Learn Aive, from first login to power user.
Where to start, which product for which need, which best practices to apply: the Academy brings together first steps on Aive Studio and Aive GEO, guides by product and five answer-first method guides, with no promise of citation or ranking.
Get started, learn, practise well.
Three ways in, depending on where you are: your first steps, your product’s guide, then the method.
Getting started
From first login to power user: Aive Studio to produce and adapt your videos, Aive GEO for your visibility in AI answers. A demo on your own content helps you choose where to start.
Guides by product
Aive Studio and Aive GEO are documented module by module on their product pages; Aive API has its own documentation, examples and changelog.
Best practices
Five answer-first method guides: video GEO, preparing a video, measuring AI visibility, governing approval, attributing an outcome.
What is video GEO?
Video GEO means making the information in a video easier for search and answer systems to understand, verify and reuse. It relies in particular on transcripts, chapters, entities, sourced statements, structured data and accessible pages. At Aive, this is the job of the Video GEO module. This preparation does not guarantee that an engine will cite the content.
- GEO
- Generative Engine Optimization: making content easier for AI answer engines to understand, verify and reuse.
- Sourced statement
- A published statement linked to a verifiable source, with status, scope and verification date.
- Entity
- A person, product, organisation, place or concept identified in the video and linked to its mentions.
Prepare, measure, govern, attribute.
Prepare a video
Transcript, chapters, sourced statements and VideoObject: the workstreams of a search-ready video.
Measure AI visibility
Panel, engines, repeated sampling, formulas and variance, without over-reading.
Governance and approval
Rules before generation, human review, logged publishing, statements register.
Attribution
Keep citation, referral, conversion and causality apart, with a declared model.
How do you prepare a video for answer engines?
Start with a video you hold the rights to. Produce a reviewed transcript, split it into chapters that each answer one question, identify verifiable statements and link them to their timestamped moment. Then publish an accessible page with a summary, transcript, chapters and accurate VideoObject markup. Every published statement keeps its source, status and verification date.
- Transcript
- The full, reviewed text of what is said in the video, aligned to time.
- Chapter
- A timestamped segment covering one question or idea, with an explicit title.
- VideoObject
- A Schema.org type describing a video: title, description, thumbnail, duration, upload date and real URLs.
Three workstreams before you publish.
Transcript and subtitles
- Rights to the video and speakers checked
- Full transcript reviewed by a person
- Proper names, products and figures checked
- Subtitles in every published language
Chapters and statements
- One chapter per question or idea
- Chapter titles phrased as answers
- Verifiable statements identified and timestamped
- Each statement linked to its source and status
Page and VideoObject
- A stable, accessible watch page
- An answer-first summary above the video
- VideoObject with real title, description, thumbnail, duration and date
- No essential information carried only by the image
How do you measure visibility in AI answers?
First define a question panel, the engines, markets, languages and period. Ask each question several times, archive the raw answers, then calculate the mention rate, citation rate, AI share of voice and prompt coverage with their denominator. Always read the score with its variance: a change smaller than the observed spread is not a trend.
- Mention rate
- Answers mentioning the brand ÷ observed answers, per engine and period.
- Citation rate
- Answers citing a brand URL or source ÷ observed answers.
- Variance
- The spread of an indicator across samples and days, displayed with the score.
Six steps to a verifiable baseline.
- 01Fix the panelList the questions and prompts tied to your intents and competitors. Do not change them during the period.
- 02Choose engines and access methodFor each engine, record whether it is a provider API or an observation protocol, with model, region and date.
- 03Set a dated baselineMeasure before any action. Without a baseline, no change can be read.
- 04SampleAsk each question several times, at a set frequency, and archive every raw answer.
- 05Calculate with a formulaPublish the formula, denominator and period of each indicator, per engine.
- 06Compare two comparable windowsSame panel, same engines, same method. Interpret the gap against the variance.
What not to conclude
A sound measurement is still an observation within a scope.
- Citation ≠ click
- Referral ≠ causality
- API answer ≠ every user’s experience
How do you govern the approval of generated content?
Set the rules before generation: brand, formats, markets, rights and approvers. Every draft goes through a human review that corrects, approves or rejects it, with separation of roles and a log. Publishing only targets activated destinations. Every published statement points to a source and carries a status and an expiry date. AI prepares; your team decides.
- Playbook
- Rules set before generation: objectives, brand, markets, formats, rights and approval.
- Statements register
- A single base of published statements: wording, source, status, scope, owner and expiry date.
- Activated destination
- A channel or CMS connected and authorised for publishing, with logging.
Four controls, in this order.
Rules before generation
Playbook: brand, formats, markets, rights and approvers defined before any production.
Human review
Corrections, approvals, separation of roles and version snapshots.
Logged publishing
Draft, schedule or publish to activated destinations, with a log.
Sourced statements
Drafts, statements, sources, article and structured data linked to proof.
How do you attribute an outcome to a GEO action?
Keep the levels apart: what was observed, recommended and published, then the outcome recorded. Measure citations, referred sessions and conversions separately, over two comparable windows and with the same method. Declare the attribution model used for assisted conversions. Only assert causality when a robust protocol supports it; otherwise, speak of an observed outcome.
- AI referral
- A session arriving from an identifiable AI referrer, measured separately from crawling and citations.
- Assisted conversion
- A conversion with exposure or referral, under a declared attribution model.
- Observed outcome
- A change recorded between two comparable windows, with no implied causality.
Where should I start with Aive?
With the product that meets your need: Aive Studio to produce and adapt your videos, Aive GEO for your visibility in AI answers, Aive API to plug Aive into your own tools. A demo on your own content helps you choose.
Do I need a new video to do video GEO?
No. An existing video is enough if you hold the rights and its content answers real questions from your audience. Preparation covers the transcript, chapters, sourced statements, watch page and structured data.
Is VideoObject markup enough to get cited?
No. VideoObject describes the video for the systems that read it; it guarantees neither indexing nor citation. It must match a real, accessible page with useful, verifiable content.
Do these guides describe what Aive does with its customers?
Yes, at the level of method. The exact features available depend on your plan and configuration; every module mentioned links to its own page, and limits and statuses are grouped in the methodology.