Skip to content

afeezadebayo/research-learning-skills / research-planning

research-planning

Verified

Plan research before searching: the decision it serves, a sharp bounded question, an issue tree of sub-questions, hypotheses, depth matched to stakes, methods from desk research to interviews, surveys and experiments, sources, evidence standards, confidence labels, time plans and agreed outputs. Use whenever someone needs to research something, gets a vague research request, plans a study, market or competitor research, wants to scope how deep to go or needs a research plan before starting.

Research Planning

Most research goes wrong before the first search. The question is vague ("look into competitors"), the decision it should inform is unclear, nobody agreed how deep to go, and the researcher collects everything interesting instead of what matters. Days later there is a long document that answers a slightly different question. Good research starts with a sharp question tied to a decision, a scope that fits the time available, the right methods and sources, and a plan for what "done" looks like.

This skill plans research of any kind, from a two-hour desk scan to a multi-week study: framing the decision and research question, breaking it into sub-questions, setting scope and depth, choosing methods (desk research, data analysis, interviews, surveys, experiments), selecting sources, defining evidence standards, planning time, and agreeing the output format before work begins. It is the entry point for the Research and Learning set.

Operating principles

  1. Decision first. Every research question serves a decision or a clear need to understand.
  2. Sharp questions. Specific, answerable and bounded in time, place and population.
  3. Depth matches stakes. A quick decision gets a quick scan; a costly, hard-to-reverse decision gets deeper work.
  4. Mixed methods where useful. Combine sources so weaknesses in one are covered by another.
  5. Agree the output before starting. Format, length, confidence levels and deadline.
  6. Track sources from minute one. Every claim needs a source you can find again.

Question framing

Weak questionSharp question
Look into competitorsWhich three competitors win most deals against us in UK mid-market, and why do buyers choose them?
Research AI toolsWhich AI meeting-notes tools meet our security requirements and cost under £20 per user per month?
Is remote work productive?What does recent research say about productivity in hybrid versus fully remote software teams?
Find out about the Nigerian edtech marketHow many secondary schools in Lagos use paid digital learning tools, and what do they pay?

A useful test: could two people independently tell whether the question has been answered?

Research depth levels

LevelTimeMethodsUse for
Quick scan30 minutes to 2 hoursDesk research, a few trusted sourcesLow-stakes, reversible decisions, background
Standard1 to 3 daysDesk research, data, a few expert conversationsMost business decisions
Deep study1 to 6 weeksMixed methods: interviews, surveys, data, literature reviewHigh-stakes or strategic decisions

Methods

MethodBest forWatch for
Desk researchExisting facts, markets, regulations, prior studiesOutdated or low-quality sources
Data analysisPatterns in your own or public dataBiased samples, tracking changes
Literature reviewWhat research says on a topicCherry-picking studies
Expert interviewsContext, judgement, hidden knowledgeIndividual bias; few voices
User or customer interviewsNeeds, behaviours, reasonsLeading questions; small samples
SurveysHow common something isPoor questions; low response rates
ExperimentsCause and effectTime, design quality, sample size

When to use, and when not to

Use it before any research task longer than an hour, when a request is vague, when several people will research together, or when research must inform an important decision.

Don't use it for a quick factual lookup; just search. For specific methods, hand off to the specialist skills.

Inputs

  • The decision or need, and who will use the research.
  • What is already known (and assumed).
  • Time, budget and access (data, people, paid sources).
  • Required output and deadline.
  • Any constraints: confidentiality, regions, languages.

Workflow

Step 1. Clarify the decision

Who decides what, by when, and what would change their mind? Write it in one sentence.

Step 2. Write the main question

Specific, bounded and answerable. Note what is out of scope.

Step 3. Break it into sub-questions

Three to seven sub-questions that together answer the main question, using an issue tree: each branch is distinct and together they cover the question.

Step 4. State hypotheses (optional but powerful)

Your current best guesses, so research can test them rather than wander. Be ready to drop them.

Step 5. Choose depth and methods

Use the tables above; match methods to each sub-question.

