00:00:00 Introduction and recap

Welcome to today's session on conservation, which focuses specifically on turning a results chain into a data model using a case study as a working example. This is the second part of a three-part series. The first session covered the foundations of what makes conservation monitoring and evaluation its own distinct discipline, along with the frameworks that guide it. Today, we take that conceptual thinking and turn it into a design blueprint. In the next session, a full demonstration of a conservation monitoring system will be shown in the ActivityInfo software.

To quickly recap the key takeaways from the previous webinar, conservation runs on ecological time, and ecological change is not linear. A flat line in your data might indicate a failure, or it might represent a system that is quietly healing, and you cannot tell the difference from the line alone. Secondly, every program carries dual targets—nature and people—and they must be tracked together. Logframes alone are insufficient for conservation work; results chains are necessary because they turn every assumption into something you can test. Finally, threat reduction indicators and ecological outcome indicators must be read together. One tells you the pressure came off, while the other tells you if anything recovered. They do not move at the same speed, and reading either one in isolation will mislead you.

Zooming in on the concept of dual targets, conservation almost always involves tracking two pathways simultaneously. The nature pathway involves tracking populations, poaching, and habitat conditions. The people pathway typically involves livelihoods, such as income, jobs, and governance. You cannot run one without the other. If you track only nature, you might watch a species population fall without ever knowing why, because the root cause sits on the people pathway. Conversely, if you track only people, you might have a livelihoods project reporting rising income while the species you were funded to protect quietly disappears. The two pathways are part of the same program and the same system. The challenge is tracking both in one data model. While today's focus is on Natural Resource Management, the logic applies equally to species and biodiversity monitoring, protected area management, and nature-based solutions. The case study used today is fictional but is based on the real structures of actual programs.

00:06:11 The case study: Community wildlife areas

To introduce the case study, consider this premise: the people who carry the cost of wildlife are rarely the people who collect its value. For example, imagine a project working to protect a lion population. If that lion eats a farmer's livestock, that animal is not a conservation asset to the farmer; it has destroyed how his household eats. Meanwhile, a tourist vehicle might pay to photograph that same lion, generating money for a lodge operator or a government agency, but not for the farmer. The farmer carries the entire cost of an animal that earns someone else a living. From the farmer's perspective, the only rational solution is for the lion to be gone. This is not a failure of awareness that can be fixed with posters or training; it is the result of incentives working exactly as expected.

The model we are looking at attempts to change these incentives through Community-Based Natural Resource Management. This philosophy involves giving the people who live in a natural resource the legal right to manage and benefit from it. Our specific unit is a Community Wildlife Area, which is a registered area with agreed boundaries and an elected committee holding conditional rights over its wildlife. This takes place on communal land held collectively by the community, which is crucial because people cannot have legal rights to the wildlife if they do not have rights to the land.

Before this intervention, wildlife belongs to the state by law. The revenue generated from lodges, tourism, and licensed hunting goes to the state or licensed operators. Because the local household carries all the costs and receives none of the returns, wildlife is a pure liability, and removing it is rational. After the intervention, rights over wildlife are conditionally devolved to the community. Income from tourism and hunting is paid to the Wildlife Area, and meat and jobs are distributed to members. The wildlife itself hasn't changed, but the rights holders have, giving them a reason to keep the animals alive.

The financial side of this deal relies on four concrete revenue streams: joint venture tourism where the community takes a share of a private lodge's revenue, regulated licensed safari hunting based on annual quotas, game meat distribution allocated to households, and jobs/dividends such as community scout positions and cash distributions. The wildlife has structurally transformed from a liability into a cash crop. Underneath all four streams is a specific data shape: somebody receives something on a specific date in a specific amount.

In theory, this works as a loop. Wildlife generates income, meat, and jobs. These benefits make the wildlife worth tolerating. Increased tolerance causes poaching and retaliatory killing to fall. The animal population holds or recovers, which in turn sustains the income that started the loop. A logframe cannot draw this circular logic, but a results chain can. A loop like this does not fail all at once; it fails at a specific link. The income might not reach the household, or it might reach them but tolerance doesn't shift because losses still outweigh benefits, or tolerance shifts but populations fall due to external pressures like a drought. These are three different problems requiring different solutions. Hiring more scouts solves the third problem but wastes money if the issue is actually the first or second. Monitoring must sit in the middle of this loop to tell you which link is weakening.

