What does source file search time reveal about a GTM research folder design?

A protocol for measuring how long analysts spend locating source files and what the pattern says about folder structure and naming.

What does source file search time reveal about a GTM research folder design?

Question and scope

This protocol measures the time analysts spend locating source files in a GTM research folder. The scope is internal search for a file the analyst has used before or knows exists. It excludes open-ended browsing for new information, which is a different activity with a different purpose. The question matters because search time is invisible in most reports yet it compounds across every brief and every refresh.

The unit is one search episode that ends when the analyst opens the intended file or gives up. A search that ends with a substitute file is recorded separately, because the substitute may change the evidence. The protocol does not try to measure the value of the research itself. It measures the cost of the folder design that surrounds the work.

Measure a search episode

Ask analysts to note the start time, the end time, and the path they took for a sample of searches. A path can be a folder sequence, a search query, or a request to a colleague. Record whether the file was found and whether the analyst had used it before. Keep the note short enough that it does not become a second job.

Use a small timer or a browser extension only if the team already accepts that level of monitoring. A lightweight self-report is often more sustainable than precise tracking that people resent. If self-report is used, ask for the estimate immediately after the search rather than at the end of the day. Delayed estimates drift toward round numbers.

Ask the analyst to record the first place they looked, not only the successful path. The first place reveals the team's mental model of the folder, which is often different from the designed structure. A mismatch between the first look and the final path is a strong signal for a naming or placement fix. Record it beside the duration.

Sample searches without disrupting the team

Sample across a full week rather than a single busy day. Include searches from different analysts and different research topics so the result is not dominated by one project. Aim for a sample large enough to show a pattern, and record how many searches were logged versus how many were requested. A low response rate is itself a finding about the measurement burden.

Do not ask analysts to change how they work during the sample. A special folder or a temporary naming scheme would measure the intervention rather than the current design. Tell the team the purpose and how the results will be used. People report more accurately when they know the goal is to fix the folder, not to judge their speed.

Explain how the sample will be stored and who will see it before collection starts. Analysts are more willing to report slow searches when the data is used for design rather than performance. Do not attach names to the published result. If a manager needs to follow up on an access problem, do it privately.

Separate search from browsing

A search has a known target, while browsing has a question but no specific file in mind. Mixing the two inflates the search time and leads to the wrong fix. Ask the analyst to state the target before starting, even if the target is a category such as the latest packaging capture. If the target is vague, classify the episode as browsing and exclude it from the search measure.

Record the outcome of each search as found, found with help, substituted, or abandoned. A search that required help points to a naming or permission problem. A substituted file may indicate that the original was missing or that the analyst did not trust it. The outcome categories turn a duration into an actionable pattern.

Include the searches that ended with a request to a colleague. Those episodes are often the most expensive because they interrupt two people. Record the colleague's role and whether they found the file quickly. A pattern of helped searches points to a knowledge distribution problem rather than a folder problem.

Compare folder structures

If the team is considering a new folder structure, run the sample before and after the change with comparable tasks. Keep the tasks similar so the comparison measures the design rather than the difficulty of the topic. Record the number of clicks, the number of queries, and the time to the target. A structure that looks tidy in a screenshot may still be slow in use.

Compare structures by task type rather than only in total. A topic-based folder may be fast for finding a market note and slow for finding a source export. A project-based folder may do the reverse. The comparison should show which tasks improve and which ones get worse, because most reorganizations trade one kind of search for another.

Keep the task list identical between the two measurements. If the second round uses harder tasks, the comparison will show a worsening that comes from the test rather than the folder. Store the task list with the results so the next reviewer can rerun it. A stable test is what makes a small sample useful.

Interpret the distribution

Report the median and the long tail rather than only the average. A few very long searches can dominate an average while most searches are quick. The long tail is where the design causes the most frustration and where a small fix may help. Show the share of searches that took longer than a defined threshold and the reasons they did.

Look for repeated targets. If several analysts struggle to find the same file, the problem is the file's location or name rather than the searchers. If the slow searches are spread across many unique files, the problem may be the overall taxonomy. The pattern determines whether to fix one entry or redesign the structure.

Compare search time with file age and file owner. Older files without a named owner may be harder to find because no one has maintained their location. A file that is frequently sought but rarely updated may need a stable home and a review date. The pattern should guide both naming and ownership fixes.

What to change first

Start with the repeated targets and the searches that needed help. A short list of the ten most-sought files can guide a naming and pinning fix without a full migration. Add a stable shortcut for the files used in every research cycle. These changes are cheap and can be measured again in a week.

Only consider a full folder migration after the cheap fixes have been tested. A migration is expensive and often moves the problem rather than solving it. If the team does migrate, keep the old paths as redirects or index notes for at least a quarter. The measure should continue after the change so the team can see whether search time actually fell.

Limits and sources of error

Self-reported time is approximate and may be rounded. Analysts who know they are being measured may search more deliberately or ask for help sooner. The sample may also miss searches that were abandoned before they were logged. Report the collection method and the response rate so the reader can judge the quality of the estimate.

The cited public sources describe research practice, business survey methods, and information protection. They do not provide a search-time benchmark for this folder. Use the result as a local baseline and compare it with the team's own later periods. A small sample can show a direction, but it should not be presented as a precise cost saving.

Keep the measure alive

Repeat the sample after the first round of fixes rather than relying on a single study. Use the same task types and the same outcome categories so the comparison is meaningful. A short quarterly sample is usually enough to catch a regression. The measure should become part of the folder's maintenance, not a one-time project.

Publish a short note with the baseline, the change, and the next review date. Tell the team which files were pinned or renamed as a result. Visible follow-through makes future requests for search data easier to get. It also shows that slow searches are a design issue the team is willing to fix.

Related GTM research