Choose a video format by observing the conditions in which your intended viewers can start, understand and complete the material, then publish the smallest version that preserves the communication task. For most important work, keep captions and a transcript alongside the video and test at least one lighter alternative such as compressed video or audio. The recommendation changes when visual detail is essential, a platform controls delivery quality, or your intended viewers demonstrate that data is not the limiting condition.
“Nigerians use mobile data” is not a format strategy. It does not tell you which device a particular viewer has, whether they are using a purchased bundle or shared connection, how stable the connection is, how much storage remains, whether they can play sound, or whether the video has to work during travel.
The National Bureau of Statistics publishes operator-reported active internet subscription data by quarter. Its current catalogue includes reports through Q4 2025 (NBS Nigerian Telecoms Data). Those figures describe subscriptions, not unique people, personal data allowances, viewing situations or preferences. They can establish national telecommunications context, but they cannot choose your export settings.
The broader planning principle is explained in There Is No Single Nigerian Audience. The narrower production question here is how to turn observed conditions into a defensible set of media files.
Use the Format Burden Matrix
The Format Burden Matrix is an editorial method for comparing what each version asks of a viewer against what the message requires. It is not an industry standard, and it does not produce a universal Nigerian bitrate. It gives your producer, editor and distribution lead a shared decision record.
Complete one row for each intended viewing situation:
| Condition to investigate | Evidence to collect | Production choice it affects | Observable result |
|---|---|---|---|
| Device and screen | Actual device class, browser or app, available storage | Frame size, text size, aspect ratio, downloadable file | Essential text and action remain legible. |
| Connection | Where viewing happens, connection type, interruption and retry behaviour | File size, streaming host, download option | Media starts and resumes under the observed condition. |
| Data allowance | Voluntarily reported plan terms or an agreed test allowance | Video bitrate, audio version, visible size label | The required version fits the project's defined burden limit. |
| Sound situation | Headphones, shared room, commute, work or quiet setting | Captions, transcript, audio mix | The viewer can understand the piece without assuming sound is available. |
| Visual requirement | Which information exists only in pictures | Audio description, transcript detail, minimum image quality | The task remains possible in each supported version. |
| Language and reading | Spoken language, subtitle need and comfortable reading pace | Subtitle track, translated version, transcript | Meaning survives the selected access route. |
| Time and attention | Available session length and reason for watching | Chapters, shorter cut, summary, return path | The viewer can reach the promised information. |
The responsible producer should record the evidence source beside each entry. “Young audience” and claim national relevance, include research conditions beyond one neighbourhood or city and report the limits of that sample.
Start with the communication task
Before exporting anything, write one sentence describing what the viewer must be able to do afterwards. Examples include:
- identify the three documents needed for an application;
- compare two creator partnership options;
- understand why a production choice affected a scene;
- decide whether to watch a full interview;
- remember a deadline and the action attached to it.
This sentence prevents a false choice between “beautiful video” and “low data”. If a reader can complete the task from a 70 KB transcript, the transcript is a valid access route. If the task is to compare movement, composition or a product demonstration, the pictures may be indispensable and compression has a quality floor.
Do not make audio merely by stripping the picture track. Listen without the screen. Add concise spoken descriptions when a diagram, name caption or physical action carries necessary information. W3C's current media guidance recommends planning accessibility from the start and identifies captions, transcripts, visual description and accessible media players as distinct components (W3C Web Accessibility Initiative).
Build a comparable access set
Create versions from the same locked script so you can compare burden without changing the argument:
- Primary video: the visual version that preserves all necessary detail.
- Compressed video: a smaller version checked for legible text, faces and demonstrations.
- Audio: a properly mixed and, where needed, described version.
- Caption file: reviewed against the final sound, including meaningful non-speech information.
- Transcript: a separate readable page or file, with visual information added when it is necessary to understand the work.
You may not need to publish every version. Producing a small comparison set before release shows which compromises are real. It also exposes scripts that rely on silent screen text or editors who assume everyone can hear the soundtrack.
On YouTube, the viewer does not necessarily receive one fixed rendition. YouTube says it changes stream quality according to connection speed, screen size, source upload and browser, while viewers can also change quality manually (YouTube Help). An uploaded 1080p master therefore does not establish one data cost for every play. For a downloadable file you control, measure the actual file. For adaptive streaming, record observed playback conditions and treat any bitrate calculation as an estimate.
Illustrative scenario: an eight-minute creator interview
The following numbers are assumptions for a reproducible comparison, not Nigerian market measurements and not promises about a platform stream.
Assume an eight-minute programme, or 480 seconds, is exported in four access versions:
| Version | Assumed total bitrate or file size | Calculation | Approximate transfer |
|---|---|---|---|
| Primary video | 2 megabits per second | 2 × 480 ÷ 8 | 120 MB |
| Compressed video | 0.5 megabits per second | 0.5 × 480 ÷ 8 | 30 MB |
| Audio | 0.096 megabits per second | 0.096 × 480 ÷ 8 | 5.76 MB |
| Transcript | Assumed measured file | Not bitrate-based | 0.08 MB |
The formula divides megabits by eight to obtain megabytes. It uses decimal units for simplicity. Container overhead, captions, adaptive streaming, caching and retries can change actual transfer.
