Nordic Naturals and live sports: avoiding double credit in full-funnel budgets

01 Facts & context

The published goals, partners and results

Amazon Ads describes how Nordic Naturals worked with Gigi and Code3 while expanding its vitamins and supplements assortment. The project sought 10% year-over-year Amazon revenue growth, alongside awareness, consideration and acquisition objectives. It also planned 30% more budget while seeking a lower cost per thousand impressions. Those targets belong to this particular project; they are not standard targets for brands using Amazon Marketing Cloud.

The reported approach used AMC observations to inform inventory, audience, frequency, attribution and Performance+ campaign choices. The public case reports 33% year-over-year Amazon revenue growth, a branded-search rate 3.9 times its display benchmark, 16% lower CPM, 21% higher new-to-brand detail-page-view rate and 87% higher new-to-brand ROAS. Its footnotes identify Nordic Naturals data from the United States during 2025–2026. These are source-reported results, with definitions and comparisons that should be read in their stated scope.

The useful analytical question is how those decisions fit together, and which conclusions the available evidence supports. More spend, new product categories, different audiences and revised reporting windows changed together. Some audience overlays were tested, but that does not establish a complete counterfactual for the whole program. GSP uses the case to examine the relationship between demand creation, demand capture and measurement, rather than attributing every revenue change to live sports or presenting the outcome as a forecast for another brand.

02 GSP analysis

Brand search is both an outcome and an input to later advertising

Nordic Naturals connects live sports, branded search, product-page visits and new-customer purchases. Meridian’s full-funnel methodology explains why such a chain can double-count credit. Brand advertising may create search interest that search ads later capture. Last-touch evaluation can undervalue the earlier exposure; treating all branded search as a fixed control can instead remove part of the upstream effect from a model. This is a methodological issue, not something solved completely by extending an attribution window.

GSP recommends mapping the brand’s demand path before deciding which metrics are observations and which can guide budget allocation. Sports exposure can be assessed through reach, memory and subsequent active search; conversion touchpoints through friction and completed orders. Two-stage modeling can examine the indirect path, but needs more complete data and control of confounding. When these conditions are absent, retain a simpler model and add focused experiments. A more elaborate formula should not conceal weaker inputs or transform a plausible mechanism into a demonstrated outcome.

03 GSP analysis

A longer attribution window does not establish greater demand

The case reports very different attributed product-page activity under 60-day and 14-day windows. This is consistent with a longer purchase journey, but extending a window does not create additional visits. IAB’s incrementality principles add a necessary constraint: the assessment needs a credible counterfactual. A longer window can include other campaigns, underlying demand or returning customers in the same attributed path. It broadens observation rather than automatically strengthening causal evidence.

Reports and experiments should therefore have different jobs. Reports reveal behavioral sequences after sports exposure; experiments compare total outcomes with and without the intervention. Higher annual spending, category expansion and creative changes belong in the explanation alongside ROAS. Before allocating budgets, define whether the objective is new audiences, repeat purchase or contribution efficiency, then choose appropriate controls. Consistent customer definitions, channel coverage and conversion delays allow evidence to accumulate across touchpoints instead of allowing every department to claim the same order independently.

04 GSP analysis

Test audience restrictions rather than assuming more is better

The case describes AMC overlap analysis and tests of audience overlays on streaming television. The team selected relevant live-sports inventory using first-party insights and Amazon signals, then removed audience overlays from that inventory. The source reports a 30% reduction in cost per branded search after the removal. This provides a concrete hypothesis: in some contexts, additional targeting restrictions reduce useful reach without improving the net outcome enough to justify the restriction.

That hypothesis does not mean context always outperforms audience selection. Content, viewers, creative, pricing and available inventory all matter. The clean-room matching and programmatic guaranteed arrangement were this project’s implementation choices, not prerequisites for every contextual test. A smaller advertiser can choose a design suited to its information and buying options. The source’s reported comparisons of engagement and purchase rates remain project evidence, not a promise that removing targeting works in every category.

For a hypothetical supplements brand, the first step is to define the question and select an appropriate comparison. If two content environments are compared, keep creative, budget, period and eligibility as comparable as possible while recording differences in the audiences they naturally attract. New-customer page visits can be informative, but one comparison cannot prove that the underlying bottleneck is data rather than inventory. Turn a reproducible observation into the next testable hypothesis, instead of treating a favorable result as a complete explanation of how targeting should work.

