Betting Campaign Incrementality: How to Measure Lift
How to Measure Betting Campaign Incrementality
A betting campaign can deliver thousands of tracked registrations and still create very little new commercial value.
That is the limitation of attribution.
Advertising platforms, affiliate systems and analytics tools are designed to assign credit for conversions. They can show that a player clicked an advert, registered after visiting an affiliate or deposited within a platform's attribution window.
What they cannot automatically prove is that the marketing activity caused that behaviour.
A player credited to paid search may already have intended to register. A sportsbook customer receiving a CRM reactivation message may have been planning to deposit that weekend anyway. An affiliate may appear immediately before conversion without having introduced the player to the brand.
Betting campaign incrementality asks the harder question:
What would have happened if the campaign had not run?
The difference between the observed result and that expected baseline is the incremental effect of the marketing activity.
For operators facing rising acquisition costs, overlapping media channels, volatile sporting calendars and increasing scrutiny of player quality, that distinction matters. Incrementality moves the discussion away from who received conversion credit and towards whether marketing genuinely changed player behaviour.
In short: Betting campaign incrementality measures the additional registrations, verified players, first-time depositors, deposits, revenue or profit caused by marketing activity above what would have happened without it. Operators can estimate this using methods such as geo experiments, audience holdouts and platform lift studies, then validate whether the incremental players acquired are commercially valuable.
What betting campaign incrementality means
Incrementality is the lift created by an intervention above the outcome that would otherwise have occurred naturally.
That natural outcome is known as the counterfactual.
The difficulty is that the counterfactual cannot be observed directly.
Once an operator launches a campaign, it can see what happened with the campaign running. It cannot simultaneously observe exactly what the same players, in the same market, during the same sporting events, would have done without it.
Incrementality testing attempts to estimate that missing outcome using a suitable control group.
For example, suppose a paid social campaign is credited with 1,000 first-time depositors.
Attribution reporting may therefore suggest:
Attributed FTDs: 1,000
But a comparable group that was not exposed to the campaign produces an equivalent rate corresponding to 800 expected FTDs.
The estimated incremental effect is therefore:
1,000 observed FTDs - 800 expected FTDs = 200 incremental FTDs
The campaign may still have influenced 1,000 tracked conversions.
Commercially, however, the test suggests only around 200 were additional to what would probably have happened anyway.
That is a very different basis for evaluating acquisition spend.
Attribution and incrementality answer different questions
Attribution and incrementality are related, but they should not be treated as interchangeable.
Attribution asks who should receive credit
An attribution model might determine that a conversion belongs to:
paid search;
paid social;
an affiliate;
direct traffic;
CRM;
another measurable source.
Different models can distribute that credit differently.
Last-click may favour search.
First-touch may favour paid social.
An affiliate platform may claim the player if its tracking link appeared shortly before registration.
Incrementality asks whether the activity caused additional behaviour
Incrementality is less concerned with which touchpoint appeared in the conversion path.
It asks whether the conversion would probably have happened without the marketing intervention.
This is especially important in betting because natural demand can be unusually strong.
Major sporting events, brand recognition, organic search activity, price boosts, media coverage and competitor behaviour can all increase registrations and deposits regardless of a particular campaign.
A channel can therefore have an excellent attributed CPA while generating limited incremental growth.
Why incrementality matters in betting
Gambling acquisition contains several characteristics that make incrementality particularly useful.
Sporting calendars create natural demand
A major football final, racing festival or international tournament can cause registration and deposit volumes to rise even without additional marketing.
A simple before-and-after analysis may therefore exaggerate the effect of a campaign running during that period.
Players encounter several acquisition channels
A potential bettor might:
See a paid social advert.
Search for the operator.
Visit an affiliate comparison site.
Click a branded search advert.
Register.
Receive an onboarding CRM message.
Deposit.
Several systems could legitimately claim involvement.
Only an incrementality test can begin to answer whether a particular intervention actually increased the probability of the outcome.
Brand search can capture existing intent
Branded paid search can often look exceptionally efficient.
That does not make it unnecessary.
But operators may want to understand how many players would have reached the website organically if the advert had not appeared.
CRM can receive credit for behaviour that was already likely
A reactivation campaign may generate deposits shortly after messages are sent.
Some recipients would have returned anyway.
A holdout group helps estimate the additional behaviour caused by the communication.
