How to track programme KPIs from KoboToolbox data

To track programme KPIs from KoboToolbox, design each form around the indicators it feeds, write down exactly how every KPI is calculated from the fields, check data quality as submissions arrive, and connect the form to wherever your targets live so the numbers update without retyping.

Milestra team · 26 October 2026 · 7 min read

Why KoboToolbox data often stops at the export

KoboToolbox is excellent at what it was built for: collecting structured data in the field, offline if needed, on phones and tablets. Many NGOs run every household survey, water point check and training attendance list through it.

The gap usually comes afterwards. Data sits in the form's table until someone exports it to Excel, filters it, builds a pivot table and copies the result into a donor report. Each export is a snapshot, so by the time the report is written the numbers are already a month old. When two people do the export slightly differently, you get two different answers to the same question.

Tracking KPIs well means closing that gap: deciding in advance how each number is produced, and producing it the same way every time.

Step 1: Start from the indicator, not the form

Before you build or edit a form, list the KPIs it needs to feed. For each one, write a short definition:

  • Indicator: % of water points functional
  • Unit: percentage of water points visited
  • Source: water point monitoring form
  • Calculation: water points where status is "functional", divided by all water points visited in the period
  • Frequency: monthly
  • Disaggregation: by district
  • Target: agreed in the logframe

This is essentially an indicator reference sheet. It takes ten minutes per indicator and saves hours of argument later, because everyone knows exactly which field and which rule produce the number.

Step 2: Design forms that make KPIs easy to calculate

A few design choices make a large difference when you come to calculate KPIs.

Use choice lists, not free text

"Is the water point working?" as a select_one question with fixed answers (functional, partly functional, not functional) can be counted reliably. The same question as free text will produce "yes", "Yes", "working", "ok" and "mostly", which nobody can count without cleaning.

Give fields clear, stable names

In the form builder, or in XLSForm, each question has a name as well as a label. Labels can change and can be translated. Names should be short, lowercase and stable, for example wp_status or hh_size. If you rename a field halfway through a programme, every calculation that depends on it breaks.

Capture the date of the activity

KoboToolbox records when each submission arrives (_submission_time). That is not always when the activity happened: an enumerator might sync a week of visits on Friday. If the timing matters for your reporting period, add a date question for the visit or session itself.

Record the location and the group

Include district, site or school as a choice list, and the demographic questions you need for disaggregation, such as sex and age band. It is much harder to add these after data collection has started.

Use calculations sparingly

A calculate field can do useful work, such as flagging a household as "at risk" based on several answers. Keep the logic simple and document it, because whoever reads the KPI later needs to understand it.

Step 3: Choose the right calculation for each KPI

Most programme KPIs from survey data fall into a handful of calculation types.

CalculationUse it forExample
Count of submissionsActivities completedNumber of hygiene sessions held
Sum of a fieldTotals reported per submissionPeople reached (sum of participants per session)
Percent where a condition is metShares and rates% of water points functional
Average of a fieldScores and sizesAverage reading score in the assessment
Count of distinct valuesCoverageNumber of villages reached
Latest valueStatus that changes over timeCurrent stock of test kits at a facility

Two common traps:

  • Counting submissions when you mean people. If one session form records 25 participants, counting forms gives you sessions, not people. Use the sum of the participant field.
  • Double counting repeat visits. If the same household or water point is visited every month, counting submissions over a quarter will count it three times. For coverage, count distinct IDs; for status, use the latest visit per site.

Step 4: Filter to the right population and period

Most KPIs need a filter as well as a calculation:

  • Period: only submissions from this quarter, or since the start of the grant
  • Population: only programme schools, only the WASH component, only women-led households
  • Status: only submissions that have passed review

That last one matters. KoboToolbox lets reviewers set a validation status on submissions. If your team uses it, filter your KPIs to approved submissions so that test entries and rejected records do not inflate the numbers.