05 GSP analysis

Turn attribution and frequency observations into testable settings

The source reports that attributed detail-page-view rate under a sixty-day window was twelve times the rate under a fourteen-day window. It also associates at least five weekly impressions with a substantially higher page-view rate. These observations raise questions about purchase delay and exposure. Changing the attribution window changes which behaviors are counted; it does not create twelve times as many actual visits. Frequency associations may also reflect differences in interest and opportunities to receive advertising.

The operating chain is useful: observe behavior in AMC, propose a parameter, execute it through media tools and review subsequent outcomes. Each change should have a stated reason and a way to check the result. But changing the measuring rule is different from improving the business. Keep a continuous report under fixed definitions while testing whether a new exposure policy improves reach, customer quality and contribution. This avoids mistaking more credited activity for more activity actually occurring.

A longer window requires waiting for later behavior, so reports should distinguish mature results from values that may still update. A frequency experiment should compare allocation policies, including cost, repeated reach and customer response. If costs and a new-customer metric both rise after two weeks, the system is not necessarily correcting an earlier bias; audience composition or promotions could have changed. Set review timing using actual purchase delay and sample availability, and retain the original reporting basis before deciding whether to expand the new setting.

06 GSP analysis

Different objectives for acquisition, consideration and retention

The case used Performance+ for consideration, remarketing and retention. It reports changing a consideration campaign from a ROAS goal to a detail-page-view-rate goal after concerns about new-customer quality. The lesson to test is that the chosen objective influences optimization behavior. It does not establish that every ROAS goal only captures existing demand, or that a browsing goal necessarily creates valuable new customers. Connect each early objective to later orders, returns and the continuation of the relationship.

The audience design linked streaming exposure, search, repeated page visits, cart activity and lapsed purchases. A hypothetical pet-supplements brand could distinguish awareness, consideration and retention tasks, but should define inactivity using the category’s replenishment cycle rather than borrowing a fixed ninety-day threshold. Each group can have a stage measure while remaining subject to shared contribution and experience criteria. Different goals do not automatically prevent budget competition, and several groups must not independently claim the same order as additional value.

The source also describes Spanish-language creative and store remarketing, reporting 6.6 million additional impressions, a $2.79 CPM, a 39% new-to-brand rate and $62,000 in additional new-to-brand sales. The public account does not disclose enough to independently audit that incrementality design. This is a reason to investigate unmet language and use-case needs, not to assume that every new language segment is cheaper. Product explanations, purchase pages and service support should be able to fulfill the promise made in the advertising.

07 GSP analysis

Coordinate brand, technology and agency responsibilities

Nordic Naturals, Gigi and Code3 connected brand objectives, AMC observations, media settings and creative execution. This illustrates a handoff from analysis to action, but the public case does not prove that every company needs the same three-party structure. GSP focuses on whether each handoff has a clear input and output: what business problem is being addressed, what the evidence supports, who can change a parameter and who checks the customer and commercial result.

The brand should retain authority over growth objectives, budget limits and measurement definitions while partners contribute analysis and execution. The original revenue and efficiency targets are project choices, not performance requirements for another brand. An attribution window changes the scope of an advertising report; it does not directly change financial revenue recognition. When reconciling the two, record customer definitions, time windows, returns and channel coverage rather than treating them as interchangeable accounts of the business.

Assign a result owner for each category and connect analysis, media, product and service responsibilities. Evaluate partners on evidence quality, traceable changes and outcomes as well as execution, rather than rewarding only adjustment speed or the number of experiments. A decision record can contain the observation, proposed mechanism, owner and review date. If results differ from expectations, the team can then identify whether the hypothesis, data, implementation or wider conditions need revision. A small team can maintain this responsibility chain without reproducing a large-agency organizational chart.

08 GSP analysis

Account for budget changes and alternative explanations

The case reports 30% more budget, 16% lower CPM and 33% revenue growth. Those figures cannot simply be joined into a causal claim that media efficiency produced all the growth. New product categories, promotions, inventory mix and media prices could contribute. The public account does not fully separate new-category and existing-product effects or provide a complete program-level counterfactual. Changes from fourteen-day to sixty-day attribution also require a fixed-basis comparison before reports can be interpreted consistently.

