Methodology
How the figures are produced
Most measurement covers only part of the audience. Eyeballr estimates audiences across formats on one method, and says which values are observed, which are modelled, and which are calculated.
This page is an overview. A fuller methodology document is available to clients and prospective clients on request.
The audience estimate
Audience is estimated at fixture and market level. Every property runs through the same four stages, in the same order.
- Population. The people in a market who could plausibly watch, shaped by that market's history with the property.
- Access. How many of them could reach it, worked out per broadcaster and per fixture from confirmed rights and schedule information. This stage explains most of the difference between a large audience and a small one.
- Conditions. What made this fixture different from the same fixture in another week.
- Split. The market total is divided across the broadcasters showing it, and between licensed and unofficial viewing.
The five inputs shown on the product page (access, demand, context, unofficial viewing, and calibration) feed these stages.
What the model considers
- Local time of day, in the viewer's own time zone
- Day of the week
- Point in the season
- Competition stage
- Table position and the importance of the fixture
- Star players, who lift their home market while they are at the club
- Each market's history with the property
- Rivalries
- Broadcaster type and reach, per market and per fixture
- Clashing events that draw on the same audience
Player and team effects are dated, so a mid-season transfer moves a market from the day it takes effect. A new sport changes the unit of calculation, how importance is judged, and whose following matters. The stages stay the same.
Licensed and unofficial viewing
Audience figures are split into licensed and unofficial viewing. Unofficial viewing is estimated rather than measured: it is modelled at league and market level, anchored to validated benchmarks, and is never a count of streams.
Eyeballr treats unofficial viewing as a signal of demand. It does not host, mirror, index, or link to infringing content. Measurement is aggregate and at market level, and no individual viewer is identified.
Where the data comes from
- Built by Eyeballr. Reference data that Eyeballr compiles and maintains, such as which broadcaster shows which fixture in which market, and how strongly each market follows each team and player.
- Licensed and public. Population and market statistics, schedules and rights announcements, competition data, and aggregate signals of interest such as search and coverage volume. Signals of interest shape relative weightings. They are never converted into a viewer count.
- Supplied by a client. Rates, asset lists, or existing measurement. It stays the client's, is held separately, and is not used for any other client.
Eyeballr does not use panels, metered devices, device identifiers, cookies, or any individual-level data. No personal data is processed to produce an audience figure.
Sponsor exposure
Eyeballr records when, where, and for how long each sponsor appears on screen, then ties each appearance to the audience watching that fixture in each market.
- Read. The broadcast is sampled to still frames at a documented rate, and vision models propose where brands appear.
- Name. Each proposal is matched to one brand register, whatever script the mark is written in. A mark that cannot be positively identified is held, not assigned to the likeliest candidate.
- Place. Each appearance is put in a zone, a sub-zone, and a named asset, the level at which sponsorship contracts are written.
- Review. A person approves, rejects, or corrects before anything is published.
- Join. Reviewed exposure is joined to the audience for that fixture and market, giving impressions by market.
What exposure measurement does not do
- It shows that a brand was on screen, at what size, where, and for how long. It does not show that a viewer looked at it.
- It does not cover assets that never enter the broadcast frame.
- It depends on access to the broadcast. Where footage cannot be obtained, a fixture is reported as audience only and labelled as such.
- Broadcast feeds differ between markets. Exposure found on one feed is not assumed for another.
From exposure to value
Each appearance is weighted for its size, position, and length, multiplied by the audience in that market, and converted to a value using a rate per thousand. The rate is an input, set by the client: its own negotiated rates, an agency rate card, or a published benchmark. The rate, its source, and its date are shown with each figure.
The value is the equivalent cost of buying the same attention as paid advertising. It leaves out association and brand fit, hospitality, rights usage in owned channels, content and access, and category exclusivity.
How the figures are tested
- Against published figures. Estimates are compared with reported audiences for the same fixture and market.
- Like for like. An average audience is only compared with an average audience, and reach with reach.
- On shape as well as level. The ranking of fixtures by demand is checked against independent signals of public interest.
- Automatically. Tests confirm that each fixture returns a figure and that the licensed and unofficial split is filled in.
- By version. Each calibration produces a new model version, and earlier versions are kept, so change over time is visible.
Each market figure is labelled as anchored to published figures or as modelled where none are published.
Scope
Audience is estimated by country, with DMA-level detail in the United States. Each figure carries its unit, period, and scope. Where data is unavailable for a market or period it is labelled as unavailable, never shown as zero.
When figures are published
No figure is published for a match or race before it has finished. Estimates can change when new reference data arrives.
Independence
Eyeballr is independent. It is not affiliated with any league, governing body, team, or tournament organiser. Published figures from broadcasters and measurement panels are attributed to their publishers and kept separate from Eyeballr estimates.