Android Surveys & Feedback

Supported on: Android SDK 1.0.33+

You can A/B test this survey with no extra code — create an experiment on it in the Console and the SDK serves the assigned variant automatically. See Servable Surface Experiments.

The surveys module lets you collect user feedback through in-app surveys. Surveys can trigger automatically based on events (like in-app messages) or be presented manually from your code.

How It Works

Surveys are configured in the Console with content, question types, and trigger rules. Like in-app messages, they can appear automatically when trigger conditions are met — no code required for trigger-based surveys.

Present a Survey Manually

AppDNA.surveys.present("nps_q1_2026")

The SDK uses the foreground Activity captured at configure time to launch its internal SurveyActivity. If the survey ID is unknown or the config hasn’t loaded, the call is a no-op and an SDK log line explains why.

Module Access

val surveys = AppDNA.surveys

Module Methods

Method	Signature	Description
`present`	`present(surveyId: String)`	Present a survey by ID
`setDelegate`	`setDelegate(delegate: AppDNASurveyDelegate?)`	Set a delegate for survey lifecycle callbacks

9 Question Types

Type	Display	Response
`nps`	0-10 numeric scale	Integer 0-10
`csat`	satisfaction scale, 3 / 5 / 7 wide	Integer 1-`scale`
`rating`	star, heart or thumb rating	Integer 1-`max`
`emoji_scale`	Emoji options	Selected emoji value
`yes_no`	Binary choice	Boolean
`single_choice`	Radio button options	Selected option string
`multi_choice`	Checkbox options	List of selected strings
`free_text`	Text input field	Free-form string
`likert`	numeric scale with optional anchor labels	Integer `min`-`max`

Question types and their content are defined entirely in the Console. The SDK renders them automatically based on the survey configuration.

scale is an accepted alias for likert, and scale_config for likert_config — both SDKs read either spelling, because questions were stored under the older one before likert became authorable. The Console writes the canonical names, so a survey created today uses likert and likert_config.A max above 10 is refused at authoring time: the scale draws as one un-scrolling row of fixed-width buttons, so a wider range would run off the edge of a phone. A min greater than max is read as the range the other way round rather than drawing nothing.

Asking for a store review

A review prompt is never filtered by how someone answered. Google’s policy is explicit — “Don’t ask the user any questions before presenting the rating button or card, including questions about their opinion” — and Apple’s App Review Guidelines say the same for the App Store. An app that screens users first risks rejection.So the SDK never offers a screener, and a survey cannot be configured as one: if any sentiment branch asks for a review, every respondent who finishes the survey is eligible, whatever they answered. The OS decides the rest — Google’s In-App Review API applies its own quota, and it deliberately does not report whether a prompt appeared.

Set prompt_review on whichever branch of follow_up_actions fits your product; it is the ASKING that is configured, not the audience. The respondent’s own branch still runs, so a detractor who lands on show_feedback_form sees the form as well.

Question configuration

Each type reads its own config object on the question. Everything is optional; an unset value takes the default below.

Type	Config	Keys
`nps`	`nps_config`	`low_label`, `high_label` — the anchors under 0 and 10
`csat`	`csat_config`	`scale` (3, 5 or 7 — default 5), `labels` (one per point), `style`
`rating`	`rating_config`	`max` (default 5), `icon` — `star` (default), `heart` or `thumb`
`single_choice` · `multi_choice`	`choice_config`	`options`, each `{ id, label }`
`emoji_scale`	`emoji_config`	`emojis` — the scale, in order, worst to best. Default 😡 😕 😐 😊 😍
`free_text`	`free_text_config`	`placeholder`, `max_length` (default 500)
`likert`	`likert_config`	`min` (default 1), `max` (default 5, capped at 10), `low_label`, `high_label` — the anchors under the first and last button
any	—	`required` (default `false`), `image_url`, `show_if`

The emoji scale is read in order. The first entry is the worst and the last the best, and that order is what sentiment routing uses — see Follow-up actions. A scale of any length works.

Conditional questions

A question can carry a show_if naming an EARLIER question and the answers that reveal it:

{
  "id": "why_not",
  "type": "free_text",
  "text": "What would have made it better?",
  "show_if": { "question_id": "would_recommend", "answer_in": [false] }
}

The condition is re-evaluated after every answer, so the question appears and disappears as the respondent moves. Two rules worth knowing:

Unanswered means hidden. If the question it references has not been answered yet, the conditional question is not shown.
A blank question_id means no condition. A half-configured show_if shows the question rather than hiding it — otherwise a survey could render with nothing in it at all.

When a survey appears

trigger_rules on the survey decides this; no code is needed for a triggered survey.

