Media audience research

Media audience research answers four questions: who is eligible to buy, what media they use, what evidence supports that finding, and whether a specific placement can reach them. Define the intended audience first, then match each decision to evidence that measures the right people, place and unit.

A media-use result is a starting point, not a buyable audience. Check the placement’s inventory, geographic coverage, targeting definition and forecast basis before treating it as an opportunity.

Define the audience and the decision

Describe who is eligible for the offer, where the business serves them and any limits on purchase. Separate what customer and enquiry records show from assumptions about people the business has yet to reach. If a brief says “regional homeowners”, specify the regions and why a household would qualify.

Write eligibility as observable rules, and state whether the audience is defined as people, households or another unit. Use the same definition when comparing population counts, media-use findings and placement forecasts.

State what the assessment must decide. Is the question whether a medium deserves investigation, whether an audience is large enough to plan against, or whether a particular placement covers the service area? Each needs different evidence.

Give each source a job

Choose evidence for the question it can answer: customer records describe past customers and enquiries, population data sizes a defined area, surveys describe reported activity, and planning data estimates a specified placement. Record the provider or study name, date, population, geography and audience unit so unlike evidence is not treated as interchangeable.

Roy Morgan Single Source provides a named example of Australian media-use evidence. Its newspaper and magazine results report cross-platform audience estimates, while its published summary describes a population aged 14 and over—not buyers who meet a particular offer’s eligibility rules.

For a survey, capture the exact question, fieldwork period, sample size, base and any reported sampling error or confidence interval. Check how the sample was selected and whether the surveyed population matches the intended buyers. If key details or uncertainty are not reported, treat the finding as directional rather than as a precise local audience count.

Roy Morgan’s results for the six months to June 2020 draw on personal interviews with a representative sample of 28,112 Australians aged 14 and over in their homes, alongside machine-based data. The published summary does not state a sampling error or confidence interval, so do not imply finer precision than the reported estimates.

ABS QuickStats can describe a selected area's Census characteristics. In the 2021 QuickStats, person figures use place of usual residence; family and dwelling figures use place of enumeration. Check the Census year and geography. A Census count is not a current advertising audience.

For any survey figure, keep the survey question and fieldwork period with it. A general media-use finding can support a channel hypothesis, but it does not count eligible local buyers or measure delivery of an advertiser's ad.

Read area data carefully

QuickStats can help turn a broad audience description into a checkable local profile. Its Persons tables include characteristics such as age, education, language, ancestry, health and employment; Families tables include family composition and income; and Dwellings tables include dwelling type, bedrooms, household income and vehicles. These categories describe different units, so choose the one that matches the audience definition rather than treating household characteristics as person counts.

To find an area profile, use the ABS Search Census data tool and select the Census year and area of interest. All persons QuickStats provide data from 2001 onwards for geographies down to Statistical Area Level 1, and compare the selected area with its state or territory and Australia. The comparison can give context, but it does not establish how many residents meet the offer’s eligibility rules.

Map each eligibility rule to a matching Census characteristic and unit. Do not combine separate counts to claim the number who meet several conditions unless the data directly supports that intersection.

QuickStats applies small random adjustments to cell values to protect respondent confidentiality. The ABS notes that these adjustments can make row or column sums differ slightly from table totals. Treat small discrepancies as a data limitation, not as evidence that the audience definition or total is exact.

The ABS points to the 2021 Census Privacy Statement for further detail on Census confidentiality. That statement concerns Census data; it does not establish rules for handling a buyer’s customer or enquiry records.

Test the buyable opportunity

If the audience uses a medium, find out what inventory is offered, where it runs and whether its targeting matches the intended market. Check whether a figure counts people, accounts, devices, households, visits or impressions. Ask whether it describes content users, people addressable by advertising, or the expected reach of the proposed plan.

Roy Morgan reported that 18.5 million Australians aged 14 and over (88%) read or accessed newspaper content in some way in an average four weeks; 16.1 million (77%) accessed metropolitan titles. The same results put the Sydney Morning Herald at almost 8.3 million in an average four weeks. These are historical cross-platform audience figures for the six months to June 2020, not a placement forecast.

For a proposed Sydney Morning Herald placement intended to reach regional homeowners, the almost 8.3 million figure does not establish how many regional homeowners qualify or whether the placement covers the service area. Ask the seller to specify the placement, its regions, inventory, audience unit, targeting definition and forecast basis. Until those details can be checked against the eligibility rules, classify the fit as unverified.

A media-use finding is a lead, not proof of a buyable audience. For “regional homeowners”, reported use of a medium only supports a hypothesis; verify the specific placement’s regions, inventory, targeting definition and forecast basis before calling it an opportunity.

Treat supplier forecasts as estimates for the defined plan, not as counts of eligible buyers. Check what inputs and definitions the supplier provides, and record any limits before using a forecast. If the audience unit, geographic coverage or counting method cannot be checked, keep the placement fit unverified.

For each channel, record the audience hypothesis, supporting evidence, available placement and main uncertainty. If the evidence establishes only general media use, describe the placement's audience fit as unverified.

Treat automated audience signals cautiously

Automated systems are part of news feeds, search and programmatic advertising. The ADM+S Centre says these systems can have positive as well as problematic results, and identifies risks to democratic processes and social cohesion from automated curation and algorithmic amplification, including misinformation and other social harms linked to selection principles it describes as “transparent and inequitable”.

For audience assessment, ask what a platform signal counts and how its selection criteria relate to the buyer’s eligibility rules. Do not mistake a platform’s audience signal for independent evidence that the selected people meet the business’s buyer definition.

Make the research usable

Compare figures only after checking their populations, periods and definitions. Show unlike figures separately; do not average them. For a reach estimate, record the audience base and delivery assumptions. For a local buy, check the placement's geographic coverage against the service area.

Separate what is measured, what is inferred and what remains unverified. A media-use finding supports a placement decision only when the proposed inventory and its audience definition can be checked against the intended market.

Revisit the assessment when the offer, service area, source definition or inventory changes. A short record of what is known, inferred and still to be checked gives the buyer a decision they can explain.