How to Build a Field Data Collection Budget in Kenya

Why fieldwork budgets go wrong

Most research budgets are assembled from the top down. A design is sketched, a total is agreed with a funder, and fieldwork receives whatever is left once the office work is priced. That is where the trouble starts, because the field is where a study meets distance, approvals, weather, market days and the working hours of the people you are asking for their time.

The symptoms are recognisable. A budget with a single line reading data collection. No pilot, because it did not fit. Supervision absorbed into coordination. Training treated as something that happens for free between recruitment and deployment. Each of these looks like a saving on paper, then shows up later as incomplete interviews, a dataset nobody can use, or a report that slips by weeks.

For studies that generate primary data, collection is rarely a small item. Reference guides for monitoring and evaluation budgeting commonly place it between 30 and 50 per cent of a programme's evaluation budget, and higher where the study is survey-heavy. Budgeting it carefully is not administrative housekeeping. It is the difference between evidence you can defend and a number you hope nobody tests.

Fix the design before pricing it

A handful of variables move the cost of fieldwork more than anything else. Settling them before you ask for a quote changes the conversation, because you can tell a supplier what the study requires rather than accept whatever they propose.

• Sample size. The gap between 300 respondents and 2,000 is not linear once travel, supervision and data checking are included.
• Geographic spread. A Nairobi study and a ten-county study need different teams, different transport and different timelines.
• Method mix. Face-to-face interviews, focus group discussions, key informant interviews and observation each carry their own costs and supervision needs.
• Number of rounds. Baseline, midline and endline studies share set-up costs but repeat field costs.
• How clean the data has to be. A dataset destined for publication or a funding decision needs validation rules, back-checks and honest time for cleaning.

The lines a credible budget contains

Ask for an itemised fieldwork budget and you should expect to see most of what follows. Where a quote collapses several of these into one figure, the missing lines are still being paid for, only invisibly.

• Enumerator fees, calculated as a day rate multiplied by field days multiplied by team size, with travel days counted.
• Per diem, transport and accommodation for teams moving between sites.
• Devices or printing. Tablets and phones for digital collection, or paper instruments and transcription where the method requires them.
• Training. Three to five paid days per round is a common planning assumption, and it is what turns a recruited person into a reliable interviewer.
• Pilot and pre-test. One to two days per instrument, with time afterwards to revise the questionnaire.
• Supervision. A common planning ratio is one supervisor for every five to eight enumerators, with their own per diem and transport.
• Translation and back-translation into the languages respondents actually use.
• Platform licences, where a commercial version of a collection tool is needed.
• Data cleaning and quality checks, listed as work rather than absorbed into someone's salary.
• Handover: cleaned dataset, documentation and, where relevant, the instruments and analysis syntax.

The Kenya lines people forget

Two costs tend to arrive late, and both are entirely predictable. The first is approval time. Formal research in Kenya often requires NACOSTI approval or institutional ethics clearance before collection begins, and counties may expect notification and a community entry process before teams arrive. Weeks and money both belong in the plan.

The second is the pre-visit. Teams that have worked across counties describe discovering, during a short exploratory trip, problems that would have been expensive mid-project: sampled schools reachable only by boat, areas where the power supply makes tablet charging impossible without power banks, or network coverage so weak that uploads have to happen from a town centre. A few days of reconnaissance is cheap insurance.

• Rainy seasons, market days and school calendars that change who is available and when.
• Community entry and local administration courtesies, which take time and sometimes a small budget line.
• Power banks, offline forms and realistic upload windows where connectivity is unreliable.
• Contingency for re-visits when a questionnaire fails validation or a respondent cannot be reached.

How to read a supplier's quote

Fieldwork quotes are easy to compare on the total and hard to compare on substance. These questions usually separate a genuinely lean proposal from one that has simply left something out.

• How many field days, and how many enumerators and supervisors are those days priced for?
• Is the pilot a separate line, and what happens to the instrument after it?
• What share of interviews will be back-checked, and who does the checking?
• Is data cleaning included, and who reviews incoming data during fieldwork?
• Who pays when a questionnaire has to be re-administered because of a programming error?
• What is the plan if approvals arrive late or access is refused in one area?

A planning estimate before you get quotes

You can get a rough sense of scale yourself. Suppose your team achieves eight completed interviews per enumerator per day and your sample is 800 respondents. That is 100 enumerator days of interviewing, before travel. Add a travel day for every two field days in dispersed areas, three to five training days and a pilot day for the whole team, and supervision at one supervisor for every six enumerators, and the total grows quickly and visibly.

Treat that as arithmetic rather than a price. Its value is that it shows which assumptions dominate: interviews per enumerator per day, the travel burden and team size. Change one of those in a conversation with a supplier and you can see what actually moves the number, instead of negotiating a total you do not understand.

Where to save without cutting quality

If the first estimate sits beyond what you have, the useful question is what you can remove without damaging the evidence, not what you can shave off the percentage.

• Shorten the instrument. Fewer well-chosen questions cost less field time and produce better answers than a long form completed carelessly.
• Reduce the number of areas rather than the sample within them, so each area still supports a conclusion.
• Use phone collection where a face-to-face interview adds nothing, and reserve in-person work for what needs it.
• Share one field team across rounds so training and set-up are paid once.
• Build internal capacity so later rounds need less external support. Go Sonic Academy runs a self-paced course in field data collection fundamentals for teams that want that capability in-house.

What Sonivance does with a budget like this

Sonivance Limited is a research, data, monitoring and evaluation consultancy based in Kenya. We scope fieldwork before quoting it: what the design implies, what the areas require, what the approvals path looks like and what supervision will cost. Quotes come itemised, with the team, the ratios and the quality checks named.

If you already have a questionnaire and a sampling design, send them across and we will tell you what fieldwork would take. If the study is still forming, we can help you shape it around the evidence a decision actually needs, which is usually a cheaper place to start than a full tender.