BigCat Research

Commissioning field observation research in Turkey: observer training, forms and sampling

Short answer: Field observation research is only as good as three things you can inspect before fieldwork starts: how observers are trained and calibrated, whether the observation form records concrete behaviour rather than impressions, and how sites, days and time slots are sampled. Add a written consent and photo policy, and insist on receiving site-level data, not only a summary deck.

What observation adds that surveys and mystery shopping do not

A survey tells you what people remember and are willing to say. Mystery shopping tells you how one scripted scenario was handled. Structured observation records what actually happens: how long the queue is at 12:40 on a Friday, whether signage sends customers to the wrong counter, which process step is skipped. That strength is also the risk: two observers in the same branch can produce two different pictures unless the design forces them to look at the same things in the same way.

Observer training: what "trained" should mean

Any provider will say its observers are trained. Ask what the training consists of. A credible answer usually includes:

If the provider cannot describe calibration, you are buying individual opinions arranged in a table.

The observation form: behaviour, not adjectives

Ask to see the draft form before signing, or a redacted form from a comparable project. A usable form has a defined unit of observation (one interaction, one 15-minute slot, one touchpoint), items that are counted or binary (yes/no, seconds, number of people) rather than unanchored "quality" ratings, start and end times, context fields such as site, weekday, time slot and staffing on duty, and a mapping from each item to a written service standard or process step. Warning signs are "the branch" as the unit, free text as the main output, and items nobody can act on. Paper or tablet matters less than whether each field has a definition the observer learned in training.

Sampling sites, days and time slots

Observation samples are small, so where and when you look matters. Three decisions should be visible in the proposal:

  1. Site selection. Which branches, dealers or stores, and why. Random selection within strata (region, size, performance tier) is defensible. "The sites head office suggested" is a case study, not a sample, and should be labelled as one.
  2. Day and time coverage. A quiet Tuesday morning and a Saturday afternoon are different realities. Ask for a visit matrix showing how peak and off-peak slots are spread across sites.
  3. Observations per site. One visit tells you about that visit. To compare sites, each needs enough observations that one unusual event does not define it. Ask the provider to justify the number.

In Turkey, also consider seasonal footfall, religious holidays, and whether urban and provincial sites are both covered.

Consent, KVKK and site access

Observation involves employees and customers sharing the same space. Personal data processing in Turkey falls under Law No. 6698 (KVKK). Settle in writing whether observation is announced to site staff; whether individual employees are identifiable or only sites and roles (role-level data is usually enough and lowers risk); photo and audio rules, including blurring of faces and name badges, storage location and retention; consent wording for any short interviews; and who the data controller and processor are. Your legal team should see this before fieldwork, not after.

Vendor-neutral checklist before you sign

1. Objective tied to a decision

The proposal states what will change depending on what is observed.

2. Written protocol and item definitions

You have seen them, or a redacted equivalent.

3. Calibration described

Method, result, and how disagreements were resolved.

4. Pilot included

At least one site, with time budgeted for form revision.

5. Sampling justified

Site selection logic, visit matrix, observations per site.

6. Supervision during fieldwork

Who checks completed forms, how quickly, and what happens when data looks off.

7. Consent and photo policy in writing

Aligned with KVKK and your internal policy.

8. Site-level data delivered

Observation-level or site-level file (Excel, CSV or SPSS) plus form and codebook, not only a presentation.

9. Findings separated from interpretation

What was observed versus what the researchers think it means.

10. Operations-ready output

Quick fixes versus structural changes, with the evidence for each.

Questions to ask before signing

  1. Who will the observers be, how many, and how were they recruited?
  2. How were they trained and what did the calibration exercise show?
  3. Can I see the observation form and protocol before fieldwork?
  4. How did you choose sites, days and time slots, and how many observations per site?
  5. Will observation be covert, announced or mixed, and who informs site staff?
  6. How are completed forms checked during fieldwork, and how fast?
  7. Will I receive observation-level data, the form and the codebook?
  8. If a site refuses access or an observer is identified, what is the fallback?

FAQ

How many sites and observations are enough? There is no universal number. It depends on how many groups of sites you want to compare and how much visits to the same site vary. Ask for planned observations per comparison group and a commitment to flag thin bases.

Should observation be covert or announced? Covert observation shows routine behaviour but limits what can be recorded and raises consent questions, especially for employees. Announced observation gives better access and cleaner consent but may change behaviour at first. Many projects combine both.

Can observers be our own staff? They can, but internal observers know the people observed and carry assumptions about which sites are "good". Protocol, calibration and supervision then matter even more, and the data should not identify individual employees to their managers.

What does field observation cost in Turkey? It depends on the number of sites, geographic spread, hours per site, travel, and how much training and supervision is built in. Request an itemised quote so a lower price can be traced to a specific cut, such as no pilot.

How is observation data combined with surveys or interviews? Observation shows what happens; interviews explain why. A common design pairs structured observation with short employee or customer interviews at the same touchpoints, then maps both against the written service standard, saying which frictions more than one source confirms.

How BigCat Research approaches this

BigCat Research is an independent research company based in Istanbul. Field observation sits in its operational experience and service quality module. The method set on the service page is a structured observation form with touchpoint coding, short interviews, photo notes and process tracing, and matching of findings with service standards and business impact. Outputs are a field observation and touchpoint report, a friction, risk and quick-win list, and an implementation brief for operations teams. Details: Field Observation Research.

This page makes no claims about BigCat's observer network, certifications or past clients. Apply the checklist above to BigCat's proposal and to every other proposal in the same way.

Primary sources

Author: Murat Akşit, BigCat Research. Last reviewed: 3 October 2026.