Conflict analysis is a structured way of understanding a conflict: its history, the groups involved, each group's perspective and the causes. That's how the GSDRC topic guide defines it. It's the first step in any peace plan, and the base of conflict sensitivity. A county can do a useful version in a few days, if it talks to the right people.
Key points:
- Look at the profile, causes, actors and dynamics.
- Hear from every group, not only leaders.
- Separate facts from claims.
- Update the analysis as things change.
Why do a conflict analysis?
It stops you from acting on assumptions. Peace Insight says that success in early warning starts with a reliable analysis of local conflict dynamics. The guide shows what can go wrong without it, with a water project in rural Kenya that raised conflict.
It also helps partners to agree. A shared picture avoids wasted effort.
What are the main parts?
A standard analysis covers four things: the profile (what's the place like?), the causes (why is there tension?), the actors (who's involved?) and the dynamics (how's it changing?).
What goes into the profile?
Describe the area: people, land, economy, services, politics and history. Note past violence and peace agreements. Add maps. Keep it short.
How do you find the causes?
Separate them into layers. The AFSC study is a useful model, with structural drivers, proximate factors and triggers.
Structural causes are deep: land, inequality, exclusion. Proximate factors are recent: a tight election, drought, new gangs. Triggers are single events: a killing, a disputed result, a viral clip.
How do you analyse actors?
List the people and groups involved. What do they want? What do they fear? What power do they have? Who influences them? Who could help?
Include those who keep the peace: elders, teachers, women's groups and traders. Include those who profit from conflict. Be careful with names, and don't write down allegations you can't support.
How do you map a conflict?
Draw it. Put actors in circles sized by their power. Use lines for alliances, and crosses for conflict. A map of who relates to whom can show gaps that words can miss.
Geographic maps help too. Mark incidents, water points, boundaries, markets and police posts. Then look for patterns.
What about dynamics?
Ask what's getting better, what's getting worse, and what's coming. Mark dates: elections, party primaries, the dry season, school fees, court rulings. The AFSC study says that violence peaks during nominations, campaigns and after results.
Where do you get information?
Use a mix. Read court cases, county records, news, police data and past reports. Talk to people. Hold small group discussions with women, youth, elders and traders separately, since they'll tell you different things.
Use ACLED for recorded events and parliamentary reports for history. Check how old each source is.
How do you handle conflicting accounts?
Expect them. Write down each group's account as its own view. Note what's confirmed, and how. Disagreement is a finding in itself.
How do you protect people?
Don't record who said what. Keep notes secure. Meet in safe places. Don't make promises. A conflict analysis can be dangerous if it leaks, so limit who can see the detail.
Who should do it?
A small mixed team works well: someone who knows the area, someone skilled in analysis, and people from different groups. Outsiders bring distance. Insiders bring knowledge. You need both.
How long does it take?
A quick version takes a few days. A detailed one takes weeks. Use what fits your decision. A rapid analysis before a borehole is fine. A county election plan needs more.
How do you write scenarios?
Describe three futures: best, most likely and worst. The AFSC study does this for 2027. Its most likely scenario is contained but persistent local tension, with early warning systems detecting risks and responses being inconsistent.
For each scenario, ask what you'd see first, and what you'd do.
How do you turn analysis into action?
List options against each cause. Decide who does what, by when, at what cost. Note what you won't do, and why. A good analysis ends in decisions.
How often should it be updated?
Update it when something big changes: a killing, a new candidate, a drought. In an election year, review monthly. Keep old versions, so that you can learn how your thinking changed.
How does it link to early warning?
Your analysis tells you what to watch. It gives you the local indicators. Early warning reports then feed back into the analysis.
How do you share it?
Write two versions. One is detailed, for a small group. The other is a short public version that avoids naming people or ranking communities. Think about who could misuse it.
What are common mistakes?
Treating the loudest voice as the truth. Using old data. Writing a long report that nobody reads. Forgetting women and young people. Stopping at analysis, with no decision.
What does a worked example look like?
Imagine a ward where herders and farmers share a river. The profile notes two villages, a market and a new sand-harvesting business. The structural cause is unclear land boundaries. The proximate factor is a failed rainy season. A possible trigger is a cow killed in a maize field.
The actors include elders on both sides, sand traders who want the riverbank, a ward aspirant, and women who fetch water. The analysis suggests reviving the old grazing agreement before the dry season, and talking to the sand traders early. That's a plan you can act on next week.
How do you check your analysis?
Show a draft to people from each group, without revealing who said what. Ask what's wrong or missing. If everyone agrees with everything, you may not have asked the right people.
Read next
- Conflict early warning system for Kenya: the TI pack page for this work
- Peacebuilding in Kenya: all the pages in this section
- Conflict sensitivity
- Election hotspots in Kenya
- A county peace and cohesion plan
- Takwimu Intelligence (TI) home page
- National cohesion in Kenya: the guide to this library
- The TI platform and how we check our numbers
Where TI fits
Takwimu Intelligence (TI) brings survey answers, complaints and public media into one place for analysis, and it keeps what's confirmed apart from what's only claimed.
To see it on your own work, book a walk-through.