"Lead quality" is the most used and least defined phrase in performance marketing. A sales manager means leads that close. A compliance officer means leads with clean consent. A publisher often means leads that pass validation. All three are right, and a programme only works when it measures all three. This guide breaks lead quality into signals a buyer can actually check, at the moment a lead arrives, in the first contact attempt, and in the weeks that follow, without relying on anyone's headline claims.
Layer 1: the record itself
The first signals are visible in the data before anyone picks up the phone.
- Valid, reachable contact details. Phone numbers should be correctly formatted and in service, and the line type (mobile or landline) should be known. Email addresses should have valid syntax and a domain that accepts mail. Many publishers validate both in real time, which removes the most obvious waste.
- Internal consistency. The ZIP code should match the state, the stated age should match the product, a vehicle year should be plausible, and a loan amount should fall inside the range the form allows. Contradictions are a sign of careless entry at best, and fabricated data at worst.
- Plausible answers to qualifying questions. A consumer who answers every question with the first option, or completes a ten-step form in a few seconds, deserves scrutiny.
- Duplicates. Records should be checked against your own recent leads and customers, not only the publisher's database, within a window both sides agree.
Layer 2: consent and provenance
A record that converts but carries weak consent is a liability, not a good lead. Every record should arrive with, or be linked to, the evidence of how it was created: submission time, IP, browser details, the page address, the disclosure version displayed and the seller named. Where you require it, an independent third-party consent certificate adds an external record of the session.
Check that the evidence fits together. The timestamp should precede delivery by seconds, not hours or days. The IP location should be broadly consistent with the stated address. The page URL should be a funnel you approved. And the disclosure should name you. Provenance matters too: leads created on the publisher's own funnels are easier to verify than leads passed through several hands.
Layer 3: behaviour signals
Session behaviour separates real shoppers from bots and form-fillers. Useful signals include time on page and time to complete the form, whether fields were typed or pasted, how the visitor arrived (the campaign, keyword or placement), and the device and browser profile. Few buyers receive all of these directly, but it is reasonable to ask a publisher which behaviour checks run before delivery and what is filtered out as a result.
Layer 4: intent signals
Intent is the reason the consumer filled in the form, and it is the strongest predictor of whether a conversation will go anywhere. Intent signals are vertical-specific:
- In auto insurance, a renewal date in the near future or a recent change such as a new car or move.
- In mortgage, a defined purchase timeline or an identified property, or a clear refinance goal.
- In home improvement, a specific project, an owner-occupied home and a target start date.
- In legal intake, a recent incident date within the relevant limitation period and whether the person already has representation.
- In debt relief, an unsecured debt amount within the provider's programme range and current financial hardship.
Funnels that ask these questions honestly, and route answers to the right buyers, produce better conversations than funnels that ask as little as possible.
Layer 5: outcomes
Ultimately, quality is measured in your own systems. The useful outcome metrics, roughly in order, are contact rate, conversation or qualification rate, quote or appointment rate, and sale, bind, funded loan or signed case rate. Pair them with returns and complaints. A source that converts slightly worse but generates no complaints may be worth more than one that converts well and produces disputes.
Measure by source and sub-source, and look at trends over time rather than single days. Quality often drifts slowly as a publisher changes media buying, and a weekly view catches it before it becomes expensive.
Separating publisher problems from process problems
Before blaming a source, check your own process. Speed to contact has a large effect on whether a consumer can be reached, and a lead worked hours later will look worse than the same lead worked in minutes. Calling outside the consumer's local hours, using a number that displays poorly on caller ID, or routing leads to agents not licensed in the consumer's state will all depress results in ways the publisher cannot fix. Comparing sources only makes sense when they are worked the same way.
Turning signals into a feedback loop
- Agree which fields and evidence arrive with each record.
- Reject or return records that fail agreed validation rules, with the reason recorded.
- Share outcome flags (contacted, qualified, sold) by lead ID, weekly.
- Review results by sub-source together with the publisher, and agree changes.
- Audit a sample of consent evidence against live pages each month.
Publishers can only improve what they can see. A buyer that shares outcomes gets a source tuned to its business; a buyer that only sends returns gets a source tuned to avoid returns.
How Growth Box approaches quality
Growth Box validates phone and email in real time, applies buyer filters and duplicate windows before delivery, and attaches the full consent evidence to every record. Because we run our own funnels and media, we can tell you exactly where each lead came from and change it quickly when your outcome data says so. To set up a test with a feedback loop from the first week, get in touch.
This article is general information about measuring lead quality, not legal advice.
