Skip to main content

Product Experimentation

Amplitude Experiment is our product that allows teams to remotely modify and configure the product experiences for unique audiences.

  • 114 Topics
  • 198 Replies
114 Topics
Amplitude Admin
Community Manager
Amplitude AdminCommunity Manager
published in Product Experimentation
Welcome to the product experimentation forum!

Amplitude Experiment is our product that allows teams to remotely modify and configure the product experiences for unique audiences This is where you can ask and get feedback on questions such as: ✅ Am I able to export the data I have in Experiment similar to Analytics?✅ How can I re-assign variants to my users?✅ Why are my users not getting targeted the right variant?✅ What happens if we merge projects in Experiment? Helpful resources: ❓Overview of Amplitude Experiment❓The experimentation workflow: from configuring to troubleshooting your experiment❓Amplitude Experiment key terms ❓Set up mutually-exclusive experiments❓Understand the Experiment Analysis view❓Advanced topics in Experiment: duration estimates, sequential testing for statistical inference, MDE and confidence intervals, randominzation, performance and scaling, sample ratio mismatches❓Amplitude Academy course: Getting Started with ExperimentStill have questions? Post them here! 

5381
T
T
Tuyen NguyenNew Member
asked in Product Experimentation
Web experiment does not run

I set up the first web experiment 2 days ago but now I don't receive any data for this experiment. Don’t know why? 

41
T
C
ccixdNew Member
asked in Product Experimentation
question regarding surveys SDK

Dear Support Team, We have added surveys to our product and tested them internally. However, we encountered some bugs and unexpected behavior where surveys cannot be snoozed or dismissed even after users have completed them. We have triple-checked the limits and settings of the surveys we created, but the issue persists. Could you please assist us in resolving this? 

50
C
N
NitishKumarNew Member
asked in Product Experimentation
What is the true conversion(Event level conversion) from "Select Song or Video" to "Play Song or Video", when the Song or Video is selected from "Main Landing Screen". Compare this conversion number by different "Sources" , in the same chart.

Hi, I feel my answer is a little close https://app.amplitude.com/analytics/demo/chart/hjqmr0zx, Is this correct, if not I need little assistance.

8942
H
V
VMGMNew Member
asked in Product Experimentation
Missing events make it impossible to trust funnel chart

Hello, I’m logging an event for every action my user does in onboarding. No actions are skippable and they are always in the same order. However, about 20% of the time there are a few missing events, meaning the user goes from Screen A → Screen B → Screen C but there are only events for Event A → Event C. These missing events are ruining my funnel charts, making it look like my conversion is much lower than it is.What can I do to fix this?

200
V
B
BKMNew Member
asked in Product Experimentation
After signout, Next set of amplitude events till the next login session are capturing under the previous logged out user event stream even after reset

To stop tracking events under the same UserID after logging out of the application used reset method.Analytics.shared().reset()With this change, a new session is initiated by closing the previous session, and the userID appears as "-" in User Properties. However, the events are still being displayed under the same old user's event stream. instead of displaying under the anonymous user. 

321
B
Experienced
M
Matthew DoNew Member
asked in Product Experimentation
How to pull the best blog post of the day based off most likes/comments

am working on a social video platform, and i use amplitude for my analytics. i made a segmentation graph charting the amount of blog posts we get uploaded per day, and i am trying to figure out how i can pull the blog post onto my dashboard? like what steps would i need to take or is this even possible

404
Saish Redkar
Expert
R
Raphael Rodrigues SilvaNew Member
asked in Product Experimentation
How to use A/B Tests with Amplitude Experiment for Server-Side Email Campaigns and Unauthenticated Users

I need to do an A/B test so that 50% of people receive a certain email and the other 50% receive another email. The email is sent on the server and I am using the experiment flag to decide which email the person should receive. Through email, these people have a link and when they click, they access the same page on my app. On this page there are some events that I want to monitor and compare the email conversion rate. Even though the person is not authenticated, I can know the email because this information exists in the url.In this case, how should I proceed?I have seen that when I fetch the flags providing the user_id it generates the [Experiment] Assignment event but if I don't provide the user_id it doesn't. What is the impact on the experiment of not having this [Experiment] Assignment generated?Do I need to verify the user in order for the experiment to work properly? I'm asking this because we verify the user after they logged in, but at this point the user is not logged in.I'v

761
liuaud
Peer Moderator
A
Ahmed zoheiryNew Member
asked in Product Experimentation
How to change experiment allocation/ targeting trigger ?

