Ecommerce conversion reviews: a screening method

What to take away

No conversion service can be ranked until its dossier clears every admission gate.
A dossier holds the contracting entity, the edition, the price and the recheck date.
Software and agency candidates share one template but never one score.
A synthetic journey is a non-production test that starts only after the gates clear.
Privacy, accessibility and payment gates never compensate for each other.

Start with one merchant job

For ecommerce conversion reviews, define the job before collecting any certificate. Here it is an England-based online store moving an adult customer from product page through basket and checkout.

Record the contracting entity, trading name, service, edition and check date. The government's Companies House information service shows registered address, incorporation date, officers and filed documents. None of that proves trading capacity or fitness for this store.

Freeze a baseline: journey, evidence sources, event definitions, price content and the decision owner.

Build the candidate table

One template serves software, payment and agency candidates. Every material statement needs a source, access date, scope and a recheck trigger.

Candidate	Type	Price on record	Primary stop gate
Optimizely	Experimentation and flags	Quote-based, often annual	Data and contract scope
VWO	A/B testing and surveys	Published plans from the low hundreds of pounds per month	Price and traffic limits
Convert.com	A/B testing	Published plans below £100 per month	Price and tracking
Contentsquare	Replay and experience analytics	Quote-based, often annual	Privacy and retention
Stripe	Payments	UK card rates typically around 1.5% plus 20p	Payments and money states
Mollie	Payments	European card rates often around 1.2% plus 20p	Payments and money states
Conversion.com	Agency, London	Quote only, no day rate on file	Contract and measurement
Hallam	Agency, Nottingham	Quote only, no day rate on file	Contract and exit
PRWD	Agency, Manchester	Quote only, no day rate on file	Accessibility and data

Conversion service candidates

Optimizely

Type: Experimentation
Pricing: Quote-based
Stop gate: Data scope

Adobe Target

Type: Enterprise testing
Pricing: Annual quote
Stop gate: Contract exit

VWO

Type: A/B testing
Pricing: Low hundreds
Stop gate: Price limits

AB Tasty

Type: Experimentation
Pricing: Quote-based
Stop gate: Accessibility

Rates appear once, in the price column, and are indicative published figures to recheck. Agency rows are not software, so ask each for a day rate and a fixed-scope price. No agency day rate sits on file here, which keeps every agency price gate open.

Example: the VWO dossier

Take VWO. Identity fields record the contracting entity, the plan tier and the traffic limit. The price field holds the published plan starting in the low hundreds of pounds per month.

The reviewer marks every field verified, unknown or failed. Price and traffic limits clear for that tier. Consent handling under PECR stays unknown until a source is filed, so the tracking gate remains open.

Verdict: HOLD. No synthetic journey starts while a gate is open. The complete order states note sets out which payment and fulfilment fields need mapping first.

Run a synthetic journey

Only a cleared dossier is tested. The run happens in a non-production environment with a written script and a rollback route.

Write the journey from landing page to order confirmation.
Assign one tester to each device class and record the exact build.
Capture each step's screen, timing and error message.
Log every consent banner, price change and refund link encountered.
Repeat the run twice and compare both records.

Synthetic journey test script

Show product evidence, price components, optional extra
Send valid, missing, late, duplicate events; keep raw record
Refuse non-exempt tracking; keep journey usable
Exercise payment authorised, captured, declined
Record cancellation, return, refund, dispute distinctly
Complete accessibility tasks with specialist method
Remove access; inspect logs, privileges, residual access
Roll back configuration; prove baseline restored

Differences between runs are observations, not conversion lift. A synthetic journey shows whether a service behaves as described. It cannot prove a revenue effect, so define the events with the conversion metrics method before you script anything.

Stop gates stay separate

A stop gate blocks admission on its own. A strong analytics feature cannot answer a confusing total price. The refund route is judged separately from checkout speed.

Confirm the total price is visible before payment.
Confirm the security and rollback evidence is on file.
Confirm the refund route is reachable.

The CMA's price-transparency guidance covers mandatory fees, taxes, charges and drip pricing in online retail.

The ICO's storage and access technologies guidance covers cookies, pixels, scripts and tags, plus the exceptions and objections that must be honoured.

A refused consent choice is a required test path, not missing data to be repaired.

A candidate that fails one gate is held, not scored down. Before signing, put the conversion supplier contract questions to each shortlisted candidate, because exit terms decide what a poor result costs you.

Common questions

Which services should a UK store compare first?

Start with the candidate whose job overlaps your blocked journey. A payment provider and a testing platform answer different questions.

What evidence goes into a dossier?

Contracting entity, edition, price, data map, source access date and a recheck trigger. Any field left unknown keeps its gate open.

How long should a synthetic journey run?

Long enough to repeat the script twice per device class. It measures behaviour and defects, not statistical significance.

What does a hold verdict change?

It keeps the candidate out of procurement until evidence arrives. It does not rank anyone, and it is not a rejection.