The footnotes limit the data to one brand, the United States and 2025–2026. A platform-published success case can help identify mechanisms and candidate tests, but it does not provide the distribution of outcomes for similar projects. This does not remove its value. It means distinguishing the reported observation, the conditions that might transfer and the outcome that remains to be tested locally. Evidence from a selected case cannot establish a typical expected return or the likelihood of success elsewhere.

Start local review with two separate questions. First, how many total new customers and how much contribution did additional spend create relative to a credible comparison? Second, does the delayed behavior captured by a longer window fit the observed purchase path and its other influences? The second question improves interpretation; it does not replace the first. If the only demonstrated change is broader attribution coverage, state that conclusion rather than claiming the window created revenue. Individual parameter changes can be tested separately without copying the entire tool combination.

09 Suggested actions

Set definitions before inventory and audience decisions

GSP recommends agreeing on the principal measures and their purposes before changing inventory or audiences. Define the denominator, window and coverage for branded search, page visits, new customers, ROAS and CPM, alongside budget and contribution objectives. Growth, finance and media can review a concise definition record and keep versions when it changes. Shared definitions make it easier to distinguish the effect of a campaign action from a change in reporting and reduce duplicate claims for the same result.

Next, test audience overlays where suitable. Compare keeping or removing a restriction under comparable conditions, observing reach, cost, branded search and final outcomes. If removal produces a worse cost, investigate content fit, sample, media price and audience differences rather than immediately concluding that the category does not fit sports. Check the design and implementation, then choose whether to revise creative, audiences or inventory. Recording alternatives alongside the initial hypothesis is more useful than producing only a list of restrictions that appeared to work.

Then evaluate new language, need-state or lapsed-customer groups using the evidence. The sequence can change with the company’s bottleneck; it is not mandatory to implement every module. Give each group a service promise, stage objective and later outcome measure. New-language expansion should include page, support and returns information. Retain comparisons and stable definitions while expanding so that changes in spend, audience and observation do not become inseparable. The decision should identify which action merits continuation, not simply celebrate a larger campaign footprint.

10 Suggested actions

Leading signals, later outcomes and recurring reviews

Leading observations can include branded-search cost, detail-page activity, frequency distribution and CPM, with reporting scheduled around actual purchase delay. Frequency and window choices should be explanatory labels rather than performance goals. The source case’s five impressions and sixty days are not universal targets. Regularly align parameter changes with results while retaining an unchanged measurement series. That helps distinguish an implementation problem, more reported behavior and more behavior actually occurring in the market.

Later outcomes include new-customer orders, net contribution, repeat purchase, returns and the acquisition cost associated with subscription programs. Subscribe & Save is the name of an Amazon program; a lower cost per enrollment should not be translated as a financial saving the company has already realized. Review timing should reflect the sample and purchase cycle. A fixed four-week rule may be too short or unnecessarily long. When evidence is insufficient, state that the conclusion is unresolved rather than treating absence of a clear result as proof of failure.

Useful comparisons include overlays retained versus removed, a fixed set of events recomputed under short and long windows, alternative objectives and language-specific creative with store remarketing. The window comparison is descriptive unless paired with a design that changes actual exposure. Predefine criteria and stopping rules, and inspect product suitability, price, service and promotional changes when results deteriorate. A single disappointing metric does not establish poor audience quality. Several consistent observations and an adequate comparison provide a stronger basis for changing the budget.

11 Suggested actions

Conditions for application and revision

The analysis is most useful where purchase delay can be observed, media settings can be compared and enough data exists for the decision. First-party information and clean rooms can support particular designs, but they are not necessary for every contextual test. Programmatic guaranteed buying is also one inventory choice rather than a universal requirement. Management needs to understand result maturity and fixed definitions, not simply agree to a longer attribution window. Smaller teams may answer their immediate question with a lighter audience, creative or geographic comparison.

Adjustment signals should trigger investigation rather than immediate labels. An unchanged search cost after removing an overlay calls for review of context, creative, price and sample. More new customers with lower repeat purchase calls for examination of buying cycles, product fit and audience mix. A large difference between long and short windows calls for examination of delay and other activities. None of those observations alone proves that the audience is low quality or the context is invalid.

The central capability is translating evidence into traceable decisions and tests, then revising them using customer outcomes. Tool access supplies more observations; it does not guarantee a sound explanation or a repeat of the source’s growth figures. At each review, ask what was observed, what else could explain it and how the next comparison can distinguish those explanations. Accumulating that evidence can make allocation less dependent on a single report or success story, while keeping the commercial decision connected to what the available information can actually support.