Incremental Lift
Mira's standard measurement product
Incremental Lift Analysis is Mira's standard measurement product, and the only type of Study the platform runs. It compares the change in conversion rate from before a campaign to during it, in a group that was exposed to the media against a group that was not.
A Study is a child of a Campaign, and it holds the parameters that define the measurement: the dates, the attribution window, the cuts, and whether a pre period and a control group are used. One Campaign can have more than one Study - most commonly to measure the same campaign at two different attribution windows.
One Study Type, Two SwitchesEarlier versions of the platform exposed several named study types. There is now a single type, and what used to distinguish those types is expressed as configuration: whether the Study uses a pre period, and whether it uses a control group.
What Is Measured
A fully configured Study produces four conversion rates - one for each combination of group and period:
| Period | Control | Test (Exposed) |
|---|---|---|
| Pre | Pre Control | Pre Test |
| Post | Post Control | Post Test |
From these, three figures are derived:
- Exposed lift - the change in the exposed group's conversion rate from the pre period to the post period
- Control lift - the same change, in the control group
- Net lift - the difference between the two. This is the headline result, and it is the part attributable to the media
The control group's job is to absorb everything other than the campaign. Seasonality, other media, a product launch - all of these move both groups, so subtracting the control's lift from the exposed group's lift leaves the incremental effect of the OOH.
Configuration and What Each Produces
Both the pre period and the control group are optional, and the combination determines what the Study can report:
| Pre Period | Control | Result |
|---|---|---|
| Yes | Yes | Net lift - the full analysis, and the recommended configuration |
| No | Yes | A single lift, comparing the exposed and control groups during the campaign only |
| Yes | No | A single lift, comparing the exposed group against its own pre-campaign baseline |
| No | No | Conversion rates only, with no lift |
A control group requires no advance preparation, because it is constructed after the fact - see Control for how it is built and what it is matched on. A pre period does: the conversion environment has to already be collecting data before the pre period starts. See Conversion Properties for which environments need setting up ahead of a campaign.
Partial ResultsA Study that requests a control may report its exposed figures before the control is available, and gain the control-based metrics later. What appears in a report depends on what has been computed, which will not always match everything the Study requested.
Attribution Window
The attribution window, sometimes called a lookback window or conversion window, is the amount of time a user has to convert after being exposed in order to be counted as an attributed conversion.
If a user is exposed at 2026-03-02 10:21:00 and the window is 7 days, they count as converted only if a conversion event falls between that moment and 2026-03-09 10:20:59. A conversion on 2026-03-11 would not be attributed.
Choosing a window is a judgement about the use case rather than a statistical decision. We typically recommend at least 7 days and no more than 30. The shorter the window, the stronger the claim that an attributed conversion was actually caused by the exposure; the longer the window, the more conversions are captured.
Study Periods
The post period runs from the start of the media campaign to the end of the campaign plus the attribution window. For a four-week campaign running 1 June to 28 June with a 7-day window, the post period spans 1 June to 5 July. Exposures only occur during the campaign itself - the trailing stretch exists so that the attribution window can lapse for users exposed near the end.
The pre period is a stretch of time before media begins, used to establish a baseline conversion rate for each group. It lasts as long as the attribution window. Ideally the control and exposed groups show similar conversion rates during the pre period, since neither has been exposed yet.
| Name | Start | End | Notes |
|---|---|---|---|
| Pre period | (Campaign start) - (attribution window) | Campaign start | Usually immediately precedes the campaign, but does not have to |
| Post period - exposures and conversions | Campaign start | Campaign end | Both exposures and conversions are accruing |
| Post period - conversions only | Campaign end | Campaign end + attribution window | No further exposures; the attribution window lapses |
Conversion Data Is Needed For Every PeriodExposures only happen during the campaign, but conversion data is required for the entirety of both periods - including the pre period and the trailing window.
Pre period dates are derived from the campaign dates and the attribution window rather than set directly. If conversion data does not reach far enough back to cover a full pre period, the pre period is shortened to what the data supports.
Cuts
The simplest Study reports a single set of numbers: a control conversion rate, an exposed conversion rate, and the lift between them. Cutting is the practice of grouping those results into subgroups and observing the effect within each. Market of exposure, property, and goal cuts are always included, but users can request additional cuts on other fields, such as:
- a standard metadata field from the source data, such as
venue_type,vendor_name, orcreative_id - a custom metadata field, carried through from the source under the custom prefix
- a label applied when a Source was associated with the Campaign - useful for a dimension that isn't in the underlying data at all, such as
{"flight": "summer"}
See Standard Schemas for OOH Data Onboarding for the standard and custom metadata fields available, and Configuring Sources for Campaigns for applying labels and filters.
Exposure-Side and Conversion-Side Cuts
Cuts originate on either side of the attribution join, and the two behave differently.
Exposure-side cuts group on a property of the media - market of exposure, vendor, format, creative, device type. These reduce both the numerator and the denominator of a conversion rate, because they narrow the set of people under consideration.
Conversion-side cuts group on a property of the conversion - property, goal, market of conversion for footfall. These reduce only the numerator, because they narrow only the qualifying events.
Cuts Need To Be In The DataSome cuts we can produce without anything extra from you, because we can infer them - market of exposure, property, and goal. Anything else has to be present in the files you provide. If you want results broken out by format or creative name, those fields must be in the exposure data.
Cuts are also computed in combination, so a Study cut by market and creative reports overall, by market, by creative, and by market and creative together.
Results
A Study's results are refreshed as exposure and conversion data accumulates, and are final once the Study has been delivered. See Incremental Lift Reports for what the Console presents and how to read it, and Feasibility for what makes a campaign large enough to produce a conclusive one.
For setting a Study up, submitting it, and tracking its status in the Console, see Configuring Measurement Studies.
Updated about 13 hours ago