Creating and Running Surveys

Pertento runs on-site surveys alongside experiments. Three instruments are supported: NPS, CSAT and CES.

The four steps

The builder asks four things, in the order the decisions actually happen:

• Instrument: which of the three you are running, plus an internal name for your library. The name is never shown to a visitor, so it can be as specific as you need.
• Questions: the scored question and its optional follow up.
• Appearance: how the widget looks on your site.
• Trigger: when it should fire.

Every step opens with the right answer already in it. Choosing an instrument fills in the standard question wording, the correct scale and a sensible follow up, so a survey you do not touch further is still a correctly constructed one. A live preview sits beside the form, so the effect of a colour, a corner radius or a longer question is visible while you are deciding rather than after you publish.

Question types

Three kinds of question are available: a score, free text, and single choice.

When a survey fires

Triggers are page view, a delay, exit intent, scroll depth, and a custom event. The widget can sit in one of four positions on the page, and frequency capping limits how often the same visitor is asked.

Sentiment on written answers

Each written answer is classified as positive, neutral or negative, and the response feed filters on that classification. It is a per answer sentiment reading rather than a clustering of what people wrote about.

Surveys inside experiments

A survey running while an experiment is live records which variant of which experiment each respondent was in. That single fact is what makes running both in one platform worth doing.

An experiment can tell you that variant B converted better. It cannot tell you why, and the why is what decides what you test next. Reading the same question per variant usually can. It works the other way too, and that is the case worth planning for: a variant that wins on conversion while scoring worse on satisfaction is a warning rather than a win, and it is invisible to an experiment reading conversion rate alone.

The attribution is recorded at the moment of answering rather than inferred later, and the reporting counts only respondents who were genuinely enrolled in the experiment you are asking about.

What it costs the page

Survey code is a separate bundle from the experiment runtime, and a website with no survey running never requests it at all. When it is requested it is requested last, after the page has been revealed and after measurement has been requested, so no survey code path can leave a visitor looking at a hidden page and no survey failure can cost you a conversion event.

Surveys can also be switched off from the page itself, which is the escape hatch you want on a live site when something needs to stop immediately.

Notes

• The survey name is internal: no visitor ever sees it.
• Sentiment is per answer: it classifies each response rather than grouping responses into themes.
• A survey inside an experiment needs no extra wiring: the variant is recorded automatically for anyone enrolled.