Short answer
The first round of interviews often raises questions nobody in it can answer. People say "ask the night inventory controller" or "only finance knows how that approval works". Following up with those colleagues, and going back once to people whose answers left gaps, fills in the picture.
Set limits before you start:
- who may be contacted;
- how many times;
- how people can decline.
Tell the organiser before anyone new is contacted. Invite each referred colleague once, without chasing. When someone does not reply, record what remains unknown.
Why the first list is never complete
The people in a study are usually chosen by role, or because the sponsor thought of them. The person who actually knows how a workaround came about, or who reconciles the numbers at night, is often not on the list. Interviews reveal them. When three people independently name the same colleague, that colleague's account probably matters.
Researchers call this referral or snowball sampling. It reaches people the organiser did not know to ask, and it has a known weakness: it follows existing networks, so people outside them stay unheard. Keep that limit visible in the findings.
Two kinds of follow-up
| Follow-up | When | What to ask |
|---|---|---|
| Back to someone who already answered | Their answer raised a new question or contradicted someone else's | One or two specific points, referring to what they said |
| A colleague they mentioned | They named someone who knows a step, a decision or a history | The specific process or event they were named for |
Keep both short. A follow-up is not a second interview; it is a question about a gap.
Set limits at the start
| Limit | Example |
|---|---|
| Who may be contacted | People in the study, and colleagues they name within the organisation |
| How often | Each person asked again at most once in the study |
| Notice | The organiser is told the day before anyone is contacted, and can remove someone or pause |
| Chasing | A referred colleague receives one invitation and is never chased |
| Stopping | Anyone can stop follow-ups at any time |
| Scope | Follow-ups stay within the topics approved for the study |
These limits protect people's time and the study's credibility. A referred colleague who gets three reminders about a study they never agreed to join will remember it, and so will their colleagues.
Do not turn referral into reporting
Asking "who else would know about this process?" is different from asking "who is responsible for this problem?". The first finds expertise; the second invites blame. Phrase referral questions around knowledge of the work. Do not tell the referred colleague who named them unless that person agreed.
A fictional example: month-end close
A fictional manufacturing group asks its finance team why month-end close takes eight working days. Eight people answer:
- five describe waiting for intercompany balances from the Spanish subsidiary;
- two describe manual journal corrections;
- one did not know.
Three people name the same person as the only one who understands the reconciliation: the group's intercompany accountant, who was not on the list. One of the five also says the Spanish balances arrive late "because of the bank feed", but nobody else mentions a bank feed.
The organiser is told the next day's plan: invite the intercompany accountant, and ask the person who mentioned the bank feed one follow-up question. The organiser removes nobody and lets it go ahead. The accountant explains that the balances wait for a currency revaluation that runs only after a local deadline. The person asked about the bank feed clarifies that it was a problem last year and has since been fixed.
The finding changes from "Spain is late" to "the close waits for a revaluation step scheduled after a local deadline". That is something the group can reschedule.
Record what stays unknown
Not everyone will reply. A referred colleague may be on leave, may not want to take part, or may know less than their colleagues thought. Record the gap: "The intercompany process was described by three people; the person they named did not take part." An honest unknown is better than a finding stretched to cover it.
For knowledge that sits with only a few people, see capturing employee knowledge across thousands of people and turning interviews into training.
Where Lontra fits
This is what Deep Study does. When the first answers leave gaps, Lontra reads what people said and spots what is still unclear. It then sends short follow-ups to people who already answered and to colleagues they suggest, within the limits you set at launch.
- It emails you the day before, so you can remove someone or pause.
- Colleagues they suggest get one email and are never chased.
- People can be asked again once more in a study; then the findings are final.
- Participants can stop follow-ups at any time.
- A follow-up with someone you already heard does not use another interview; a new colleague does.
Deep Study follows the people your participants know. It does not find people outside those networks, so check who is missing.
Follow the evidence, within limits
Referral follow-ups are how an interview study reaches the person who actually knows. Set the limits before the first invitation, keep to them, and record what the follow-ups could not reach.
Frequently asked questions
- What is snowball sampling in interviews?
- Finding participants by asking people already interviewed who else would know about the topic. It reaches people the organiser did not know to ask, but it follows existing networks and can miss people outside them.
- Should referred colleagues be told who suggested them?
- Only if the person who suggested them agreed. The invitation can say they were suggested as someone who knows the process, without naming who suggested them.
- How many times should someone be asked to take part?
- Set a limit before the study starts. A considerate limit is one invitation for a referred colleague, with no reminders, and at most one follow-up for someone who already answered.