Grow conversion and average order value with ML across the entire customer journey. Data from every sales channel in one CDP.
| User | Segment | AOV | Activity | Trend |
|---|---|---|---|---|
| Anna M. | VIP | 14 200 ₽ | 3.2× / mo | |
| Igor S. | At risk | 5 400 ₽ | 0.3× / mo | |
| Maria D. | New | 2 100 ₽ | 0.8× / mo | |
| Dmitry K. | Loyal | 8 900 ₽ | 2.1× / mo |
The store does not answer individual needs. UX and the customer journey ignore cross-channel history, status, the current funnel stage and other behavioral data.
Segmentation exists in the CRM but is never applied on the site. That breaks the omnichannel journey: the offers and content in email campaigns differ from what the landing page shows.
The store does not adapt to real-time signals — scrolling, repeat visits, dwell time on a product card. Without dynamic product offers, high-margin sales are lost at the moment of decision.
Gravity Field tracks every action in real time: views, add-to-carts, clicks, time on page, traffic source. Online and offline data merge into a single profile with no more than 50 ms of latency.
// core.js listens to on-page behavior
window.GF = window.GF || {};
GF.section = "6315dda2f1c33ba4cd04655b";
// Event → ML recalculates the segment
GF.API("event", {
name: "add_to_cart",
data: { sku: "sku_9281", price: 12990 }
});
// returns the segment in < 50 msThe platform builds up to 25 base segments automatically: new users, churn risk, premium customers, discount-sensitive shoppers, regular buyers. No analyst and no manual rule setup.
A machine learning model analyzes hundreds of signals and assigns the user to the right segment in milliseconds. The segment updates dynamically within the current session.
The segment becomes available instantly on the site, in the app, on digital screens in stores, in AI assistants, in your email platform and in the contact-center interface. One consistent experience at every touchpoint.
Research Online, Purchase Offline is the revenue your digital work generates in stores and nobody attributes to it. Promo codes, call tracking, offline conversion import and…
Three or four behavioural segments that cover 100% of traffic beat a hundred narrow ones. A step-by-step guide to defining segments, testing personalization on each, and…
Turn UTM parameters from an analytics label into a personalization trigger: five steps to make the landing page mirror the ad, with a walkthrough in the…
Five levers that make paid traffic pay off — without raising the budget: the first 10 seconds, purchase barriers, returning users, recommendations and A/B testing.
How churn prediction models are built and validated, five retention mechanics they unlock, and how to check with an A/B test that retention costs less than…
A web script, a server-side API, an MCP server for AI agents and native mobile SDKs — the actual methods from our documentation.
A single script core.js on every page — the platform collects behavior and returns the user segment. No backend development; SPA and SSR supported.
<link rel="preconnect" href="//cdn-01.gravityfield.ai">
<script>
window.GF = window.GF || {};
GF.section = "YOUR_SECTION_ID";
GF.pageContext = { type: "CART", data: ["XD-154"] };
</script>
<script src="//cdn-01.gravityfield.ai/core/core.js"></script>Request campaigns and the segment from any backend via /ssapi/choose. Bearer authorization, CRM segment passed in pageAttributes, response in <50 ms.
curl --request POST \
--url 'https://evs-01.gravityfield.ai/ssapi/choose' \
-H 'Authorization: Bearer your-api-key' \
-d '{
"user": { "id": "638073ccfb53d530340043f2" },
"context": { "pageAttributes": {
"crm_segment": "loyal_buyers" } },
"selector": { "names": ["pdp_recs"] }
}'The specification Gravity Field API (OpenAPI) is published together with an MCP server. AI agents and LLM tools read the method schema and call the platform directly.
{ "mcpServers": {
"gravity-field-api": {
"command": "npx",
"args": ["-y", "apidog-mcp-server@latest",
"--site-id=<SITE_ID>"]
} } }
# spec: openapi-v2.gravityfield.aiNative SDKs work as a thin client: they pass screen context and events and receive campaigns and the segment. Initialization takes a couple of lines with apiKey and section.
import ai.gravityfield.gravity_sdk.GravitySDK
GravitySDK.initialize(
context = this,
apiKey = "YOUR_API_KEY",
section = "YOUR_SECTION_ID",
gravityEventCallback = ::handleGravityEvent
)We handle the peak load of large retailers — up to 2 million requests per minute — with no loss of speed. Even on Black Friday response time stays within 300 ms.
The Gravity Field platform is included in the unified register of Russian software (registry entry No. 24502 of 18.10.2024) and accredited by the Ministry of Digital Development as an IT company. That confirms our status as a reliable vendor and entitles your business to tax benefits when working with us.
Machine learning segmentation assigns users to groups automatically and in real time, with no manual rule building. Unlike static CRM segments built on rigid conditions (“spent more than a certain amount in six months”), an ML model weighs hundreds of signals: views, clicks, add-to-carts, time on page, pauses and returns. The segment updates within milliseconds of each user action. You get dynamic groups — “likely to churn in the next 30 days”, “discount-sensitive right now” — that shift together with shopper behavior.
The platform builds up to 25 base segments automatically, for example:
You can add your own rules on top of the ML models, or configure the segmentation logic for your business from scratch.
We use a site script (installed as a single code snippet) and built-in libraries for iOS and Android apps. They track events: product views, add-to-carts, page scrolling, button clicks, checkout starts. Every signal reaches the platform with no more than 50 milliseconds of latency. Offline transactions from your CRM or CDP and data from the contact center are loaded as well. Gravity Field merges online and offline into a single profile, available for personalization during the current visit.
Yes — that is a core job of this module. You load segments from your CRM via REST API, via streaming (Apache Kafka) or by batch export. Gravity Field links CRM segments to behavioral signals and makes them available on the site and in the app: on page load the platform returns the user segment (“VIP customer”, for instance), and your site or app adapts the content instantly — serving product recommendations without a discount to a premium customer, or a welcome banner to a newcomer. CRM segments stop sitting idle and start driving sales.
The Gravity Field platform is included in the unified register of Russian software (registry entry No. 24502 of 18.10.2024) and accredited by the Ministry of Digital Development as an IT company — which confirms our status as a reliable supplier and entitles your business to tax benefits. Data is stored and processed on servers in Russia and encrypted at rest and in transit. Technical characteristics: up to 2 million requests per minute at peak load (during a Black Friday sale, for example), and no more than 300 milliseconds to process a personalization request.
Average client figures for the first 90 days after the module goes live:
The first changes show up about 2 weeks after integration, once enough behavioral signals have accumulated. The full effect arrives by the end of the second month, as the ML models train on the data of your specific store.
A segment determined by Gravity Field is instantly available at every customer touchpoint:
So the customer gets one consistent personal experience moving from an email to the site, or from the app to the call center.
Ordinary CDPs and email services work with batch data: they load history once an hour or once a day. They cannot see what a user is doing on the site right now. Gravity Field specializes in real-time segmentation with under 50 milliseconds of latency and segment updates within a single visit. That makes it possible, for example, to move a user from a “hesitating” segment to a “ready to buy” segment the moment an item lands in the cart, and immediately open a chat window or offer free delivery. Neither CDPs nor email services deliver that speed and depth on the site itself. On top of that, we do not store personal data in the clear (only anonymized identifiers), which makes compliance simpler.
We will build demo segments on your traffic and show how personal scenarios work for your key audiences in real time.