What Should Count as a Marketing Conversion?

The short answer

A conversion is an action that changes what the business does next. A quote request does: someone has to respond. A booking does: a room or a slot is now taken. A scroll to the bottom of the services page does not. Nobody picks up the phone because of it.

That is the whole principle, but applying it is where most setups go wrong. Analytics tools make every interaction countable, and a count that rises feels like progress. So scrolls, video plays, PDF downloads, phone clicks and newsletter signups pile into one “conversions” total. The total looks good. It also cannot tell leadership whether marketing produced business, and when ad platforms are told to chase it, they find the cheapest version of whatever is in the pile.

This article sets out a taxonomy for deciding what counts. It goes deeper on one part of the marketing measurement plan: the definitions. The plan decides which KPIs matter; this decides which actions feed them.

Three questions for any candidate action

Before adding anything to the conversion list, ask:

1. Does it show commitment? Did the person give you something (contact details, time, money, a specific request) or just look at something?
2. Would the business act on it? Does someone need to respond, schedule, quote, ship or follow up?
3. Can it be verified? Is there a record outside the analytics tool, in a form system, inbox, booking engine, phone log or CRM, that confirms it happened?

An action that passes all three is a conversion. One that passes the first only partly, such as a newsletter signup, is a useful signal worth tracking but not worth counting alongside inquiries. One that passes none is engagement: diagnostic, sometimes interesting, never a conversion.

The four tiers

Tier one: engagement signals

Page views, scroll depth, video plays, time on page, visits to a pricing or careers page, clicks on an outbound link. These describe behaviour. They help explain why conversions moved (fewer people reached the form, or more people left the page early) and they help diagnose page problems. Track the ones you will actually look at. Do not mark them as key events in GA4 and do not send them to ad platforms as conversions.

Tier two: leads

Actions where a person identifies themselves and asks for something: a quote or contact form, a booking request, a demo request, a call, an event registration with real intent, a job application if recruitment is a goal. These are your primary website conversions. Each needs a precise definition: what counts, how it is confirmed and how duplicates and spam are excluded.

Leads are where most marketing reporting stops, and that is the problem with most marketing reporting.

Tier three: qualified leads

Leads that meet written criteria agreed with sales: the right kind of client, the right service, the right area, a plausible need and timeline. Qualification usually happens in the CRM, after a person has reviewed the inquiry or had a first conversation. It is the first tier that tells you whether marketing attracted the right people rather than just people.

Tier four: revenue outcomes

The sale, the paid booking, the signed contract, the first invoice. For e-commerce and direct online booking, this can happen on the website and be tracked immediately. For most service businesses it happens weeks or months later in the CRM or accounting system, and connecting it back to the original source is the work described in connecting marketing reporting to sales outcomes.

Report every tier, separately, by source. The ratio between tiers is often more telling than any single number. A channel with many leads and few qualified ones has a targeting or message problem. A channel with few leads and many qualified ones may deserve more budget, not less.

The edge cases everyone argues about

• Phone number clicks. A tap on a phone link shows intent, but it is not a call. Many taps go nowhere. Track them as a secondary signal. If calls matter, use call tracking that records connected calls of a minimum length, and count those as leads.
• Email address clicks. Even weaker. The click opens a mail app; whether an email follows is unknown. Secondary at best.
• Chat. A conversation started is engagement. A conversation where the visitor leaves contact details and asks for something is a lead. Most chat tools can distinguish the two if asked.
• Newsletter signups. Valuable for building an audience, but a different thing from an inquiry. Count them as their own goal, not as leads.
• Downloads. An ungated brochure download is engagement. A gated one that captures contact details is closer to a lead, but treat it as a separate, lower-intent category unless sales actually follows up on it.
• Directions and location clicks. For clinics, retailers and attractions, a request for directions is a strong signal, but unverifiable. Report it alongside Business Profile actions as a secondary indicator.
• Existing customers using the contact form. Not new business. Give them a separate route, or a form option, so they can be excluded.
• Job applications. A conversion if recruitment is a marketing goal, and nothing to do with sales leads. Report it separately or it will flatter the lead count in a hiring season.

What this looks like for different businesses

These are illustrative, not prescriptive. The right list depends on how each business makes money.

• A professional services firm (say, an accounting practice with three offices): leads are consultation requests and connected calls; qualified leads are those matching the client profile; outcomes are engagements signed. Newsletter signups and guide downloads are tracked as secondary.
• A tourism operator selling through a third-party booking engine: the conversion is a completed booking with its value, which requires tracking through or back from the engine. Click-outs to the booking engine are a secondary signal, useful when the booking itself cannot be seen. Our article on what a tourism board should measure covers destination organizations, where the equivalent is a referral to a partner.
• A multi-clinic healthcare practice: appointment requests and booked appointments by location are leads and outcomes; directions requests and phone taps are secondary. Location must be captured on every one.
• A B2B manufacturer: quote requests and spec sheet requests with contact details are leads; sales-accepted opportunities are qualified; purchase orders are outcomes, often months later.

What to tell the ad platforms

Automated bidding in Google Ads, Meta and LinkedIn optimizes toward whatever you call a conversion. That makes the definition a spending decision, not just a reporting one.

• Never optimize toward engagement. The platform will find people who scroll and play videos, and it is very good at it.
• Optimize toward the highest tier that occurs often enough to learn from. If you have a steady volume of qualified leads, importing them from the CRM as conversions teaches the platform to find more of them. If qualified leads are rare, optimize toward leads and judge campaigns on qualified results in your own reporting.
• Keep secondary actions secondary. Most platforms let you record an action for observation without using it for bidding. Use that for phone clicks and downloads.
• Assign values where you can. If a consultation request is worth more than a newsletter signup, tell the platform, even with rough relative values.

Our guides to lead volume versus lead quality and measuring lead quality from paid campaigns cover the advertising side in more depth.

Writing the definitions down

A conversion taxonomy is only useful if it is written, agreed and used the same way everywhere. For each conversion, record its tier, its exact definition, how it is confirmed, how duplicates and spam are excluded, its event name in GA4 and the ad platforms, and where it lands in the CRM.

“Form submission” is not a definition. “A successful submission of the quote form, counted once, confirmed by the form tool, excluding submissions marked as spam or from existing clients” is. It feels pedantic until two suppliers report different numbers and someone has to decide who is right. When GA4 and the CRM disagree anyway, which they will, why GA4 and your CRM report different lead numbers walks through the reconciliation.

Review the list once a year and whenever the business changes: a new service line, a new booking system, a new market. In our analytics and measurement work, agreeing this list with marketing, sales and leadership is one of the first deliverables, and the one most likely to change what an organization believes about which of its channels work.