Key	Effect
`event`	the event name that arms the survey — it is checked against every event the SDK sees
`conditions`	`[{ field, operator, value }]` over that event’s properties. Operators: `eq`, `neq`, `gt`, `gte`, `lt`, `lte`, `contains`, `in`, `not_in`. `in` / `not_in` take a list (a single value is treated as a one-element list). An operator outside this set never matches — it is a configuration error, and a survey that never appears is easier to notice than one shown to the wrong audience
`frequency`	`once` (default), `once_per_session`, `every_time`, `max_times`. A value outside this set behaves as `once` on both platforms
`max_displays`	the cap for `max_times`
`love_score_range`	`{ min, max }` against a love score held on the device. No SDK computes one yet (the value is always 0), so a `min` above 0 never shows the survey: leave it unset. A survey skipped by this rule (or by `min_sessions`) says so in the debug log.
`min_sessions`	the survey waits until the user has had this many sessions
`delay_seconds`	how long after the event to present

Presenting with present(...) from your own code ignores all of this except the survey’s own status and audience.

Follow-up actions

When a survey completes, the SDK classifies the FIRST answer as positive, negative or neutral, and runs the matching branch of follow_up_actions.

`survey_type`	positive	negative	neutral
`nps`	9-10	0-6	7-8
`csat`, `rating`	≥ 4	≤ 2	3
`emoji_scale`	above the middle of the authored scale	below it	the middle
anything else	—	—	always `neutral`

A numeric type whose answer is NOT a number falls through to the emoji row, so a csat survey whose first question is an emoji scale is classified by that scale rather than coming back neutral. An answer that is neither a number nor a member of the scale is always neutral.

Sentiment is read from the survey’s survey_type, not from the question’s type. A survey left as custom whose first question is an NPS scale is always neutral, and its branches never run. Set survey_type to match the first question.

Each branch is { action, message }:

`action`	What the SDK does
`prompt_review`	asks the OS for a store review prompt
`show_feedback_form`	opens a built-in free-text form, titled with `message`
`trigger_winback`	signals only — emits `survey_followup_winback` with the `message`, and your app launches the campaign
`dismiss`	nothing

A review prompt is never filtered by how someone answered. Apple’s App Review Guidelines and Google Play’s In-App Review policy both forbid asking the user a question — including one about their opinion — before presenting the rating prompt, and both forbid filtering who sees it by the answer.So: if any branch asks for prompt_review, every respondent who finishes the survey is eligible, whatever they answered, and the OS quota decides the rest. Putting prompt_review on one branch configures the ASKING, never the audience. The respondent’s own branch still runs, so a detractor routed to show_feedback_form sees the form as well.

Appearance

appearance on the survey:

Key	Values	Unset
`presentation`	`modal`, `bottom_sheet`, `fullscreen`	`modal`
`dismiss_allowed`	whether the respondent can close it	`true`
`show_progress`	a progress indicator across the questions	`false`
`corner_radius`	points	the platform default
`question_text_style`	a text style object for every question	inherited
`option_style`	an element style object for every choice row	inherited
`theme`	the colours and media below	—

appearance.theme may be a flat object, or { "light": { … }, "dark": { … } } — in the second form the active variant is sparse-merged over the other, so a dark theme only needs the keys it changes.

`theme` key	What it colours
`background_color`	the sheet
`text_color`	question text
`accent_color`	the selected option, the filled rating icon, the NPS selection
`button_color` · `button_text_color`	the submit button
`font_family` · `question_font_size` · `font_weight` · `text_align`	typography (`font_weight`: `normal`, `medium`, `semibold`, `bold`)
`gradient` · `button_gradient`	a gradient behind the sheet or the button

Plus the rich-media keys in the next section.

AppDNASurveyDelegate

All 3 methods on this delegate fire from SurveyManager alongside the analytics events. Register your delegate via AppDNA.surveys.setDelegate(...).

onSurveyPresented(surveyId) — fires when the survey view appears (alongside the survey_shown event).
onSurveyCompleted(surveyId, responses) — fires when the user submits the final question. responses is a List<SurveyResponse> (one entry per answered question with questionId + the raw answer), suitable for forwarding to your own analytics or CRM.
onSurveyDismissed(surveyId) — fires when the user closes the survey before completing it (back button, swipe-to-dismiss, outside tap).

Implement the delegate to capture survey responses and lifecycle events:

interface AppDNASurveyDelegate {
    fun onSurveyPresented(surveyId: String)
    fun onSurveyCompleted(surveyId: String, responses: List<SurveyResponse>)
    fun onSurveyDismissed(surveyId: String)
}

SurveyResponse

Property	Type	Description
`questionId`	`String`	Identifier for the question
`answer`	`Any`	The raw answer — `Int` for NPS/CSAT/rating/Likert, `String` for free-text/single-choice, `List<String>` for multi-choice, `Boolean` for yes-no
`metadata`	`Map<String, Any>?`	Optional per-question metadata (question type, timing, etc.)

Cast answer to the type matching the question (see Question Types).