In the current experimentation UI in amplitude. we have the option to target all user or based on user property, but what we need is to target based on specific event.lets imagine that we have a funnel and Viewed landing page (first event in the funnel)  Amplitude normally assign users on variants before or after this event get fired  Clicked on product carousel  this is the event that we want to start assigning when it happen to include only in the experiment users who have done this event and the variants (treatment / control) should get exposed after firing this event.How to work on implementing solution 2. is there is a way from amplitude ui or it should be done using code ? and if with code, how to do it ?       

602
C
Team Member
L
Novice
lanceNovice
asked in Product Experimentation
Consistent bucketing across unique amplitude experiments

Hi there!We're working on configuring three experiments, one for each platform (web, ios and android) but we want the bucketing to be consistent across all platforms (i.e., a user that sees variant A on web should also see variant A on iOS and Android). If the rollout %, # of variants and variant weight all remain the same, could we apply the same bucketing salt string to all three experiments and expect consistent bucketing for all of our users in these experiments? 

3784
C
L
Laura JelenichNew Member
asked in Product Experimentation
iOS SDK manually changing the experiment variant on the client side

I’m working on an iOS app using the Amplitude SDK, and multiple developers are testing experiments. However, not all testers can access the Amplitude dashboard to change the assigned variants. To facilitate testing without using the dashboard, I’m looking for a way to override or manually set the variant for specific experiments directly on the client side.Does the iOS SDK currently support something like this?

764
liuaud
Peer Moderator
T
Tom HubenNew Member
asked in Product Experimentation
Delay on clicks

when i added the snippet to my web site. All the clicks starting adding 3 to 5 second delay.I assume i am doing something incorrect? for example a click to open Modal took 5 seconds. Take away Amplitude and instant

620
T
S
Novice
Son JoohoNovice
asked in Product Experimentation
Do Experiment and Session Replay work in Flutter platform?

Hi Experts, I would like to have experts’ advice how Experiment and Session Replay works in Flutter Platform.I couldn’t find any doc and SDK on this topic.Please give your advice. Cheers,Jooho

1872
L
J
Jiseong KimNew Member
asked in Product Experimentation
Why number of active user is same with new User

"Despite having retention, why are the numbers of new users and active users coming out the same?" 

340
J
M
milleryoonNew Member
asked in Product Experimentation
Visual Editor doesn't work

I was making an Experiment, But I can’t run visual editor.When I run visual editor, I met a screen like below.I supposed that it caused by Arc Browser, but I faced same screen at Chrome Browser.Can you help me to do experiment? Thanks.

630
M
L
Novice
lanceNovice
asked in Product Experimentation
Exposure tracking provider for NodeJS

Hi! We currently utilize the NodeJS SDK to handle server-side analytics and experiments. We were wondering if there are any plans to build an exposure tracking provider for NodeJS, similar to the JS SDK implementation

3055
I
J
Japser YooNew Member
asked in Product Experimentation
How to use reverse proxy for Experiment SDK

We have integrated the Experiment SDK (JavaScript) into our React application and are currently utilizing both the experiment and feature-flag functionalities.To fetch flags data, we initialize the SDK as follows:```javascriptthis.experiment = Experiment.initializeWithAmplitudeAnalytics(apiKey);await this.experiment.start();```Then, we retrieve the data using this.experiment.variant(variantKey).However, we are encountering an issue where the flags data is blocked by ad blockers or in certain regions, preventing us from accessing the data. To resolve this, we attempted to use a reverse proxy URL like below:```javascriptExperiment.initializeWithAmplitudeAnalytics(apiKey, { flagServerUrl: "proxy-url" });```Unfortunately, the SDK automatically appends `sdk/v2/flags` to each request, which makes it difficult to bypass the blockage. Ultimately, we are trying to achieve a setup similar to how we apply a reverse proxy to Amplitude event requests, where we can set a reverse proxy URL for flags