Affiliate journeys can contain interception
An affiliate positioned close to conversion may receive credit for players whose original intent came from another source.
Incrementality testing can help operators distinguish genuine customer introduction from conversion interception.
Start with a commercial decision, not a dashboard
The most important part of an incrementality test comes before any data is analysed.
Define the decision the experiment needs to inform.
Possible questions include:
Does paid social create new sportsbook customers?
Is generic paid search generating incremental FTDs?
Does brand search increase acquisition or mostly capture existing demand?
Does a welcome offer generate additional player value?
Does an affiliate placement introduce genuinely new players?
Does CRM reactivation change deposit behaviour?
Should an operator increase spending in a particular region?
Does video advertising generate additional registrations that later become valuable players?
The experiment should be designed around one primary question.
Write a clear hypothesis before the test
A useful hypothesis should state the intervention, the expected outcome and the commercial threshold.
For example:
Increasing investment in generic football-betting search terms will generate incremental 30-day net gaming revenue at an acceptable cost per incremental FTD.
That is more commercially useful than:
Does generic search generate conversions?
The second question can be answered by attribution reporting.
The first requires causal evidence.
Define the success metric upfront
The team should agree the primary outcome before seeing the result.
That might be:
incremental registrations;
incremental verified accounts;
incremental FTDs;
incremental deposit value;
incremental 30-day NGR;
incremental profit.
Secondary metrics can still be evaluated.
However, choosing the main success measure in advance reduces the temptation to search afterwards for whichever KPI makes the campaign appear successful.
Define the test population and exposure rules
A credible betting incrementality test needs a clear definition of who is eligible to enter it.
Teams should document:
target audience;
market;
player eligibility;
test period;
control population;
media exposure;
spend;
channel exclusions;
success threshold.
This becomes particularly important when players can be reached through several marketing systems.
A customer withheld from an email campaign but simultaneously targeted through push notifications and paid retargeting is not a clean control.
Exposure rules should therefore account for the other channels capable of influencing the same behaviour.
Choose a representative betting period
Sporting calendars can materially distort campaign performance.
An experiment conducted entirely during a major tournament may not represent normal acquisition conditions.
Likewise, testing sportsbook advertising only during an international break could understate the channel's normal contribution.
Where practical, the test window should contain a representative mix of demand.
Relevant contextual factors may include:
major football fixtures;
racing meetings;
international tournaments;
seasonal betting peaks;
competitor promotions;
operator offer changes.
The objective is not to remove every source of variation.
It is to avoid interpreting an unusual trading period as universal behaviour.
How to measure betting campaign incrementality
There is no single incrementality methodology suitable for every betting campaign.
The strongest approach depends on:
channel;
available identity;
geography;
audience size;
conversion volume;
technical infrastructure;
market restrictions.
Three particularly useful approaches are geo experiments, audience holdouts and platform lift studies.
Use geo experiments for broad acquisition campaigns
Geo experiments divide comparable geographic areas into test and control groups.
The campaign runs, increases or changes in the test regions while activity is withheld, reduced or left unchanged in comparable control regions.
The difference between the groups helps estimate incremental lift.
This can be useful for channels such as:
paid social;
display;
online video;
broader awareness activity;
some paid search strategies;
controlled affiliate activation.
Example of a geo incrementality test
Imagine an operator identifies two groups of regions with historically similar:
FTD volumes;
deposit behaviour;
channel mix;
sporting demand.
Paid social spend is increased in the test regions while control regions continue at baseline activity.
Before the experiment:
Region groupWeekly FTDsTest regions1,000Control regions980
During the campaign:
Region groupWeekly FTDsTest regions1,300Control regions1,100
A simple comparison showing that test-region FTDs increased by 300 would exaggerate the likely campaign effect.
Control regions also increased.
The more useful analysis considers the relative change between both groups.
The test regions increased more strongly than the underlying market movement suggested by the control.
That additional change provides the basis for estimating incrementality.
Why before-and-after analysis is weak
Suppose registrations rise 25% after a new advertising campaign launches.
That looks encouraging.
But during the same period:
a major fixture occurred;
competitors reduced promotional activity;
organic demand increased;
another operator experienced downtime.
Without a control, it is difficult to know how much of the uplift came from the campaign.
A matched control provides evidence of what might have happened without the additional media.
Match geo groups carefully
Poorly matched test and control regions can undermine the experiment before it begins.
