Your retail partner reports a successful store campaign. Your China-based production team is ready to make the next assets. Should the brand extend the campaign, add screen advertising to the existing materials, or replace commercial A with commercial B?
Those are three different decisions. ONCE recommends commissioning the comparison that matches the change you might actually buy. More versions will not resolve a question the measurement arrangement cannot distinguish.
Keep the reported result attached to its treatment
On October 6, 2026, Love’s Media Group reported a 23% incremental unit lift for a beverage promotion. The campaign ran February 26–April 22, combining in-store and pump screens with printed materials. MarketDial compared campaign stores with matched holdouts. This was a company-reported result for a combined campaign, not a video-specific finding or an October 16 launch.
We have not reviewed raw store data, complete implementation records or all analytical settings. Do not invent whether the comparison stores had the same promotion. The release identifies digital screens, not the exact video content played on each one.
A useful campaign-level result does not assign credit to its components. Splitting the reported lift according to each asset’s production cost would not solve that problem. Nor would choosing the most visually prominent asset and calling it the growth driver.
ONCE’s Three Questions, Three Comparisons
This is a proposed commissioning aid, not a new statistical estimator or a reconstruction of the Love’s test. The rows are alternative questions. They are not instructions to run a large multi-arm experiment regardless of budget or available data.
| Decision to make | Comparison to plan | What it does not establish |
|---|---|---|
| Continue the complete campaign | The specified package versus a clearly described business-as-usual condition, recording offer and material differences | Which screen, poster or individual video produced the result |
| Add screen advertising to existing materials | Comparable product, promotion and other materials, with versus without the specified screen advertising | A universal standalone screen effect, or creative B’s advantage over A |
| Replace creative A with B | Comparable screen placements, schedules, delivery and other campaign conditions, with A versus B as the main planned change | Video versus no video, or the whole package’s total incremental effect |
Adding screens here means adding screen advertising, not necessarily installing hardware. Its effect can depend on the materials already present. An A/B creative result similarly belongs to those versions and the conditions in which they ran.
If A gets more plays in busier stores while B receives less delivery elsewhere, the comparison is no longer just about creative. Version files must connect to actual execution and the relevant sales data. Where those conditions cannot be established, narrow the claim rather than explaining the difference as a storytelling success.
Establish a feasible comparison before producing the variants
MarketDial’s store-selection guidance discusses relevant historical matching, pre-period checks and sample requirements that depend on effect size, variability and statistical design. Another campaign’s store count is not a ready-made requirement for your brand.
Ask the media and measurement leads what the available stores and data can support. Agree on assignment, the intended change, what should remain comparable, and how stockouts, promotions and delivery deviations will be recorded. Define when a result can guide the decision and when it remains inconclusive.
A rotation design also needs assessment for timing changes, repeated exposure and carryover. Simply alternating two files is not automatically a fair A/B test. An unmatched store that happened not to run the ad is not automatically a useful control either.
If the design cannot support a reliable distinction, an exploratory result may still inform the next question. It should not be the sole basis for declaring a winner or expanding the purchase. Failure to distinguish A and B does not prove that they are equally effective.
A hypothetical renewal decision
Imagine a beverage brand sourcing from China receives a positive package-level report. The media seller proposes additional screen advertising; the production partner proposes fresh films. If the unresolved purchase is the screen add-on, while the other materials are already settled, the second comparison is the relevant starting point. This is a fictional commissioning situation, not an ONCE client test.
The design would need comparable products, promotions, printed materials and timing, with the specified screen advertising as the planned addition. If the retailer can only deliver a combined package and cannot establish those conditions, the object being evaluated remains the package. Renaming its result “screen contribution” would not fix the limitation. The brand can accept a broader answer, change the execution plan or defer the add-on.
If the screen allocation is already fixed and only the replacement-production fee remains undecided, the third comparison becomes more useful. Versions must be distinguishable in actual delivery and measurement. Alternating two files on one store screen while receiving only total campaign sales does not automatically create a version-level result. Establish which comparison is feasible before ordering ten films.
Connect the production order to the measurement job
The production partner can supply precisely identified versions, their deliberate differences and the applicable product information. The media partner supplies the available placement plan and execution records. The measurement lead owns the design, data requirements and analytical limitations. The brand specifies which future purchase the finding will change.
Keep creative production, execution reconciliation and measurement scopes visible in the proposal. A campaign-level evaluation does not automatically require a large variant library. Conversely, buying two finished films does not buy the conditions needed to compare them. A producer should not guarantee store sales it cannot control.
For ONCE, the potential long-term asset is not an unqualified ranking of “winning videos.” It is the record of what changed, what remained comparable, what deviated in execution and what decision followed. Such records could support continuing creative services, but recurring revenue and margin improvement still require actual client work and validation.
Before the next shoot, write the purchase question into the brief: renew the package, add screen advertising, or replace A with B? Bring that brief to ONCE’s video production solutions discussion. Agree on the question before multiplying the files.
Sources checked October 16, 2026. Love’s is a first-party campaign announcement; MarketDial’s undated guidance is methodology background, not independent validation of this campaign. We also reviewed Take 10 Media’s October 9 discussion of measurement without device IDs, which explicitly treats the Love’s finding as a whole-mix result. The commissioning framework and beverage brief are ONCE’s analysis; no client results or testing outcomes are claimed here.