If a viewer has chosen to reserve 500 MB for all remaining activity in a period, one complete transfer would consume approximately:
- primary video: 120 ÷ 500 × 100 = 24%;
- compressed video: 30 ÷ 500 × 100 = 6%;
- audio: 5.76 ÷ 500 × 100 = 1.152%;
- transcript: 0.08 ÷ 500 × 100 = 0.016%.
This calculation does not say that Nigerian viewers have 500 MB available. It shows why your team should state the file burden instead of hiding it behind a resolution label. If you need a monetary estimate, use the participant's actual current bundle price and allowance, identify the operator and date, and show the arithmetic. A simple pro-rata calculation may still misrepresent bonuses, expiry, rollover, zero-rating or out-of-bundle charges, so label its limitations.
The preferred version is not automatically the smallest. If the interview includes visual demonstrations that the audio does not describe, a 5.76 MB file may fail the communication task. The right decision may be a 30 MB compressed video with readable close-ups, plus transcript and audio for other situations.
Run a viewing-conditions check
A small check can improve one release, but it is not national audience research. Recruit people because they match intended viewing situations, then disclose the number and selection basis. If your distribution claim includes several locations, do not recruit only in Lagos and generalise the result.
Give each participant a real task, not a preference survey. For example: “Find the two criteria the guest says should change a creator brief.” Capture:
- start success: successful starts divided by offered play attempts, by version;
- start delay: seconds from the participant's action to usable playback;
- completion: completed plays divided by successful starts;
- task comprehension: correct task responses divided by completed attempts;
- transfer burden: measured bytes for a controlled download, or device-reported data where reliable;
- recovery: whether the version resumes after a realistic interruption;
- access failure: the specific visual, audio, language or control barrier reported.
Keep the denominators. Ten completions could mean ten of ten starts or ten of fifty. Do not add plays from separate platforms and call them unique people unless you have a valid deduplication method.
Platform analytics are useful after publication, within their limits. YouTube's Audience reports can show watch-time distribution by device type, top geographies, subtitle languages and the formats viewers watch. YouTube also warns that some geography and demographic data can be limited (YouTube Audience reports; limited analytics data). These reports do not reveal an individual's data
Distribution choices that reduce avoidable burden
Once a format passes the task check:
- show the duration and download size before a user commits to a file you host;
- avoid autoplay, especially when the user has not chosen to spend data or hear sound;
- place the transcript beside the player rather than behind an email form;
- provide accurate captions that can be turned on and off;
- use clear chapter labels for longer material;
- keep essential words large enough to survive the compressed version;
- state when an audio version omits visual material;
- preserve a high-quality master even when the public delivery copy is smaller.
W3C notes that captions are useful in loud and quiet environments as well as for people who are Deaf or hard of hearing, while a transcript supports additional access situations (W3C on video captions). Low transfer size and accessibility are related in some situations, but they are not the same objective. A tiny inaccessible file is still a failed version.
A nine-working-day decision sequence
Day 1: The editor and commissioning lead define the one action or understanding the material must produce. They identify visual information that cannot be discarded.
Days 2 and 3: The audience lead documents intended viewing conditions through existing first-party evidence or a disclosed recruitment plan. Unknowns remain marked unknown.
Day 4: The editor exports the comparable access set and records exact sizes, bitrates, dimensions and durations. The language reviewer checks captions and transcript.
Day 5: The producer verifies keyboard access, controls, transcript placement and the meaning of audio-only playback. Any legal or policy requirements receive qualified review.
Days 6 and 7: The team runs the viewing-conditions check. Results are recorded by version with denominators and specific failures.
Day 8: The analyst compares task comprehension, start success, completion, measured burden and accessibility. Platform analytics are used only as supporting historical evidence.
Day 9: The commissioning lead chooses the default and alternatives. The decision record states which viewer conditions are supported, which are not and when the choice will be reviewed.
Choose a version only if it preserves the critical information and passes the pre-agreed start, comprehension and accessibility requirements for the intended situations. If the lightweight version fails on meaning, improve its script or retain the larger file. If observed data burden is not material for the defined audience but image detail is, prioritise visual quality and continue offering captions and a transcript for accessibility.
Related reading
Frequently asked questions
Should every Nigerian video have a low-resolution download?
Not automatically. Provide it when the intended distribution route supports downloads and your evidence shows it solves a real access condition. Always retain a high-quality master and check that the smaller file remains usable.
Can national subscription figures tell me the right resolution?
No. Subscription totals establish market context, not an individual's connection, allowance, device or purpose. Use them to frame research, then measure the actual viewing situation.
Do high views prove that data was not a barrier?
No. Views describe recorded playback under a platform's definition. They do not count the people who declined to start, failed before measurement or chose another format, and they do not establish why someone watched.
Is a transcript a replacement for captions?
No. Captions are synchronised with the pictures and sound, while a transcript is a separate reading route. They serve overlapping but different needs. Plan both when the material and applicable requirements call for them.
By Isong Godswill
By Omodot Jackson