Relevant matching variables may include:
historical FTD volume;
deposit value;
brand awareness;
player density;
product mix;
seasonality;
media mix;
previous campaign exposure.
Teams should also watch for local factors that could contaminate results.
For example, football club sponsorship, local sporting interest or region-specific promotions may make two otherwise similar areas behave differently.
Where regulatory treatment differs between regions, operators should assess whether the comparison remains valid.
Use audience holdouts where player identity is available
Audience holdouts are often cleaner than geo tests when users can be identified and randomly assigned.
A percentage of eligible users is deliberately withheld from the marketing activity.
The remaining users receive the normal campaign.
The difference between the groups estimates the incremental effect.
Holdouts are particularly suitable for:
CRM;
reactivation;
push notifications;
logged-in website personalisation;
retargeting;
certain first-party audience campaigns.
Example: CRM reactivation incrementality
Suppose 100,000 dormant sportsbook users are eligible for a reactivation programme.
The operator randomly assigns:
90,000 to receive the campaign;
10,000 to a control group.
During the test:
Treatment group deposit rate: 6.0%
Control group deposit rate: 4.5%
The campaign is associated with 6% deposit behaviour.
However, 4.5% appears likely to have happened without it.
The estimated incremental lift is therefore around 1.5 percentage points.
The operator can then calculate how many additional depositors the campaign created and compare that with promotional and communication costs.
That is far more informative than simply reporting every deposit occurring after an email as campaign-generated revenue.
Keep holdouts genuinely unexposed
This sounds obvious but is often difficult operationally.
A player may be excluded from an email journey while still receiving:
push notifications;
SMS;
onsite messaging;
paid retargeting;
affiliate promotions.
If those activities communicate an equivalent proposition, the control becomes contaminated.
Testing teams should document which channels can reach the audience and establish appropriate suppression logic where feasible.
The more similar the test and control groups are apart from the intended intervention, the stronger the experiment.
Treat platform conversion-lift studies as useful evidence, not absolute truth
Major advertising platforms may offer their own lift-testing products.
These can be valuable because the platform may have access to impression-level exposure information that is difficult for the operator to reconstruct independently.
A conversion-lift study can compare exposed and withheld audiences to estimate additional outcomes generated by the campaign.
That can be especially useful for:
paid social;
video;
display.
However, the operator should still understand exactly what the study measures.
Questions to ask about a platform lift study
Review:
conversion definition;
test population;
holdout methodology;
sample size;
attribution window;
test duration;
statistical confidence;
campaign exclusions.
The platform might measure incremental registrations.
The operator may care primarily about incremental verified FTDs or 30-day NGR.
Both results can be useful, but they are not the same.
Platform studies should therefore be joined with first-party player data where appropriate rather than treated as the entire commercial answer.
Measure incremental player quality beyond first deposit
An incremental FTD is more valuable evidence than an attributed FTD.
It is still only an early acquisition event.
A campaign that genuinely generates new depositors can remain commercially unattractive if those players produce poor downstream value.
For acquisition tests, useful quality measures can include:
incremental registrations;
incremental verified accounts;
incremental FTDs;
first deposit value;
cost per incremental FTD;
bonus cost;
second deposit rate;
active betting days;
stake;
30-day NGR;
90-day NGR;
cross-sell into other products where relevant.
The appropriate measures will depend on the market, product and operator.
Cost per incremental FTD is different from reported CPA
Suppose a campaign costs £40,000 and attribution reports 1,000 FTDs.
The reported FTD CPA is:
£40,000 / 1,000 = £40
That may appear highly efficient.
But an incrementality experiment estimates that only 250 of those depositors were genuinely additional.
Cost per incremental FTD becomes:
£40,000 / 250 = £160
Neither figure is necessarily wrong.
They answer different questions.
The first describes attributed acquisition efficiency.
The second estimates the cost of creating genuinely additional depositors.
For strategic budget allocation, the second can be considerably more informative.
Measure incremental revenue, not only conversion volume
Betting campaigns can generate the same number of incremental players while producing very different economics.
Consider:
Campaign A
200 incremental FTDs
£100 average 30-day NGR per incremental FTD
Campaign B
200 incremental FTDs
£40 average 30-day NGR per incremental FTD
Conversion incrementality is identical.
Commercial value is not.
Operators should therefore extend incrementality analysis into revenue once cohorts have had enough time to mature.
Incremental profit is often the strongest commercial measure
Ultimately, operators need to understand whether incremental behaviour is profitable.
