Conversions in marketing

What is a conversion is and why it is important to track them?

Impressions or clicks

What is a conversion in digital marketing?

In digital marketing, a conversion is an event that is tracked and recorded when a campaign goal or interaction is complete. In a lot of cases, this will mean a sale.

For example, if a user clicks on a PPC search ad and buys a t-shirt from landing on the advertiser page, that would count as a sales conversion. A conversion is tracked via a conversion pixel (Javascript code that fires when a conversion event has been triggered), which tells the ad server that a conversion has occurred. This conversion will then be attributed to the ad that influenced the conversion.

Other conversion types could include clicks on a URL, a landing page visit, a newsletter sign-up, a white paper or brochure download, or a telephone call via a number that is linked to the digital campaign. Any trackable action or event that the marketer can add a conversion pixel to can count as a conversion.

In analytics systems (e.g. Google Analytics), these are often referred to as “events” or "goals", rather than conversions.

Conversion formula

To estimate the number of conversions, you will need the following metrics:

  1. Impressions or clicks delivered

  2. Conversion rate


Why is it important to track conversions?

The primary reason for tracking conversions is to understand the efficacy of your campaign. For example, if you run an e-commerce website and do not know which of your ads are driving sales, you could be potentially making poor use of your budget. Once conversion tracking is set up, you will be able to understand which tactic or strategy is driving the most efficient or highest value conversions. As you test and learn, you will be able to confidently shift budget to better performing areas to make the most effective use of your marketing spend.

Alongside conversions, it is important to measure the conversion rate (of impressions or clicks to conversions) and the cost per acquisition (CPA) to fully understand how hard your activity is working.

You should set up and track conversions in a way that reflects your marketing funnel. Once you have your funnel fully tracked, you will be able to understand where people are dropping off as they are pushed down the funnel, as well as creating lookalike audiences from high-value converters.

Click-through and View-through Conversions

There are two types of conversions that are typically tracked in ad servers: click-through and view-through conversions, also referred to as post-click or post-view conversions.

  • Click-through conversions (or post-click) are conversions that were driven from a user clicking on your ad and converting then or at a later date (within your defined conversion window).
  • View-through conversions (or post-view) are conversions that were driven from a user being served your ad, not clicking on it, and then converting then or at a later date (within your defined conversion window). This type of conversion was borne from the need to attribute ads (such as digital display banners, that generate fewer amounts of clicks compared to other digital channels) that were seen and may have influenced the sale at a later date. Even if the user saw the ad but did not click, he or she could be subconsciously influenced to convert at a later date.

If view-through conversions are ignored, you could not be giving enough credit to the digital channels or formats that are influencing a conversion. Both types must be tracked and considered when evaluating the results of a campaign.

Conversion window

A conversion window is set up to define the number of days or time that the impression's influence will be valid for to count towards influencing a conversion. For both types of conversions (click and view), users often do not convert at the time of being served the ad. This means we need to define a window of influence that we feel is enough time for the user to still be influenced by it before converting.

If this window is too long, we may be giving too much credit to the ad if a user converts, for example, 30 days in the future, when in reality the influence of the ad might have worn off by then. For the inverse, we may be under-attributing the effect of the ad.

When defining a conversion window, you should take into consideration the purchase cycle of the product or service. A more expensive product will likely need a longer attribution window as people take longer to research and consider than a lower-cost item before converting.

Conversion attribution

Conversion attribution falls into either single-touch or multi-touch. Single-touch only looks at one ad impression while multi-touch takes into account multiple channels and ads.

Instead of looking at single conversions in isolation, setting up multi-touch conversion attribution will help you understand the path to conversion and which touchpoints the user was exposed to throughout the journey. In the real world, the user will be exposed to many ads before finally converting; conversion attribution modelling will alleviate these shortfalls.

In multi-touch attribution, giving different weightings to each touchpoint allows you to attribute the conversion across multiple touchpoints rather than the one that the user saw last before converting.

The most commonly used single-touch and multi-touch conversion attribution models are as follows:

  • Last click: the last ad that the user clicked on will be attributed the conversion, despite potentially being served multiple ads beforehand that are now not given any credit

  • First click: the first ad that the user clicked on will be attributed the conversion, regardless of any ad that was served afterwards

  • Multi-touch: the act of giving multiple touchpoints credit throughout the purchase journey:

    • Linear: each touch-point is given an equal proportion of the attribution
    • Time decay: the touchpoint that was closest to conversion will receive a higher share of the attribution
    • Cross-device: using probabilistic or deterministic modelling, you would track the number of devices and ads a user was exposed to before converting

Related metrics