R / 003Field note / Enterprise experimentation

Tangentix Research

A purchase path is not a funnel

How cross-channel path analysis turns an averaged conversion funnel into a ranked, testable growth agenda.

Published: July 5, 2026
Reading time: 4 minutes
Author: Tangentix Research

Executive decision brief

The conclusion, its use, and its boundary.

Paper: R / 003
Evidence status: Sanitized reconstruction

Working conclusion

A repeated journey constraint is a testable hypothesis, not a result. Prioritize the experiment that can resolve the most consequential uncertainty before changing the roadmap.

Decision to settle: Which journey constraint should enter the experiment portfolio?
Decision output: A ranked experiment brief with evidence inputs, treatment hypothesis, threshold, guardrails, and next-learning rule.
Claim boundary: The displayed path values are altered and should not be used as a client benchmark.

Figure 01 / Evidence companion

The method, made inspectable.

Sanitized reconstruction. The visual is paired with HTML interpretation so the argument does not depend on screenshot pixels.

Open evidence record Decision workspace / TX-WS-CRO-01

Artifact: Decision workspace
Evidence status: Sanitized reconstruction
Record ID: TX-WS-CRO-01
Record: Sanitized purchase-path reconstruction / values altered / pre-launch decision brief
Decision question: Which journey constraint should enter the experiment portfolio?

Purchase path analysis

Checkout initiation is the shared constraint.

12hypotheses4ready

Read every supporting value in the purchase-path record below.

ChannelLanding → cartCart → checkoutCheckout → purchaseOverall

Paid search 72.6% 7.2% 89% 4.6%

Paid social 16% 19.6% 62% 1.9%

Direct 39.5% 8.6% 77.4% 2.6%

Organic search 45.4% 7.9% 91% 3.6%

Ranked opportunity01

Recover the cart-to-checkout transition.

Users add with intent, then lose confidence before checkout. The pattern repeats across acquisition sources.

P0Side-cart reassurance + shipping progressHigh reach · high confidence
P1Benefit-led collection-to-PDP routingMedium reach · high evidence
P1Purpose-built paid-social landing pathHigh reach · validate first

Ranking logicWhy P0 is first

Reach: Four acquisition pathsThe checkout-initiation constraint appears across every observed entry route.
Value: Checkout initiationThe break occurs after demonstrated cart intent and before the purchase path can progress.
Confidence: Telemetry + session evidenceIndependent behavioral signals point to the same friction before the experiment begins.
Readiness: Pre-registered exposureA bounded treatment, primary measure, guardrails, and decision threshold are already defined.

Priority is a pre-launch evidence rank, not an outcome forecast.

First decisionShip the side-cart test with a pre-registered checkout-initiation threshold.

Experiment decision brief / Pre-launch

Side-cart reassurance and shipping progress

No outcome is claimed. This is the decision package required before controlled exposure begins.

Signal: Cart intent survives across sources; checkout initiation does not.
Hypothesis: Making fulfillment cost and progress explicit before checkout will increase initiation.
Primary measure: Checkout initiation per eligible cart session
Decision threshold: At least 3% relative lift with the 90% interval excluding zero
Owner: Growth product + analytics
Guardrail: Purchase completion and support-contact rate do not deteriorate

Use horizontal scrolling to inspect every evidence source and its role.

EvidenceStateWhat it supports

Event telemetryObservedCheckout initiation is weak across all four acquisition paths.

Session evidenceCorroboratesDelivery cost and reassurance are repeatedly sought before checkout.

Prior learningNo conflictThe experiment archive contains no completed test of this combined treatment.

Decision workspace · sanitized reconstruction · values alteredObserved signal → ranked opportunity → test decision

Figure 01 / Analytical record

Cross-channel opportunity and experiment brief

R003-F01

Source record: Sanitized client-pattern reconstruction
Evidence status: Sanitized reconstruction
Method: Cross-channel path and opportunity analysis
What to read: Read the repeated cart-to-checkout constraint across acquisition paths, the evidence supporting its priority, and the pre-registered rule required before a result can change the roadmap.
Limitation: The displayed path values are altered and should not be used as a client benchmark.
Decision use: Which journey constraint should enter the experiment portfolio?

Paper dossier

Decision question: Which journey constraint should enter the experiment portfolio?
Primary method: Cross-channel path and opportunity analysis
Evidence status: Sanitized reconstruction

UX & conversionExperimentationJourney analysis

Editorial responsibility Tangentix Research

Review state: Published / Revision 1.1
Last reviewed: July 17, 2026
Primary sources checked: July 17, 2026

Tangentix Research owns the method framing, evidence-status label, stated limitations, decision rule, and linked-source check. No client result is inferred beyond the evidence disclosed here.