A simplified commercial calculation can consider:
Incremental revenue
minus media cost
minus affiliate commission
minus bonus cost
minus payment cost
minus material servicing costs
= estimated incremental profit
The exact calculation will vary by operator.
Some businesses may evaluate contribution margin, others payback period or projected lifetime value.
What matters is that the marketing cost is judged against the additional economic value created rather than against gross conversion volume alone.
Include player quality, risk and operational outcomes
Volume should not override quality.
A campaign may appear to generate cheap incremental deposits while also producing disproportionate:
failed KYC outcomes;
bonus abuse;
fraud indicators;
chargebacks;
low-value churn.
That does not automatically make the channel commercially attractive.
Relevant player-protection and safer gambling considerations should also be reflected in how results are interpreted, subject to the appropriate requirements and operator policies.
Quality definitions should therefore be agreed across teams including:
acquisition;
trading;
CRM;
finance;
data;
compliance.
This reduces the risk of marketing celebrating an incremental result that creates poor outcomes elsewhere in the business.
Make betting incrementality analysis credible
Incrementality should not produce another unexplained number in a dashboard.
A useful test readout should show:
observed outcome;
estimated counterfactual;
incremental difference;
confidence around the estimate;
test methodology;
known limitations;
external events affecting the period.
Teams should be able to explain how the result was produced.
Check the baseline before interpreting lift
Test and control groups should be sufficiently similar before the intervention starts.
For geo experiments, operators can compare historical:
FTD volume;
registrations;
deposit value;
revenue;
seasonality;
channel mix.
For audience experiments, teams can confirm that randomisation produced reasonably balanced groups across relevant characteristics such as:
previous activity;
value band;
eligibility;
product engagement.
If substantial differences already exist before the experiment, statistical adjustment or a redesigned test may be necessary before strong conclusions are drawn.
Understand confidence and uncertainty
Incrementality is an estimate.
It should be presented as one.
A test might suggest 10% incremental lift, but the true effect could plausibly sit within a wider range.
That uncertainty matters when making investment decisions.
A narrow confidence range around a positive result provides stronger evidence than a highly uncertain estimate based on a small sample.
Marketing leaders do not need to become statisticians.
They do need to avoid treating an uncertain estimate as an exact commercial truth.
An inconclusive test is not automatically a failed test
A campaign experiment may finish without producing a clear positive or negative answer.
Possible reasons include:
campaign effect was small;
insufficient conversion volume;
test duration was too short;
control contamination;
inconsistent execution;
unusually volatile player behaviour.
That information is still useful.
An inconclusive result means the available evidence is not strong enough to support a confident decision.
Scaling significant spend on that weak signal can be more expensive than extending or redesigning the experiment.
Avoid stopping tests after one strong weekend
Betting performance can fluctuate sharply around sporting events.
A campaign may produce excellent results during one Premier League weekend and weaker results immediately afterwards.
Stopping the test as soon as a promising result appears introduces bias.
Operators should establish the decision date and minimum test duration beforehand where practical.
This reduces the temptation to continue experiments when they look bad but stop them immediately when they appear successful.
Be cautious with small numbers of high-value players
Revenue-based incrementality can be particularly volatile.
A handful of high-value bettors can materially alter short-term NGR results.
For example, a test region may appear dramatically more profitable because several high-value players happened to win or lose significant amounts during the experiment.
That is why operators may want to assess:
conversion volume;
deposit behaviour;
player activity;
revenue over longer windows;
rather than interpreting a short burst of NGR as definitive proof.
As cohorts mature, the commercial picture should become more stable.
Build incrementality into normal betting campaign operations
Incrementality should not be reserved for a large annual measurement project.
Operators can use recurring experiments around material commercial decisions.
Useful testing moments include:
major budget increases;
new-market entry;
new advertising channels;
significant affiliate partnerships;
welcome-offer changes;
reactivation programmes;
major campaign launches;
shifts between brand and generic search investment.
A sequence of smaller, repeatable experiments can create a valuable internal evidence base.
Over time, teams can learn:
where paid media is most incremental;
which channels mainly capture existing demand;
which offers change player behaviour;
where affiliate relationships create genuinely new players;
what level of spend still produces acceptable marginal returns.
That evidence can improve future budget allocation.
Connect incrementality with marginal returns
A campaign can be incremental at one spend level and less incremental as budget increases.
