Creating and Running Surveys

Pertento runs on-site surveys alongside experiments. A survey asks a visitor a few short questions right on your own site, one at a time, and the answers sit next to your experiments. This guide is the overview; each part has a guide of its own, linked below.

Four survey types

• NPS, CSAT and CES are standard instruments. Each comes with its own scale and standard wording, and its score question always comes first.
• Custom is your own mix of questions, with no single headline score.

See Building a Survey: Types, Questions and Preview.

Questions and texts

A survey asks up to five questions, of four types: score, single choice, multiple choice and text. Any question can be optional, except the score question of an NPS, CSAT or CES survey. You can also write the button texts, the thank-you message and the words at each end of a scale yourself. See Writing Survey Questions, Options and Texts.

The five steps

The builder has five steps: Template, Questions, Targeting, Trigger and Theme. You can jump between them at any time, a tick marks each finished step, and a live preview sits beside the form.

Who sees it

By default a survey can show to every visitor on every page. The Targeting step narrows that by page address, by device, to a share of visitors, and to the visitors in one experiment or in chosen variants of it. See Choosing Who Sees a Survey.

When it appears

Each survey has one trigger: on page view, after a delay, on exit intent, on a dataLayer event or at scroll depth. See Choosing When a Survey Appears. Where it sits and how it looks are covered in Survey Appearance: Position, Colour and Corners, and frequency limits stop the same visitor being asked too often, see How Often a Visitor Sees a Survey.

Going live

Starting a survey marks it as running straight away, but it reaches your site at the website's next publish, just like an experiment. See Starting, Ending and Managing Surveys.

Reading the answers

The Results tab shows the score, every question's results with every scale point and option, responses over time, and a feed that keeps each visitor's answers to the survey together. See Reading Survey Results.

If you switch it on for a survey, each written answer is classified as positive, neutral or negative, and the responses feed filters on that classification. It is off by default and uses AI credits. See Sentiment Analysis on Survey Comments.

Surveys inside experiments

Each answer records which variant of which experiment the respondent had seen. That 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. It works the other way too: 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.

Tie a survey to an experiment and only that experiment's visitors are asked. Its results then open split by variant, and any other survey can be split by any experiment its visitors were in. See Survey Results by Experiment Variant.

What it costs the page

Survey code is a separate script from the experiment runtime, about 6 KB compressed, 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 sent, so no survey code path can leave a visitor looking at a hidden page and no survey failure can cost you a conversion event. See What Surveys Collect, and What They Cost the Page.

For testing, surveys can be switched off in a single browser with ?pertento-surveys=off. To stop a survey for every visitor, end it and publish the website.