.

The Persistent Few

This customer was the exception.

Most customers sharing this behavioural pattern — same entry point, same product, same price sensitivity — abandon before reaching checkout.

Conversion Rate

Behavioural Segment 3% Site-wide 10%

27 interactions across 3 sessions over 2 days. The funnel recorded seven checkmarks.

What the funnel records

Precision decisions enabled via high-resolution pattern-based insights

📦
InventoryStock alert for Grey/L — the first abandonment was a variant availability problem, not a conversion problem.
📐
Product ClarityThe customer searched "door" at 9 minutes deep on SKU-48821 — measuring whether it fits through their doorframe. A single PDP callout confirming the sofa ships in separable modules removes a silent barrier that conventional analytics is completely blind to.
🚚
DeliveryWaive the $50 fee for basket values over $500 — the product costs $899, making the fee proportionally punishing. Also consider free first-order delivery: a one-time acquisition cost that removes the barrier at the moment of highest intent.
TimingTrigger cart recovery at 4 hours, not 24 — this customer was still in the decision window at hour 4.
💳
PaymentIntroduce BNPL at checkout — a finance decline shouldn't terminate a high-intent session.
🏷️
DiscountsSurface valid discount codes proactively — three consecutive failures compounded abandonment risk.
📧
Real-time ContextThe WELCOME20 email was already automated — but it was sending blind. Patterns reveal that it wasn't the discount alone that closed the sale; it was price sensitivity compounded by a declined finance application, in the absence of any alternative. The email landed at the right moment by coincidence. Connected to session context, it triggers by design.
↓   Scroll to reveal the journey
Continue ↓

The Informed Alternative

The same customer. The same product intent. With the six pattern-based decisions applied, the journey resolves in 10 interactions across 2 sessions — in under 4 hours. The funnel still just records the same six checkmarks.

The Optimised Journey — Decisions Applied

27
Steps
3%
Conversion
Without insights
10
Steps
12%
Conversion
With insights

Decisions That Changed the Journey

📦
Inventory
Stock alert on Grey/L means the customer finds their preferred variant available on return — no comparison loop needed.
Removes 4 wasted interactions
📐
Product Clarity
The pattern shows the customer searched "door" while reviewing SKU-48821. A PDP callout confirming the sofa ships in separable modules removes a silent barrier before it causes abandonment.
Eliminates a silent drop-off reason at deep evaluation
🚚
Delivery
Free delivery on first orders and all orders over $500 — no $50 shock on cart entry, no "free delivery" search session, no 4-hour deliberation.
Eliminates Session 2 research loop
Timing
Cart recovery triggered at 4 hours — not 24. The customer receives the re-engagement email while still in the decision window.
24-hour gap → 4-hour gap
🏷️
Discounts
SAVE10 is surfaced proactively on that PDP — no failed codes, no compounding friction, no cognitive load.
3 failed attempts → 1 successful apply
💳
Payment
BNPL is offered at checkout — the finance application step is replaced by a seamless Pay Later option.
Finance decline → accepted in one step

The funnel still only records

Why Conviva's Patterns Are Different

Most analytics tick off major milestones. Funnels tally them. Journey tools tell you in what order they occur, for gates you define. Conviva tells you precisely what, in what order, for how long, with what specific context — and which combinations of those factors actually drive conversion, abandonment, or churn, without you having to ask.

Traditional analytics(Roughly) WHAT happened + HOW MANY TIMES
Journey tools(Roughly) WHAT happened + IN WHAT ORDER
ConvivaPRECISELY WHATORDERTIMINGCONTEXT→ OUTCOMES — surfaced automatically

Automatic discovery

No human labelling. No predefined hypotheses. No knowing what to look for first.

Timing as signal

Not just the sequence of steps — the duration of each, and what that duration reveals about intent.

Dimensional depth

Not "viewed a product page" — but which SKU, from which campaign, with which search term, with what in the cart.

Census scale

No sampling. Every interaction, every user, every session. The complete picture.

One Number. Millions of Realities.

Every metric on your dashboard is an average — a single number representing millions of distinct behavioural realities collapsed into one. High-resolution decision-making requires seeing inside the number, not just reporting it.

Low Resolution
68%
Cart Abandonment Rate

One number. Millions of distinct behavioural segments inside it. Conviva sees every one.

The Agentic Reality

Decision-making has always been a function of insight resolution. The more precisely you can see individual behaviour, the more precisely you can act on it. That resolution is accelerating — and the humans making decisions can't keep pace.

The Golden Age
One voice. Big decisions.
The low-hanging fruit was still there. A single insight — TV reach, price cut, new territory — could move the needle by 20%. One smart decision made the difference.
Today
100% Human. High-resolution. Micro-decisions.
The fruit is gone. Winning requires thousands of precise, well-informed decisions — each one small, each one demanding high-resolution insight. Humans are at their limit. Most competitors are already failing here.
The Future
Agent-driven. Nano-decisions. Real time.
AI agents making hyper-personalised decisions for every individual user, at every moment, at census scale. Not possible without ultra-high resolution behavioural data — the kind only Conviva provides.

Only Conviva solves for this.

The precision. The scale.
The intelligence. In real time.

Conviva captures the exact sequence and timing of every micro-interaction within real-world user journeys — every search, click, view, and transaction — enriched with the dimensional context that makes each one meaningful [search term, SKU, filter applied, promo code entered, campaign source, what was already in the cart etc.]; automatically surfacing the patterns that matter without human labelling or predefined queries; resolving millions of distinct behavioural segments at census scale, in real time, and giving both human decision makers and AI agents the precision to make the right intervention for every user at the right moment.

Conviva conviva.com