The first £20,000 of spend may reach highly responsive new audiences.
The next £50,000 may increasingly reach players who:
already know the brand;
would have converted organically;
have seen the campaign repeatedly.
Incrementality testing can therefore become particularly valuable when operators consider scaling successful campaigns.
The commercial question becomes:
Does the next increment of spend still create enough additional value?
That is more useful than assuming historical average CPA will remain constant indefinitely.
Build one source of truth for experiment reporting
Incrementality analysis becomes difficult when campaign and player data sit across disconnected systems.
A useful workflow may bring together:
media spend;
delivery data;
clicks;
campaign identifiers;
registrations;
KYC outcomes;
deposits;
bonus cost;
revenue;
cohort information.
The reporting layer should use consistent definitions across test and control groups.
Otherwise, measurement noise can become larger than the campaign effect being tested.
Use automation to support incrementality operations
Incrementality testing contains repetitive operational work that can be automated.
Potential applications include:
refreshing test-versus-control reporting;
monitoring group sizes;
checking campaign exposure;
flagging anomalies;
updating mature cohort performance;
reconciling media spend;
checking whether control groups are receiving unexpected activity.
AI-assisted reporting can also help analysts identify unusual changes worth investigating.
However, automation cannot compensate for weak experiment design.
A perfectly automated report based on a contaminated control group will still produce unreliable evidence.
Experimental logic must come first.
Common errors that inflate perceived campaign lift
Several problems repeatedly make campaigns appear more incremental than they really are.
Using before-and-after comparisons without a control
Demand may have increased naturally.
Without a control group, the campaign receives credit for the entire movement.
Changing several things at once
Suppose a sportsbook simultaneously launches:
new creative;
a larger welcome offer;
increased bids;
a fixture-specific landing page.
Performance improves.
The operator cannot confidently isolate which change caused the lift.
Contaminating the control group
Withholding users from one channel while exposing them to an equivalent message elsewhere reduces the difference between test and control.
Ending the experiment early
A strong weekend does not necessarily prove a campaign works across normal trading conditions.
Using the wrong conversion event
Incremental registrations may look strong while incremental verified depositors remain weak.
Measurement should match the commercial objective.
Using a window that is too short
If player value develops over several weeks, a short test may understate or distort the result.
Overinterpreting small cohorts
A few unusually valuable customers can produce misleading short-term revenue conclusions.
Create clear test governance
The strongest defence against these problems is disciplined experimentation.
Before launch, document:
primary hypothesis;
eligible audience;
test and control methodology;
exposure rules;
primary success metric;
secondary metrics;
test duration;
commercial threshold;
decision date.
Afterwards, record:
result;
limitations;
market context;
operational issues;
decision taken.
That turns individual experiments into organisational knowledge.
Future teams can then refer to previous tests instead of repeatedly asking the same measurement questions.
How Cognaix approaches betting campaign incrementality
Cognaix approaches incrementality as part of practical iGaming execution rather than a standalone analytics exercise.
The objective is to connect campaign measurement with the economics and operating realities of betting acquisition.
Paid media
Paid search, paid social, display and video can all produce strong attributed conversion figures.
Incrementality testing helps determine which activity is actually expanding the player base rather than simply receiving credit for existing demand.
That evidence can support budget increases, reductions and channel allocation.
Affiliate marketing
Affiliate partnerships should be assessed not only on attributed FTDs but also on whether they introduce genuinely new players and how those players perform over time.
This is particularly relevant where several partners or channels appear in the same acquisition journey.
CRM
Audience holdouts can help operators determine whether onboarding, reactivation and promotional activity genuinely changes behaviour.
That provides a stronger commercial measure than attributing every subsequent deposit to the most recent communication.
Data and attribution
Attribution remains useful for operational reporting.
Incrementality complements it by testing causality.
Combining both views gives operators a stronger basis for decision-making than relying on either in isolation.
Reporting
Incrementality reporting should bring campaign investment, exposure, player conversion and downstream value into one understandable view.
The goal is not to create a mathematically impressive report.
It is to tell decision-makers whether the activity generated additional value and whether the evidence is strong enough to act on.
Automation
Automated data pipelines, experiment monitoring and cohort refreshes can reduce the manual work surrounding testing.
Specialists can then spend more time evaluating why behaviour changed and how that evidence should influence the next campaign.
Player quality
The ultimate goal is not incremental registrations.