Step 5: Check data quality as it arrives

Problems are cheapest to fix in the first few days after collection, while the enumerator still remembers the visit. A short routine helps:

  • Look at the latest submissions every week. Scan for blanks, impossible values and duplicates.
  • Watch the trend, not just the total. A district that suddenly reports zero may have a connectivity problem, not a programme problem.
  • Compare with what you expect. If a training of 30 people produced 300 records, something has gone wrong.
  • Fix at the source. Correct the record in KoboToolbox rather than in an export, so every later calculation uses the corrected value.

Our guides on data quality assessments and cleaning survey data go further.

Step 6: Keep the KPIs current without retyping

Once the definitions are clear, the remaining job is to get the numbers from KoboToolbox to wherever your targets and reports live, on a regular schedule.

There are three common approaches:

  1. Manual export. Export to Excel each month and recalculate. Fine for one or two indicators; fragile beyond that.
  2. Spreadsheet or BI connection. Pull data through the KoboToolbox API into a spreadsheet or dashboard tool. Flexible, but someone needs to build and maintain it.
  3. Direct link to your goals and indicators. Connect the form to the system where targets and owners already sit, so each KPI updates automatically and appears in donor reports.

In Milestra, the third approach works like this: you add your KoboToolbox server address, an API token and the form's asset ID once. For each KPI or key result, you choose the calculation (count, sum, average, percent where, distinct or latest value), add any filter, pick the date field and start date, and optionally a breakdown such as sex or district. You can test the number against the live data before saving it. After that, the form is read every night and the KPI updates, with the breakdown flowing into the donor report. The field data section of the features page has the detail.

A worked example: a multi-country WASH programme

Imagine an NGO running a WASH programme in three countries. Field teams use two KoboToolbox forms: a monthly water point check and a hygiene session attendance form.

KPIFormCalculationFilterBreakdown
% of water points functionalWater point checkPercent where wp_status = functionalThis month (each point is checked once a month)Country
Water points visitedWater point checkCount of distinct wp_idThis quarterCountry
People reached with hygiene sessionsSession attendanceSum of participantsSince grant startSex
Hygiene sessions heldSession attendanceCount of submissionsThis quarterDistrict

The regional M&E lead reviews the four KPIs in a short weekly check-in. When one country's functionality rate drops, the team can see it within days, not at the end of the quarter, and follow up with the maintenance partner.

Common mistakes

  • Changing a form mid-programme without versioning the indicator. If you change answer options, note the date and check every KPI that uses the field.
  • Mixing test and live data. Delete test submissions or filter them out before you start reporting.
  • Building KPIs nobody owns. Each KPI needs a named owner who looks at it regularly and acts when it moves.
  • Relying on one person's spreadsheet. If only one person knows how the numbers are calculated, the calculation is a risk.

Frequently asked questions

Can KoboToolbox calculate KPIs on its own?

KoboToolbox includes basic summaries and reports for each form, which are useful for checking data. Most teams still need a separate step to apply indicator definitions, filters, periods and targets, and to combine results from several forms.

How do I connect KoboToolbox to another tool?

KoboToolbox has an API. You need the server address you use (for example the global or EU server), an API token from your account settings, and the form's asset ID, which appears in the form's web address.

How often should KPIs from survey data update?

Match the frequency in your indicator definition. For routine monitoring, a nightly or weekly update is usually enough. What matters most is that the calculation is the same every time.

How do I avoid counting the same person twice?

Give each person, household or site a unique ID in the form, and count distinct IDs for coverage indicators. If you cannot record IDs, report the indicator as attendances or contacts rather than people.

Is it safe to connect survey data to other systems?

It can be, if you connect only the fields you need, avoid moving personal data where totals will do, and use a token from an account with access only to the relevant forms. Involve your data protection lead before you set up any connection.

Try Milestra for free

Put this guide into practice in 10 minutes

  • Goals, key results and indicators in one place
  • Two-minute weekly check-ins with reminders
  • Donor and board reports from live numbers
  • Free for up to 5 people · 50% off for nonprofits

Keep reading