Reviewing a lead generation agency before any live engagement

What to take away

No England lead generation agency is rated on this page. It is a review protocol you run yourself, because no client evidence or live campaign lies behind it.
Trust is granted per engagement, not per brand. Check the legal entity, the permission chain behind each lead, claim substantiation and supplier security before spend.
Six gates decide appointmentprovenance, advertising, security, accessibility, commercial control and exit. Strength on one gate does not offset a failure on another.
Run a bounded test of rejection, duplicate routing, suppression and evidence export before you allow live volume.

What a review can settle before you commit

A managed supplier that sources, routes or follows up sales enquiries is in scope. Pure software vendors, trade directories and list sellers are not. Build the candidate list from your required categories and regions, not from search ranking, and our candidate discovery method explains why that order matters.

The useful question is narrower than reputation. Can one sampled lead be traced from the notice shown at capture, through the transfer, to the contact, objection and deletion records? If not, the review stops there.

Verifying the legal entity and service boundary

Ask each candidate for its legal entity, trading names, company number, service boundary and subcontractors. Reconcile every item against Companies House and against the contract you would sign. The register warns that filed information is not independently checked, so active status proves registration only.

Each channel also needs a map from collection notice to source, disclosure, transfer, contact, objection and deletion.

Channel data journey map

Collection notice
Source
Recipient disclosure
Transfer
Contact
Objection
Deletion

The contract should name that entity and every subcontractor touching lead records.

Six gates that no proposal can offset

Controller and processor roles follow the purposes and means of processing, not the label printed in a proposal. Score each gate on its own evidence.

Non-compensating review gates

Provenance and permission
Advertising claim substantiation
Security responsibilities evidenced
Accessibility routes
Commercial control and billing
Exit and deletion evidence

The question you put

Provenance and permission: Can one sampled record be rebuilt through its full journey?
Advertising: Who approves objective claims and holds the substantiation?
Security: How are access, incidents, subcontractors and recovery handled?
Accessibility: Can participants and recipients complete, correct or object?
Commercial control: What event is billable, and who can stop spend?
Exit: Can you export evidence and propagate suppression?

Evidence that settles it

Provenance and permission: Dated consent or legitimate-interests record, source and disclosure wording
Advertising: Documentary evidence for each claim
Security: Written answers tested against your own scenarios
Accessibility: A tested route through to error recovery
Commercial control: Rejection and credit rules, named stop authority
Exit: Export schema, suppression confirmation, deletion record

The ICO's lead collection guidance supports checking bought or generated information rather than relying on a warranty.

The CAP Code's misleading advertising section requires documentary support for objective claims.

Use the NCSC's supplier assurance questions as prompts for security, not as a certificate.

Example: the provenance gate applied to a named marketplace

Take Bark.com, a UK lead marketplace for trades, as the worked case, then apply the same test to MyBuilder. Our marketplace comparison sets out how those platforms differ for buyers.

Put the request in writing. Show the collection notice at capture, the source of the record, and the disclosure wording at transfer. Then ask for the date of last contact, the objection route and the deletion trigger. A platform that cannot rebuild one record fails the gate.

Costs and commercial benchmarks

What is billed

Per accepted lead: Each enquiry meeting the agreed criteria
Retainer plus media spend: Management time and advertising spend
Share of closed revenue: A proportion of revenue from accepted leads

Evidence to hold

Per accepted lead: Acceptance and rejection log, credit record
Retainer plus media spend: Platform access and spend statements
Share of closed revenue: Agreed attribution rule and revenue definition

Per-lead prices move with sector, exclusivity and geography. A single figure is a data point, not a benchmark, unless the sample, period and source universe are published, as our benchmark research argues.

Trial, exit terms and stop authority

An agency-selected case study is supplier evidence, not independent review. With the referee's consent, check who commissioned the work, the period, the comparable scope and whether the referee can substantiate the statement.

Do not request confidential personal data, and do not read silence as poor quality.

Reconcile the legal entity, company number and subcontractors against Companies House and the contract.
Map each channel from collection notice to deletion, with one record rebuilt end to end.
Check that objective advertising claims carry documentary substantiation.
Confirm access, incident, subcontractor and recovery duties in writing.
Test rejection, duplicate routing, suppression, accessible error recovery and evidence export before appointment.
Name territory, volume ceiling, review date and stop authority in any proceed decision.

Common questions

Why is no agency rated here?

The research did not establish a complete eligible population, obtain client evidence or run a live campaign. The page gives you the gates instead, so any trust verdict rests on evidence you gather.

What should I check about an agency's references?

An agency-selected case study is supplier evidence. With the referee's consent, verify who commissioned the work, the period, the comparable scope and whether the referee can substantiate the statement.

Which tests should run before appointment?

Run synthetic or properly authorised tests of rejection, duplicate routing, named-recipient disclosure, suppression, wrong-recipient escalation, evidence export, accessible error recovery and rollback. Record who ran each test and what artefact it produced.

How should I read a published cost per lead figure?

Only with its source, period and sample. Without those, it is one data point from one campaign, not an England benchmark.