Step 6. List sources

Primary sources (original data, laws, company filings, studies), secondary sources (reputable reports, journalism) and people to ask.

Step 7. Set evidence standards

What counts as enough evidence for each sub-question; how recent sources must be; how to label confidence.

Step 8. Plan time and output

Time per sub-question, checkpoints, the final format (one-page summary, report, deck) and deadline.

Step 9. Run, then review the plan

At the halfway checkpoint, check whether the question or plan needs to change based on what you have learned.

Research plan template

Decision: <who decides what, by when>
Main question: <specific, bounded>
Out of scope: <...>
Sub-questions:
1. <...> | Method: <...> | Sources: <...> | Time: <...>
2. <...>
Hypotheses: <current best guesses>
Evidence standard: <e.g. two independent reputable sources; data from last 2 years>
Confidence labels: high / medium / low with reasons
Output: <format, length> | Deadline: <date> | Checkpoint: <date>

Issue tree example

Should we enter the Kenyan market in 2027?
  1. Is the market big enough? (size, growth, segments)
  2. Can we win? (competitors, our advantages, local needs)
  3. Can we operate there? (regulation, payments, hiring, partners)
  4. Is it worth it? (revenue potential, costs, payback, risks)

Confidence labels

LabelMeaning
HighSeveral independent reliable sources agree; recent data
MediumSome good evidence, gaps or older data
LowLimited, conflicting or indirect evidence; treat as a hypothesis

Edge cases and judgement calls

SituationWhat to do
Requester cannot state the decisionAsk what they would do differently depending on the answer
Very tight deadlineQuick scan; state confidence and gaps clearly
Answer may not existSay so early; propose a proxy or a way to generate the data
Conflicting stakeholder questionsAgree priority with the decider before starting
Sensitive or confidential topicLimit sources and sharing; anonymise interview data
Research keeps expandingReturn to the decision and the sub-questions; cut the rest

Handoffs to related skills

  • Web and desk research: desk-research
  • Judging sources: source-evaluation
  • Checking claims: fact-checking
  • AI research tools: ai-assisted-research
  • Academic literature: literature-review
  • Interviews and surveys: expert-user-interviews, survey-design
  • Turning findings into insights: research-synthesis
  • Writing the report: the Workplace Writing reports and executive-summaries skills

Worked example

A founder asked for "research on the Kenyan market". The plan clarified the decision (whether to launch a pilot in Nairobi in 2027), wrote a bounded main question, built a four-branch issue tree, chose a standard depth (three days), combined desk research, public data, five expert calls and a scan of regulation, and set a two-source evidence standard with confidence labels. At the halfway checkpoint, payment regulation emerged as the main risk, so time shifted from market sizing to regulation. The final two-page brief recommended a partner-led pilot, with confidence labelled for each finding, and the decision was made in one meeting.

Common mistakes

  • Starting to search before framing the question.
  • Researching a topic instead of answering a question.
  • Depth that does not match the stakes.
  • No source tracking, so claims cannot be checked.
  • Delivering everything found instead of what answers the question.

Output format

1. Decision and main question
2. Sub-questions (issue tree)
3. Hypotheses
4. Methods and sources per sub-question
5. Evidence standards and confidence labels
6. Time plan, checkpoint and deadline
7. Output format

Self-check before you answer

  • Is the research tied to a specific decision or need?
  • Is the main question specific and bounded?
  • Do the sub-questions cover the question without overlap?
  • Are depth, methods, evidence standards and output agreed?

Created by Afeez Adebayo, Marketing, SEO & Growth Expert Website: https://afeezadebayo.com LinkedIn: https://www.linkedin.com/in/afeez-adebayo/ X: https://x.com/za_icon Source: https://afeezadebayo.com/skills-md/productivity/research-learning/research-planning/

Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). You may copy, share and adapt this skill, including for commercial use, as long as you credit Afeez Adebayo and link to the source URL above. Keep this notice in any copy or adapted version.

add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks.Stars: 115kUpdated Claude CodeCodexVerifiedInstall