Read your results

There are two places to read answers, and they answer different questions.

• An eflow’s Feedback tab — how this one eflow performed, in aggregate: question health and respondent ratings, not individual answers. Use it when you want to know whether the eflow itself is working.
• Responses — every answer across every eflow. Use it when you are looking for a particular respondent, reading across conversations, or want to open one conversation and read it.

Open a row on Responses and you get both readings of that conversation — the Feedback tab does not drill into a single response. Transcript is what was actually said, turn by turn, and it is the tab that opens first. Answers is the same conversation reduced to one value per question — a choice question as the option that was picked, a rating as a number — which is the form that leaves for your own systems.

If you sent this person a personalized link and attached details you already held about them, the Answers tab opens with Known before the conversation — those details, above the answers and kept apart from them. They are what you told us, not what the respondent said. The assistant never sees them, so it still asks the questions they cover, and the answer can disagree with what you sent. When it does, neither one is broken: it usually means your own records are out of date. The same details come back in the exported file, in a Prefill column of their own.

Language says which language the conversation was held in — in the panel beside Status and Answers, in the exported file, in the completion webhook, and in the connected Google Sheet. It does not have its own column on the Responses list: most conversations never carry a detected language at all, so the list keeps its columns for what every row has. We work the language out from the respondent’s first typed reply, so a conversation answered entirely by tapping options reads Not detected rather than guessing. It is not the language the eflow is set up in: knowing an eflow opens in German tells you nothing about which respondents answered in it.

Status says how far a conversation got, and one of the four words means something narrower than it sounds.

• Completed — the respondent reached the end.
• In progress — they have not finished, and the window in which they could come back and pick up where they left off is still open. That window is 48 hours by default, so a conversation someone walked away from yesterday still reads In progress today. It does not mean anyone is typing right now. Check Last activity for that.
• Abandoned — they stopped answering and the window has since closed. Nothing more is coming.
• Expired — rare. Reserved for a conversation that was poked once more after its window shut. Most tenants never see it, so the filter only offers it when you have one.

Answered shows how many questions they gave you out of how many they were actually asked — 3 / 6. The second number counts every question in the eflow that everyone gets, plus any follow-up questions this particular respondent was routed into. A branch they never entered is not a question they failed to answer, so it is not in their total, and two people can honestly have different denominators on the same eflow.

A conversation only exists once someone opens the page, so some rows are page opens where nobody answered anything — a link preview, a crawler, or you checking your own eflow. Answered only is on by default so the count on this screen means people who told you something; the Include filter turns them back on when you want to see them, and the export matches whatever the screen shows.

At the end of the row, Respondent says who answered, when we know. A name with a check mark next to it arrived through a personalized link, which means we put that name there ourselves and the person answering could not have changed it. A name without a check was sent to us by the page the conversation ran on — it is often right, and we cannot confirm it, so the drawer says so rather than letting the two look alike. No name at all means the conversation was opened from a link anyone could use. A help button next to the column heading repeats this on the screen itself.

Ref is a separate, last column: whatever tag was pasted onto the plain hosted link as ?aq_ref= — a mail-merge id or a campaign name, say. The campaign tag on a QR code lands in the same column, which is how a scan of a printed poster is told apart from a click on the link. It is never a name and never checked against anything, so it carries no check mark either way; a help button next to the heading says so. The same value comes back in the exported file, in a Ref (unverified) column of its own.