00:15:45 The field register and data silos

While this loop makes for a great story in annual reports, a story without data is just an anecdote. Donors eventually look for measurable results. To prove the income arrived, reached the household, reduced poaching, and helped populations recover, you need data. The central instrument for this is the field register. A community scout employed by the wildlife area walks a patrol and records what they find—a sighting, a carcass, a snare, a crop raid—on a handheld device or paper. One event equals one entry. The system then generates the monthly and annual views needed by managers, ensuring that annual and monthly figures actually agree. Because every area uses the same form, you create one shared, comparable database.

However, the reality in almost every program is a disconnect. On one side, you have structured field registers submitted from patrols. On the other side, financial and M&E data sit in office ledgers or spreadsheets maintained by one person while everyone else emails them. Both are real, structured data, but they are collected by different people in different formats on different schedules, and nothing joins them. When asked if the benefits are actually reducing the pressure on wildlife, answering can take weeks of manual reconciliation. The goal of our data model design is to get these two isolated systems to talk to each other.

00:18:02 Drawing the results chain

Most proposals list increasing community income and increasing animal populations as parallel bullet points, but a list of goals is not a results chain. What is missing is the arrow between them explaining why income changes, what happens to the animals, and under what conditions. That arrow is the mechanism of the program, and if it is never drawn, it is never measured.

Using the CMP Open Standards version 5 notation, we can map this out. Plain boxes represent activities, dashed boxes are monitoring activities, hexagons are actions, plain rectangles are intermediate results, arrow shapes are threat reduction results, green boxes are biophysical results, and ovals are focal values (what you ultimately seek to affect, like biodiversity or human well-being).

Building the chain starts on the left with activities: training and equipping community scouts, brokering tourism and hunting agreements, and distributing benefits. A dashed monitoring activity box includes the field register and annual game counts; this doesn't change the world itself, but tells you if the other activities are working. Together, these form the action of establishing and supporting a wildlife area.

From this action, two pathways run in parallel. The people pathway is short: communities earn income and jobs (intermediate result), leading directly to the human well-being focal value of community livelihoods. The nature pathway is longer: households come to value and tolerate wildlife (intermediate result requiring attitude measurement), which causes poaching and retaliatory killing to fall (threat reduction result). This reduced pressure allows the wildlife population to recover (biophysical result), ultimately reaching the biodiversity focal value.

The most important element is the arrow running from the people pathway down to the nature pathway: benefits buy tolerance. This is the hinge of the entire program. The livelihood work is not a side benefit; it is the mechanism producing the conservation results. If poaching does not fall, the first place to look is not the scouts, but that connecting arrow to see if benefits actually arrived at the households carrying the losses. You can only answer this if your benefit data and poaching data live in the same system.

Indicators attach to specific boxes on this chain, not to the project as a whole. Total cash income and households receiving distributions attach to the people results. Reported tolerance from household surveys attaches to the intermediate result. Poaching incidents recorded attach to the threat reduction result. Estimated population by species attaches to the biophysical result. Usually, results chains are drawn and admired, while indicators are written in separate documents, and the connection is lost. We will do the opposite by turning these directly into a data model.

00:26:58 From results chain to data model

To transform the results chain into a data model, we must define a Management Information System (MIS). An MIS is a defined, repeatable process by which data is collected, stored, and handed back to the people who need it. A spreadsheet can be part of an MIS, but it is not a system by itself. The database is the container holding everything. Inside the database are forms, which are tables holding one specific category of thing (e.g., a form for species, a form for wildlife areas). A common design mistake is mixing categories in one form. Inside the forms are fields or columns (the properties of the thing), and records or rows (the individual entries).

A relational data model is the blueprint for this data. It must contain data entities (the real-world things that become forms), attributes (the properties that become fields), and relationships (how things connect). In a relational database, a fact is stored once and referenced everywhere. For example, the name of a wildlife area is entered once, and every subsequent form references that single entry rather than asking users to type it repeatedly.

There are two helpful rules for turning a results chain into a data model. Rule number one is that a monitoring activity in your results chain often becomes a data collection form in your database. The field register and the annual game count each become their own forms. Rule number two is to find the hub. One record entered once should join everything that points at it. If the wildlife area is the hub, the field register, game count, and distribution forms all point back to that single wildlife area record.

