Sponsorship research and audience evidence

Start with the sponsorship right you are assessing, then find evidence about the people who could encounter it. Check what each figure counts, how it was produced and whether it covers the proposed activity. A property-wide audience figure provides context; it does not show who saw a particular sponsor asset.

Define the evidence question

Name the right: an on-site activity, an organiser post, a programme credit or another specified benefit. Describe who would have to attend, receive or view something to encounter it. A workshop audience may be more relevant to a workshop sponsor than attendance at the whole event.

Write down the decision the evidence must support. This keeps the research focused on a useful audience question rather than the largest number in a proposal.

Sponsorship Right Relevance: Audience Figures Compared

Keep a trail for each claim

For each material figure, retain the original report or extract. Record its owner, date, reporting period, population, unit and method. Mark it as an observed count, survey estimate or forecast. Where sources are combined, ask how overlap was handled.

Do not add these figures into one unique reach total. Their units and periods may differ, and the same person may appear in several. For a digital ‘unique audience’, check whether the unit is people, devices or another identifier and how duplication was addressed.

Test whether the method supports the claim

A ticketing record can answer a ticketing question. It may not count everyone present or anyone who noticed a sponsor. A voluntary survey describes its respondents; extending its answers to all attendees needs a sampling method that supports that inference. Check eligibility, recruitment, completed responses and groups that may have been missed.

Use evidence from comparable activity where possible. Last year’s whole festival, another venue and an unusually prominent post can provide context, but none is a direct forecast for a proposed placement. Note changes in programme, channel and measurement method beside any comparison.

Use a fit-for-purpose quality check

The Australian Bureau of Statistics (ABS) Data Quality Framework assesses data quality in relation to a user’s purpose. It identifies seven dimensions: institutional environment, relevance, timeliness, accuracy, coherence, interpretability and accessibility. A source need not score equally on every dimension for every decision; judge which matter most to the sponsorship right being considered.

For example, relevance matters when the dataset’s population or measures may not match the proposed audience. Coherence deserves more attention when comparing figures across periods or sources. If credibility and trustworthiness are central, scrutinise the institutional environment of the data.

A concise quality statement can make this assessment usable. The ABS recommends communicating key characteristics that affect quality, including both strengths and limitations, so readers can decide whether the data are fit for their purpose.

ABS Data Quality Framework: Seven Dimensions for Sponsorship Evidence

Separate sampling error from other error

A survey estimate based on a sample can differ from the value that a complete count of the population would produce. The ABS calls this sampling error.

Sampling error can arise when a sample is too small, differs in its characteristics from the population, or is selected using a method that is not random. In a random sample, where each unit has a calculable chance of selection, sampling error can be measured and controlled; increasing sample size generally reduces it.

A larger sample does not resolve every weakness. Non-sampling error can arise at any stage and includes coverage error, when people are wrongly included, excluded or duplicated; non-response error, when selected people provide no or only partial information; and response error, when answers are inaccurate.

For a sponsorship audience claim, ask whether the people who could encounter the specified right were covered and whether missing responses might affect the result. Record these limitations separately from the estimate itself: they affect how confidently the result represents the intended population.

Sampling Error vs. Non-Sampling Error in Sponsorship Research

Check the digital measurement context

Digital audience figures can depend on the measurement method as well as the metric. In April 2021, IAB Australia said it could not support tagged monthly audience data in Nielsen’s soft-launched Digital Media Ratings while methodology concerns and data queries remained unresolved. Its review raised concerns about variation in unique-audience results after a methodology change.

The IAB distinguished that audience measurement from volumetric data measured through media-owner tagging, such as page views, time spent and sessions. It continued to support market use of that overnight traffic data, while noting that the methodology for untagged sites had not changed.

Use this example as a prompt to check the specific product, method and period behind a digital claim, rather than treating all metrics from a platform or publisher as interchangeable. A change in method can affect comparability even when a metric has the same name.

Digital Audience Measurement Methods: IAB Australia Guidance vs. Nielsen DMR

Make quality limits actionable

State what the evidence supports and what remains outside its reach. A dataset may be accurate for its recorded population yet not relevant to a particular right; a credible source may still have limited coverage or timeliness. The ABS framework treats quality as fitness for purpose, so the same dataset can warrant different judgements for different decisions.

Keep the limitation attached to the claim when it is shared. If a profile omits some audience members, describe the measured group rather than implying the attributes apply to everyone. If methods or definitions differ between periods, identify the break before presenting a comparison.

Record the decision and its limits

Ask what the property can report for the specific right before buying it. After delivery, keep three questions separate: did the asset run, who might have encountered it, and what response was measured? If the reporting method changed, state that before comparing periods.

A brief evidence note can label each material claim supported, supported with narrower wording or unknown, then say how the gap affects the decision. Verified event attendance, for example, may support the event’s scale while leaving a workshop audience unknown. The buyer can request workshop evidence, narrow the right or proceed with that uncertainty recorded.