Cite this paper

Tangentix Research. “A purchase path is not a funnel.” Tangentix Research, July 5, 2026. https://tangentix.io/research/purchase-path-is-not-a-funnel/

An aggregate funnel is a useful orientation device. It is a poor diagnosis.

It combines customers who entered with different intent, landed on different surfaces, used different devices, and encountered different product and checkout conditions. The average drop-off can therefore describe nobody in particular.

Start with paths, not pages

A purchase-path analysis treats the journey as a sequence of states:

entry or landing surface;
browse or product-detail interaction;
add to cart;
checkout initiation;
purchase confirmation.

The path is then segmented by acquisition source, device, landing context, product type, and other factors capable of changing intent or friction.

Look for repeated constraints

The strongest opportunity is not always the largest visible drop. It is the constraint that repeats across economically meaningful paths and can plausibly be changed.

In the sanitized client pattern reconstructed beside this paper, cart creation is healthy for several paths while checkout initiation remains weak across acquisition sources. That changes the diagnosis. More traffic or a new product page will not resolve a confidence failure inside the cart-to-checkout transition.

Translate the pattern into a causal hypothesis

An opportunity statement should name the evidence without pretending it already proves the solution.

Observed signal: customers add with intent but fail to initiate checkout across multiple channels.

Interpretation: uncertainty about delivery cost, timing, or return conditions may be suppressing the next step.

Hypothesis: presenting reassurance and progress inside a persistent side cart will increase checkout initiation for eligible sessions.

The interpretation is deliberately falsifiable. It can be wrong, and the test is designed to find out.

Rank the portfolio, not isolated ideas

Tangentix ranks experiment opportunities against four dimensions:

economic reach;
evidence strength;
expected decision value;
implementation and measurement cost.

The first experiment is the one that can settle the most consequential uncertainty—not necessarily the one that is easiest to ship.

Pre-register the decision

Before launch, define the primary metric, guardrails, minimum detectable effect, exposure rules, stopping conditions, and the action attached to each outcome.

Scale if the credible lift clears the commercial threshold without harming guardrails.
Iterate if the direction is promising but the mechanism remains unresolved.
Stop if the result rules out a decision-relevant effect.

The artifact after the test should preserve the rejected explanations as carefully as the winning one. That is how an experimentation program becomes a learning system rather than a queue of disconnected variants.

Research record

Evidence, limitations, and reproducibility.

Revision: 1.1
Published: July 5, 2026
Last reviewed: July 17, 2026

Evidence record

Use horizontal scrolling to inspect every source and its role in the argument.

SourceStatusRole in the argument

Cross-channel purchase-path reconstructionSanitized client pattern / values alteredLocates the repeated cart-to-checkout constraint

Session-evidence synthesisCorroborating qualitative evidenceSupports the delivery-cost and reassurance interpretation

Experiment decision briefPre-launch / result not observedDefines the hypothesis, threshold, guardrail, stop rule, and scale rule

Known limitations

01
The displayed path values are altered and should not be used as a client benchmark.
02
Behavioral and session evidence supports prioritization but does not prove the proposed treatment.
03
No experiment outcome is claimed; the brief remains pre-launch.

Selected references

Primary sources behind the method.

These sources provide methodological context. Their inclusion does not imply that an external study validates Tangentix client outcomes or the illustrative evidence shown here.

01 Kohavi, Ron, et al. Online Experimentation at Microsoft. Microsoft Research, 2009.
Primary practitioner reference for operating controlled experiments as a repeatable decision system rather than a collection of isolated tests.
02 Gupchup, Jayant, et al. Trustworthy Experimentation Under Telemetry Loss. CIKM, 2018.
Primary reference for how instrumentation loss can bias experiment conclusions and why telemetry integrity belongs inside the decision gate.

Reproducibility note

What another analyst would need to preserve.

Recreate the path definitions, eligibility rules, channel segments, event-quality checks, and experiment archive review; pre-register the primary measure and guardrails before exposure.

Applied method / UX & Conversion

Turn customer friction into a governed experiment.

A paper establishes the method and its limits. The applied system puts the method beside the accountable decision, working evidence, and release condition.

Decision it informs: Which journey constraint should enter the experiment portfolio?
Working output: Conversion opportunity assessment
First evidence to inspect: Journey and path data, telemetry, customer evidence, and delivery constraints around the candidate change.
Application boundary: The method informs a decision only after the available evidence, counterfactual, and accountable owner have been checked.