Example Implementation

import ai.appdna.sdk.AppDNA
import ai.appdna.sdk.AppDNASurveyDelegate
import ai.appdna.sdk.SurveyResponse

class FeedbackHandler : AppDNASurveyDelegate {

    override fun onSurveyPresented(surveyId: String) {
        Log.d("Survey", "Survey shown: $surveyId")
    }

    override fun onSurveyCompleted(surveyId: String, responses: List<SurveyResponse>) {
        for (response in responses) {
            when (val a = response.answer) {
                is Int -> {
                    // NPS / CSAT / rating / likert
                    if (response.questionId == "nps_question") {
                        if (a >= 9) showReferralPrompt()    // Promoter
                        else if (a <= 6) showSupportLink()  // Detractor
                    }
                }
                is String -> Log.d("Survey", "Feedback: $a")
                is List<*> -> Log.d("Survey", "Multi-select: $a")
                is Boolean -> Log.d("Survey", "Yes/No: $a")
            }
        }
    }

    override fun onSurveyDismissed(surveyId: String) {
        Log.d("Survey", "Survey dismissed: $surveyId")
    }

    private fun showReferralPrompt() { /* ... */ }
    private fun showSupportLink() { /* ... */ }
}

AppDNA.surveys.setDelegate(FeedbackHandler())

Rich Media

Survey questions support rich media content configured in the Console:

Key	On	What it does
`questions[].image_url`	a question	an image above the question text
`theme.intro_lottie_url`	the survey	a Lottie played once before the first question
`theme.thankyou_lottie_url`	the survey	a Lottie on the thank-you screen
`theme.thankyou_particle_effect`	the survey	a particle effect on the thank-you screen — see Rich Media for its keys
`theme.thank_you_text`	the survey	replaces “Thank you!”; supports template interpolation
`theme.blur_backdrop`	the survey	`{ radius, tint, saturation }` behind the sheet
`theme.haptic`	the survey	`{ enabled, triggers: { on_option_select, on_button_tap, … } }`

Option icons are not authorable. choice_config.options[] accepts an icon through the API and the survey editor has no control for it, so it cannot be set from the Console; it renders on iOS and not on Android. Treat choice options as { id, label } until that is resolved.

See the Rich Media guide for the Lottie, particle and haptic key tables, and for which formats each platform plays.

Auto-Tracked Events

Every event whose name begins survey_ carries survey_id. The two feedback_form_* events are emitted by the form itself and carry only the fields named below.

Event	Trigger
`survey_shown`	A survey is displayed
`survey_question_answered`	Each individual answer, as it is given
`survey_completed`	The final answer is submitted. Carries `survey_type` and the full `answers` array
`survey_dismissed`	Closed before completing
`survey_followup_prompt_review`	A follow-up asked the OS for a store review prompt. Carries `sentiment`
`survey_followup_feedback_form`	A follow-up opened the built-in feedback form. Carries `sentiment`
`survey_followup_winback`	A follow-up signalled a win-back. Carries `sentiment` and `message`
`feedback_form_submitted`	The feedback form was submitted. Carries `feedback`
`feedback_form_dismissed`	The feedback form ended without a submission. Carries `reason`
`survey_feedback_submitted`	Android only, and legacy: the same submission as `feedback_form_submitted`, kept for existing consumers. Build cross-platform funnels on `feedback_form_submitted`

feedback_form_submitted / feedback_form_dismissed come from the survey follow-up form — the one follow_up_actions opens. There is no separate standalone form.reason on the dismissal is user (the Cancel button, a back press, or a tap outside the dialog) or presenter_released (the host navigated away and the form went with it — a form the user never saw, which would otherwise be counted as one they closed).

Full Example

import ai.appdna.sdk.AppDNA
import ai.appdna.sdk.AppDNASurveyDelegate
import ai.appdna.sdk.SurveyResponse

class FeedbackManager : AppDNASurveyDelegate {

    init {
        AppDNA.surveys.setDelegate(this)
    }

    /** Present NPS survey after a key moment */
    fun askForFeedback() {
        AppDNA.surveys.present("nps_q1_2026")
    }

    // MARK: AppDNASurveyDelegate

    override fun onSurveyPresented(surveyId: String) {
        // Survey is visible
    }

    override fun onSurveyCompleted(surveyId: String, responses: List<SurveyResponse>) {
        // React to feedback
        val nps = responses.firstOrNull { it.questionId == "nps_question" }
        val score = nps?.answer as? Int ?: return
        when {
            score >= 9 -> showReferralPrompt()
            score <= 6 -> showSupportLink()
        }
    }

    override fun onSurveyDismissed(surveyId: String) {
        // User dismissed without responding
    }

    private fun showReferralPrompt() { /* ... */ }
    private fun showSupportLink() { /* ... */ }
}

Surveys are created in the Console under Feedback > Surveys. Trigger-based surveys appear automatically — manual presentation with present() is for precise control over timing.

​How It Works

​Present a Survey Manually

​Module Access

​Module Methods

​9 Question Types

​Asking for a store review

​Question configuration

​Conditional questions

​When a survey appears

​Follow-up actions

​Appearance

​AppDNASurveyDelegate

​SurveyResponse

​Example Implementation

​Rich Media

​Auto-Tracked Events

​Full Example

How It Works

Present a Survey Manually

Module Access

Module Methods

9 Question Types

Asking for a store review

Question configuration

Conditional questions

When a survey appears

Follow-up actions

Appearance

AppDNASurveyDelegate

SurveyResponse

Example Implementation

Rich Media

Auto-Tracked Events

Full Example