It is incremental commercial value from players acquired at sustainable economics while operating within relevant regulatory and player-protection requirements.
That means testing should continue beyond the first conversion whenever the scale and maturity of the data allow.
Final thoughts
Betting campaign incrementality changes the question operators ask of marketing.
Traditional attribution asks:
Which channel should receive credit?
Incrementality asks:
What additional behaviour did the channel actually create?
That distinction can fundamentally change how acquisition performance is interpreted.
A channel reporting 1,000 FTDs may only have generated 200 genuinely additional depositors.
Another campaign with fewer attributed conversions may produce a much greater incremental effect.
Neither attribution nor incrementality needs to replace the other.
Attribution remains useful for operational campaign management, tracking and partner reconciliation.
Incrementality provides stronger evidence for strategic budget decisions.
The best betting measurement frameworks use both.
Start with a clear decision.
Define the hypothesis before launching the campaign.
Choose a credible control.
Measure the difference between observed behaviour and the estimated counterfactual.
Then follow those incremental players far enough into the lifecycle to understand whether the additional behaviour was commercially valuable.
The goal is not to prove that every campaign works.
It is to identify where marketing genuinely changes player behaviour profitably.
Once operators begin treating incrementality as an ongoing acquisition discipline, budget conversations become less about who received credit and more about where the next pound of spend is most likely to create additional growth.
Frequently asked questions
What is betting campaign incrementality?
Betting campaign incrementality is the additional player behaviour caused by a marketing campaign above what would probably have happened without that activity.
It can be measured in incremental registrations, verified accounts, FTDs, deposits, revenue or profit.
The key is estimating a credible counterfactual through a control group or similar experimental design.
What is the difference between attribution and incrementality in betting?
Attribution assigns conversion credit to marketing touchpoints.
Incrementality attempts to measure whether the marketing activity actually caused additional behaviour.
A paid search campaign may receive credit for 1,000 FTDs while an incrementality test suggests that many of those players would have registered anyway.
Both views can be useful, but they answer different questions.
How do you calculate incremental FTDs?
At a simplified level, incremental FTDs are the observed FTDs generated in the test population minus the number expected to have occurred without the marketing intervention.
That baseline is estimated using a control group, matched geography or other credible counterfactual.
For example, if the observed outcome is equivalent to 1,000 FTDs and the estimated baseline is 800, the test suggests around 200 incremental FTDs.
What is cost per incremental FTD?
Cost per incremental FTD divides the relevant campaign investment by the estimated number of additional first-time depositors caused by that campaign.
For example, £40,000 of spend generating an estimated 250 incremental FTDs would result in a £160 cost per incremental FTD.
This can differ substantially from attributed FTD CPA.
Are geo experiments suitable for sportsbook advertising?
Geo experiments can be useful for broad acquisition channels when comparable regions can be identified and campaign exposure can be controlled.
Operators should account for historical FTD volume, sporting demand, channel mix and other factors that could cause regions to behave differently.
A simple before-and-after comparison is generally weaker because external market movements can be mistaken for campaign lift.
How do CRM holdout groups measure incrementality?
A proportion of eligible players is randomly withheld from a CRM campaign while the remaining group receives the normal communication.
Deposit behaviour, revenue or another outcome is then compared between the groups.
The difference provides an estimate of the additional behaviour caused by the CRM programme.
Control users should remain genuinely unexposed to equivalent campaign activity where practical.
Are platform conversion-lift studies reliable?
Platform lift studies can provide valuable incrementality evidence, particularly when the platform has access to exposure data that the operator cannot observe directly.
Operators should still review methodology, sample size, conversion definitions and attribution windows.
Platform results are strongest when they are connected with first-party player-quality data rather than used in isolation.
How long should a betting incrementality test run?
There is no universal test length.
The appropriate period depends on conversion volume, campaign type, sporting calendar and the metric being measured.
The test should generally run long enough to generate a credible sample and avoid relying on an unusual short period such as one major fixture weekend.
Longer-term player-quality outcomes can continue to be evaluated after campaign exposure ends.
Should betting incrementality be measured using revenue or conversions?
Both can be useful.
Incremental registrations and FTDs provide earlier signals, while revenue and profit give a stronger commercial view once cohorts mature.
Operators can therefore assess incrementality in stages: first confirming additional acquisition, then determining whether those additional players generate sufficient downstream value to justify the cost.