Real-time personalization that reacts to this visit.
Signals tells your recommender, search ranker or app what the customer is doing right now, in 6ms in-region, so the next thing they see fits this visit.
Below is your own visit to this page, computed by Signals as you read.
Signals, about you
connectingSignals is computing your first attributes from the events this page just sent.
Every highlighted part comes from attributes Signals computed from the behavioral data your browser sent during this visit. Open the Product menu above and watch the sentence change.
Personalization lags behind the visit.
The page decides from what the visitor did last time. What they do now reaches their profile too late to change this visit.
Every click changes what your app knows.
Signals turns behavioral data into attributes and serves them to your app in 6ms. You can also define triggers that fire when a condition is met, such as a cart left idle.
Step by step: Personalize a travel booking flow →The background: Customer profiles and identity, explained →
- Braze
- Conductrics
- Algolia
- json-render by Vercel
- Snowflake
- BigQuery
One visit, start to finish.
Pick your industry. Each moment is one place a team reads Signals, and what their app did with it.
- Arrives
A personalized homepage
entry_category: runningThe homepage leads with running, before they sign in. - Searches
Personalized search results
searches_this_visit: trail, waterproofYour ranker puts waterproof trail shoes first on the next query. - Browses
Real-time recommendations
category_this_visit: runningThe carousel swaps to running gear after the second click, not tomorrow. - Checks delivery
Delivery details moved up the page
delivery_checks: 2The page moves the delivery promise up while they are deciding. - Hesitates
A cart reminder before they leave
Triggers to Braze, Kafka and webhookscart_idle: 40sA trigger fires, and your app shows a reminder.
- Lands
A personalized front page
entry_topic: climateThe front page leads with climate for this reader. - Reads
Article recommendations
topics_this_visit: climate, energyThe next slot follows what they opened, not last week's segment. - Keeps reading
Paywall timing based on engagement
articles_this_visit: 4Your paywall rules see engagement now, not a fixed article count. - Stays
A/B tests that target engagement
engaged_seconds: 310Your A/B testing tool targets readers who are engaged in this visit.
- Signs up
Onboarding that shows the next step
setup_steps_done: 2The in-app guide shows the step that is next. - Explores
Feature flags based on usage
opened_reports: trueYour flag tool turns on the new report builder for users who just opened reports. - Hits a limit
An upgrade prompt at the limit
seat_limit_hit: trueYour app offers the upgrade while they are inviting a teammate. - Gets stuck
Help when they get stuck
Triggers to Braze, Kafka and webhooksfailed_imports: 3A trigger fires, and your app opens help on the import screen.
One read, before the page renders.
The page sends the tracker's session id to your backend. Your backend reads the storefront service and hands the values to whatever picks the modules. The credentials stay on the server.
npx plugins add snowplow/skillsimport { Signals } from "@snowplow/signals-node";
const signals = new Signals({
baseUrl, apiKey, apiKeyId, organizationId,
});
const shopper = await signals.getServiceAttributes({
name: "storefront",
attribute_key: "domain_sessionid",
identifier: sessionId,
});
// Your selector keeps its rules. It now sees this visit.
const modules = shopper.showing_price_hesitation
? ["delivery_promise", "reviews", "similar_for_less"]
: ["upsell_rail", "reviews"];Add Snowplow Signals to this app so the product page reacts to this visit. Add the Snowplow browser tracker if there is none. Define an attribute for repeat views of one product in a session, publish it in a service named storefront, and read it with getServiceAttributes in the handler that picks the page modules. Leave the selection logic as it is, so the only thing that changes is what it can see.
The read returns the attributes keyed on the session. Cart attributes take the cart id as the identifier, in a second call or a batch read.
The feature: Real-time attributes →Full reference: Retrieve attributes in your application →
Change the page while they are still on it.
To read attributes, p50 in-region
Event in to attribute updated
Streaming infrastructure for your team to run
Frequently asked questions
What is real-time personalization?
Changing what a customer sees based on what they do in this visit, not on a segment computed overnight. Signals supplies what they are doing now. Your recommender, search or page logic decides what to show.
Do I need an event stream already?
No. Collection is part of Signals. Add a Snowplow SDK to your web, mobile or server app and events start flowing, validated against your schemas. If you already run Snowplow, Signals reads the pipeline you have.
Do we need to replace our recommender or search?
No. Signals feeds the recommender, ranker or rules you already use with what happened in this visit. If you need a packaged recommendation model, this is not it.
How do I target an A/B test on what someone did this session?
Read the attributes where your app calls your testing or feature flag tool, and pass them as targeting properties. The test then targets what someone did a minute ago.
Does it work before they sign in?
Yes. Attributes key on the session from the first event, so an anonymous visitor gets a relevant page. Once they sign in, identity resolution links this device to the customer they already are, so later reads include what they did on their other signed-in devices.
Where does it run?
Managed SaaS, or a private deployment in your own AWS or GCP account. Same architecture either way, so what you evaluate is what you run.
Watch an attribute change on your own traffic today.
Add the SDK, define showing_price_hesitation or any behavior you care about, then click around your own product and watch it change, the same way the panel above changed for you. The free tier runs on the real engine: no card, no sales call.