1230
J
Y
Yiji HuangNew Member
asked in Product Experimentation
Duplicated events with different ip addresses

Hi here, We find that sometimes we will receive duplicated events (same event id, same session id). Looks like the ips are different. The client time are the same but one client upload time has huge lag (several days to even two weeks)Is this expected behavior? It brings us some trouble counting user events, and we found about 7% of the events will be duplicated

1170
Y
N
Niharika ParabNew Member
asked in Product Experimentation
Multiple First Name/Last Name Values in Amplitude

We are experiencing an issue where Amplitude is showing multiple values for first names, last names, etc., although we are sending single values (first name, last name, etc.) through Rudderstack from our side.Assumption: We believe Amplitude might be retaining previous login details and appending them to the new data. The support replied by saying:Amplitude's user property is sticky, so once the custom user property is set for users, it will associate with all subsequent events unless explicitly unsetting them.Please help us out by letting us know how to unset users.

570
N
M
Michael McNicholasNew Member
asked in Product Experimentation
Does anyone have any experience using Amplitude Experiment and Google Tag Manager together?

Does anyone have any experience using Amplitude Experiment and Google Tag Manager together? Our goal is to have Amplitude Experiment set-up like a typical client side tool since it would be difficult to integrate our servers-side code due to some vendor issues. (long-term addressable, but not in the near term) So the basic thought is1> Fire Amplitude Analytics in GTM with the  “Amplitude Analytics Browser SDK” GTM community template. 2> Fire a script tag to initialize the Expierment. <script src=“https://unpkg.com/@amplitude/experiment-js-client@1.9.0/dist/experiment.umd.js”></script> 3> Create a Google Tag Manager Variable which is a custom javascript variable and captures which the experiments and variations the visitor is in. 4> Use those values in the GTM variables to execute or not, the code in our experiments. This is not pretty.  But I haven’t come across much content on Amplitude Experiment, since it is a newer product and not for the lower tier plans. a

4683
C
Team Member
N
Navneet YadavNew Member
asked in Product Experimentation
Amplitude Logger [Error]: Unexpected token 'F', "Fixed Resp"... is not valid JSON

In the given screenshot I’m using domain proxy /amplitude-ts URL is redirecting to https://api2.amplitude.com/2/httpapi but /amplitude-ts is giving 404 as you can see in the screenshot.  

600
N
H
Heejae LeeNew Member
asked in Product Experimentation
The 400 response code from Amplitude when Adjust sends a callback

We recently integrated Adjust, and Adjust is successfully sending callbacks to Amplitude. However, it appears that Amplitude is returning a 400 response. Could you provide any suggestions on how to resolve this issue?

790
H
E
eve weinbergNew Member
asked in Product Experimentation
What happens if i leave my javascript snippet after I end my A/B test?

Hello, I just ended by A/B test and decided to ‘roll back’ - not use the variant I tested.However, i can’t remove the javascript from the page header right now due to other reasons. Is this ok?Will any users be affected, or should I just remove the js script when I can and it’s not really that harmful.

913
eddie.gaona
Team Member
E
eve weinbergNew Member
asked in Product Experimentation
Why am i getting a 'URL is required' warning, when building my experiment

I can’t get past this step, when making my experiment: 

950
E
T
Tobias Bold TechNew Member
asked in Product Experimentation
Experiments Api - "Create" endpoint returns 500

Hey, when I try to create an experiment using the API I just get Error code 500 "Internal Server Error" as response. This happens I call the API from my retool application, but also when using curl from my local machine.My networking tab shows that I send this request:"request": {    "url": "https://experiment.amplitude.com/api/1/experiments",    "method": "POST",    "body": "{\"key\":\"test-api-call\",\"name\":\"Test\",\"description\":\"Try to create an experiment using the API\",\"projectId\":******}",    "headers": {      "Content-Type": "application/json",      "User-Agent": "Retool/2.0 (+https://docs.tryretool.com/docs/apis)",      "Accept": "application/json",      "Authorization": "---sanitized---",      "ot-baggage-requestId": "undefined",      "x-datadog-trace-id": "5857767016106236026",      "x-datadog-parent-id": "5648445407639375651",      "x-datadog-sampling-priority": "0",      "x-datadog-tags": "_dd.p.dm=-1",      "traceparent": "00-0000000000000000514af810bc03187a-4e634

960
T

Recent Tips & Tricks

Terms & ConditionsCookie settings