Data can be categorized in three ways. Reference data is the controlled list you use to classify information, like species or regions; it changes very slowly and is usually maintained by someone else. Master data is the registry of core entities you operate on, like the wildlife areas; it changes occasionally. Transaction data is the event log of field activities, like register entries or surveys; it changes constantly and grows without limits. The test to distinguish them is not what the data is about, but what event causes a new row to appear. A new row in reference data appears when taxonomy changes, which is rare. A new row in transaction data appears every time a scout sees an animal. Typing reference data by hand leads to spelling errors and fragmented analysis, which is why it should always be selected from a dropdown list.

In a relational database, relationships are key. A single record for a wildlife area, like "Red Hills," is created once. The field register, annual game count, and annual area return all point to it. This hub classifies all nature data, metadata, and people data, providing the join that solves the problem of isolated ledgers.

Key fields make records unique. A serial number is used when every entry is a discrete event in time, such as a field register where a scout might legitimately see two different elephants at the same waterhole on the same day. A composite key combines multiple fields (like area, year, and species) and is used when only one record should exist per combination. This prevents someone from accidentally entering two different annual buffalo counts for the same area in the same year.

Subforms are nested tables inside a record. For example, a wildlife area's management committee is a list of people with names, roles, and genders. Instead of flattening this into awkward columns in a spreadsheet, the committee sits as a subform inside the Wildlife Area record. This allows you to easily chart metrics, like the percentage of committee seats held by women across all areas, directly from the structured data.

00:41:58 The finished data model and conditionality

The final data model contains nine forms organized into layers. On the left is the reference data, including species and regions, alongside the indicators and their target values. In the middle is the master data: the Wildlife Areas form, which includes the management committee subform. This is the hub that everything points back to. On the right is the monitoring layer, which grows every day. The field register produces poaching incidents, the distribution form produces income and households reached, the household survey produces reported tolerance, and the game count produces population by species. Every form in this diagram produces an indicator, classifies one, or proves a condition of the deal, tracing directly back to the results chain.

To understand how fields are reflected in the model, look at the concept of conditionality. To hold rights over wildlife, a community must demonstrate agreed boundaries, a register of members, an elected committee, a management plan, and an annual general meeting. You cannot prove a condition with a story; it must be proven with data. Agreed boundaries become a geographic coordinate field. The register of members and elected committee become subforms. The management plan becomes a status field and a date. This governance paperwork is entered once and is reportable across every area. The conditions fall naturally into the existing model without needing a separate, complex compliance module.

00:46:30 Generating reports and answering core questions

Ultimately, this data model produces a master pivot table generated automatically by the system. Down the side are the Wildlife Areas. Across the top are columns for total cash income, game meat distributed, households receiving distributions, reported tolerance, estimated buffalo population, and conflict incidents. These columns pull from completely different forms—distributions, household surveys, annual returns, game counts, and field registers—collected by different teams on different schedules. Yet, they sit in one table, and nobody had to manually reconcile anything because every form references the same wildlife area hub.

The columns on the left represent the people pathway, and the columns on the right represent the nature pathway. They are no longer two separate systems; they are columns on the same row. This answers the question of how to get nature and people into one report: you put them in one database with one hub, and the report falls out of it naturally.

This also answers whether the areas receiving the most benefits have the least poaching. You can view total cash income by year across all areas, which shows the model is earning money. Next to it, you can view poaching incidents by year from the field register. Holding the two together, you can see if the poaching line comes down as the income line goes up. While correlation is not causation, the model makes it very cheap to test this properly by comparing high-benefit areas against low-benefit areas with a time lag, because both numbers are already set against the same areas and years.

Open Standards lists six different audiences for conservation information, including the field team, managers, and donors, all asking genuinely different questions. Most systems fail because they are designed exclusively for the donor; the people collecting the data get nothing back, and nobody maintains a system that gives them nothing. The model designed today serves the scout, the committee, the manager, and the donor from the exact same records because it was built around what actually happened in the field, rather than just what needs to be reported.

00:54:08 Key takeaways and conclusion

There are five main takeaways from this session. First, a results chain is an architecture, not a list of goals, and the arrows represent the claims you are testing. Second, every box on the chain becomes something in the database; specifically, monitoring activities become forms. Third, you must separate reference, master, and transaction data, testing them based on what event creates the data, not what the data is about. Fourth, find your hub so that one record entered once joins everything that points at it. Finally, nature data and people data do not need one unified report; they need one common reference field to link them.

The session concluded with a Q&A, noting that the data model presented will be published as a template on the ActivityInfo website for users to adapt. A final poll revealed that the majority of attendees work in community-based natural resource management.