We went through 1,678 briefs people wrote on Briefing Fox and looked for any sign of context: a deadline, a limit, someone the output was for, something already tried. About three in four had none of those.
The missing part is rarely the task. People say clearly what they want: a proposal, a plan, a reply to a difficult email. What they leave out is everything that would let the AI tell their situation apart from the thousand others that look the same on the surface.
What the count looked like
A simple word match over the goals, so treat these as rough:
- About 8% said anything about what they had already tried or where things stood right now.
- About 8% named a constraint such as a deadline, a budget or a limit.
- About 1 in 5 said who the output was for, and that figure runs high, because the match is loose.
Put together, roughly 26% showed at least one of the three, which leaves about 74% with none. A blank goal box does not ask for any of it, and the facts that feel obvious to the person writing are the ones that never make it onto the page.
The one the AI can never infer
Of the three, the history is the one to put in first, because it is the only one the AI has no way to guess.
Give the AI “I need to handle a team member who keeps missing deadlines” and it will guess the audience (you, a manager), guess a sensible tone, and even guess a plausible constraint (you would like to keep the person). What it cannot guess is that you have already had this conversation twice, or that she was a top performer until two months ago. Those facts decide which advice is still on the table. Without them, the AI offers the first step to someone who is already on the third.
That is why “I have tried X and it did not work” is worth more than a paragraph of background. It removes options. A brief that rules out the obvious answer forces the answer that is left.
What it looks like when it is written down
The vague version:
Context: I'm a manager dealing with a team performance issue.The usable version:
Context: Six people on the team. One of them, three years in and
previously a top performer, started missing deadlines about two months
ago. I suspect something personal that she has not mentioned. I have had
one informal conversation without calling it a performance issue. I do
not want to start a formal process if the cause is personal, but client
work depends on her output and cannot move.Count what the second one holds: a history (one conversation already), a suspicion that changes the approach (personal cause), and a constraint that cuts against it (client work cannot move). Delete any one and the advice changes. That is the standard for a context line: every sentence in it should be able to change the answer.
Where it hides
Most people do not leave context out because it is secret. They leave it out because it feels obvious. You have lived in the situation for weeks, so the fact that makes it unusual no longer registers as a fact. It registers as background.
A way to find it: finish the sentence “This is not like the usual version of this problem because…”. Whatever you write after “because” is the context. If you cannot finish the sentence, you probably have not yet decided what makes the case yours, and that is worth knowing before you ask anything.
This is also why Briefing Fox asks questions instead of giving you a bigger box. A question is easier to answer than a blank, and the answers are where the history and the constraints come out.
The fifteen-second check
Before you send any brief, read the context and ask whether it could have been written by someone else with a similar problem. If it could, it describes a type of situation, not yours. Add what has already happened, what you cannot change, and the thing that makes this case different. Then send it.
Try Briefing Fox free at briefingfox.com
FAQ
What is context in an AI prompt?
It is the part of the brief that describes your actual situation: what has already happened, what you cannot change, and what makes the case different from the standard version. The task says what to produce. The context says what the output has to work around.
How much context should I include?
Enough that each sentence could change the answer. A few specific sentences beat a page of background. If deleting a sentence would not change what a good answer looks like, it is not doing any work.
What is the most important kind of context?
What you have already tried or decided. The AI can guess your audience and your tone, but it cannot know what has failed, and that history removes the most options.
How common is it to include context when briefing AI?
Uncommon. In a rough pass over 1,678 briefs submitted to Briefing Fox, about three in four showed no deadline, limit, audience or earlier attempt in the goal.