Enhanced Conversion Measurement

Overview

The Enhanced Conversion Measurement (ECM) project introduces a robust, closed-loop attribution framework to measure the true impact of advertisement exposure on actual sales. By utilizing deterministic matching (see the list of match keys below), the system maps the customer journey from digital touchpoint exposure to final purchase, completely bypassing the limitations of cookie-based tracking.

ECM goes beyond simple attribution by offering insights into Sales Uplift (incrementality), allowing advertisers to filter out "organic" sales and isolate the incremental conversions generated specifically by their campaign investment.

Match keys include:

• email or hashed email
• phone number or phone_number_hashed
• mobile_identifier or mobile_identifier_hashed
• address details such as:
  • zip
  • street_name
  • street_number
  • street_number_extension
• name details:
  • first_name
  • last_name
  • birth_date

For Belgium, when matching addresses, the street_name is mandatory when you use an address.

What We Measure: Conversions vs. Incrementality

There are two options for Conversion Measurement, depending on when the measurement is requested and how the campaign is set up.

1. Simple Conversions & Conversion Rate
   Measures the reach of the exposed audience, the number of sales attributed to that audience, and the overall conversion percentage. This gives you an absolute number of conversions.
   
   
2. Sales Uplift (Incrementality)
   Compares the conversion behavior of the exposed group against a curated hold-out group. 

Pre-Launch vs. Post-Launch Requests

Because measuring Sales Uplift requires creating a clean control group before ad delivery begins, the timing of the ECM request dictates the available metrics:

• Campaign is NOT live yet + it uses audience (Ratecard or CRM) targeting: Advertiser receives Sales uplift % (Incrementality), Conversions, and CVR. (The measurement request must be made at least 3 working days prior to launch during our test phase so the data team can build the hold-out/lookalike groups)
  
  
• Campaign is already live + uses audience (Ratecard or CRM) targeting: Advertiser receives Conversions and CVR only. (Post-hoc incrementality analysis is not supported due to statistical noise).

Conversion Measurement Tiers

We classify campaigns into two distinct tiers based on the targeting method used.

Campaign Viability & Prerequisites

To ensure data integrity and avoid "noise-to-signal" issues, campaigns must meet several requirements before qualifying for ECM:

1. Identity & Targeting Requirements

• Known User Targeting
  Advertisers must target audiences that are maximally identifiable. The preferred targeting option is therefore a CRM audience as these are known by both DPG Media as well as the advertiser. If these are not available, ratecard audiences are an alternative due to the large number of logged-in users present there.
  
  
• Pre-Campaign Uploads (Tier 1)
  For CRM targeting, the targeting file must be uploaded via a Datalab or a supported Data Clean Room (Infosum, Liveramp, Snowflake)  before DPG can build the hold-out group.

2. Budget & Statistical Power

• The campaign budget must be large enough to generate a "statistically significant" number of conversions.
• Rule of Thumb: There must be enough impressions to generate at least 100–200 conversions in the exposed group to make the uplift calculation meaningful.

3. Data Integrity & Attribution Window

• 10-Day Cooling-Off Period: Advertisers must agree to a standard 10-day post-campaign attribution period after the campaign ends before uploading their sales/conversion file (via Datalab, InfoSum, Snowflake, or LiveRamp). This ensures all "latent" conversions triggered by the ad are fully captured.

Behind the Scenes: How We Measure Uplift

For advertisers who choose pre-campaign Incrementality measurement, we use advanced measurement methods to accurately calculate the real impact of a campaign while maintaining maximum campaign reach.

Ghost Bid / Small Control Group Method (Preferred Approach) 

For most campaigns, we use a Ghost Bid methodology with a small control group. The system randomly splits the target audience into two groups:

• 95% can see the campaign
• 5% are intentionally not shown the ads.

Both groups are equally likely to be interested in the product or service. By comparing sales or conversions between the exposed and non-exposed groups, we can accurately measure the additional impact generated by the advertising. This approach provides the most direct and reliable view of campaign uplift while still preserving nearly the full campaign scale.

Similar Audience Comparison

In specific cases, we use a similar audience comparison approach instead. This is typically applied when:

• the campaign targets a relatively small CRM audience, or
• the full target audience has already been exposed to the campaign.

In these situations, we compare the campaign results of the exposed CRM audience with a separate group of users who closely match the same demographic and profile characteristics but did not see the campaign. This allows us to estimate the campaign uplift without reducing delivery to the intended audience.