Customer context (CDP)

The more modules you switch on, the better each one gets. The shared customer context is why: what one module learns about a visitor, the others can use. The chat knows what was just searched. The search ranks sporty watches higher when somebody told the chat they are looking for a sporty watch. A review invitation waits while a return is in progress.

You set it per project under Manage → Project → Customer data. It is on by default; off, every module works exactly as on its own.

What Nexym knows, in four groups

Every fact has a source (the module that knows it), a trust level (below) and an age. In the dashboard you see them per visitor: click a visitor in Signals → Live, or open a customer and pick the Context tab.

Group	What	From
A · This visit	arrival (landing page, channel, utm, ad click), device, language, new or returning, consent, A/B groups; searches, searches with no results, clicked results, filters and sort; viewed products and categories with time on page, scroll depth and where they came from; out-of-stock products viewed; the cart, a coupon tried, the checkout step reached; the Product Advisor (started, answers); popups (seen, closed, clicked, signed up, prize won); the chat (topics, hand-over); stated preferences	your site (the embed script), Site search, Product Advisor, Popups, Website chat
B · Across visits	favourite brands, categories, price band, colours and sizes (recent counts more), viewed again but never bought, search themes, how often and on which device they come back	only with consent — see below
C · Customer history	orders (count, first and last, average and total spend, brands and categories bought most, sizes bought, payment and shipping method, city), RFM segment, returns and reasons, reviews, earlier chats, quotes, newsletter, segments, how the customer was recognised	the hub and your modules
D · Signals	bad experience, size uncertainty, price sensitive, high intent, abandoned cart, out-of-stock interest, churn risk, loyal customer, returning visitor	derived from A–C; never surer than what they came from

The cart comes straight from your shop’s own cart on Magento, Shopify and WooCommerce: the embed script reads it when a page loads, after every change and when the visitor returns to the tab. A quantity changed on the cart page, something added in another tab, or a cart emptied by an order or a logout is therefore right in popups and UI elements (“only with something in the cart”), in segments (“has … in the cart”), in the chat and in the abandoned cart signal. Only which products, how many and at what price are read — nothing about who. Where the cart cannot be read (or on Shopware), Nexym builds it from the cart events as before.

What is said to a visitor

Trust	Who	Used for	Said
Session	this visit, anonymous	searches, viewed products, cart, preferences	yes: “you were just looking at …”
Recognised	probably a returning customer (Recognise customers)	silently: favourite brands, a size to suggest	never: no orders, returns, reviews or name
Verified	proved who they are: order number + e-mail in the chat, or logged in	the full history	briefly, when it helps — an earlier bad experience is acknowledged once

The rule is enforced, not just promised. The chat only receives personal history once the conversation is verified, and every reply is checked: one that names an order, return, review or name to an unverified visitor is sent back and, if it happens again, replaced by an offer to verify an order first.

A customer’s Journey

Open a customer and pick the Journey tab. You see everything the customer’s recognised browsers did, across every module, per session and per day, newest first:

• at the top: first seen (and from which source), identified (how long after the first visit, and through which form), sessions (count, devices, time on site), first order and Nexym touches (which modules helped);
• filters per kind: visits, browsing, search, Nexym modules, identity & consent, orders, after sale;
• Timeline: every session with its events — pages and products viewed with time and scroll depth, searches with their result count, filters, cart and checkout, popups, the chat (linked to the conversation), the Product Advisor, consent and the moment of recognition. A product viewed, or put in or taken out of the cart, shows as a card: a small photo (the variant’s own where it has one), the name, the price, the SKU and time and scroll depth. The price is today’s catalogue price (hover to see it say so), unless the event carried its own. Click the card to open the product. “Mail me when it’s back” is a card too: Asked ‘Mail me when it’s back’, Unsubscribed from ‘Mail me’ and ‘Back in stock’ mail sent (with led to order #… when an order followed). The moment of recognition says where the address came from, not only the module: Recognised via ‘Mail me when it’s back’ (Size checker) with the product, size and masked address, Review login (Reviews), Newsletter via chat, Returns portal;
• Flow: the same sessions as a short path, from source through browsing and search to the order. Within a visit the path reads from arrival to exit; the days and visits themselves are newest on top, as in the Timeline.

Each day has a header with its date (with the year when it is not this year, e.g. Thu 16 Oct 2025) and, on the right, how long before it the previous active day was: 15 days earlier, or whole hours under two days (46 h earlier). The oldest day has none. Days follow Nexym’s time zone (Europe/Amsterdam).

The individual events are only kept on the same consent as Across visits, and cleaned up after the retention period. Orders, reviews and returns stay visible. A browser you test with yourself is marked Test.

What the modules do with it

• Site search moves products that fit a stated preference a few places up and avoided ones a few down, with a Visitor’s preference boost rule — never hidden, only within the match on the words, capped so relevance wins. Strength, searches, websites and segments are set in that rule (Search → Boost rules). A filter a visitor ticks is a preference for the others too.
• Website chat knows what was searched and viewed this visit (“what was I just looking at?”) and records preferences when somebody says what they want — only in values your catalogue really has. “Start over” clears them. For a verified customer it knows the history: a return for too small becomes one size up for that brand.
• Product Advisor hands the answers of a finished guide over as preferences.
• Reviews sends no invitation while a customer has an open complaint, a return in progress or left a low review in the last 30 days.
• Popups can target signals: only price-sensitive visitors, high intent without a purchase, returning with an abandoned cart, not after a bad experience.
• Segments can use RFM and the signals, with ready-made segments (win-back, brand fans, price sensitive, high intent, VIP, abandoned cart, out-of-stock interest).

The settings

Under Manage → Project → Customer data: the master switch; which modules may use it (chat, popups, segments, product finder — personalising the search is a boost rule, and holding review invitations back is the Reviews module’s own Never ask while … setting); which facts are kept; the signals and where their lines are drawn (e.g. high intent = viewed 3 times without buying, low review = 2 stars or fewer, the RFM bands). The page also shows which modules are on and which facts you therefore do or do not have.

Measuring what it is worth

Under Site search → A/B test → Personalised vs. not you split visitors between the search as it is (A) and the same search personalised (B), with the same twelve metrics per group. It only runs while the search itself is not under test.

Privacy and consent

• This visit is anonymous: tied to the day’s visit hash (the one Impact and the visit statistics already use) and gone two hours after the last activity.
• Across visits is only kept while Recognise customers is on, the browser was allowed to keep an id, and the visitor gave the consent you ask for there (statistics or marketing). How long it is kept is under Data & retention.
• Nothing sensitive is inferred: no health, no religion, no origin — only what somebody did on your site and bought from you.
• Deleting a customer takes their context with them: their browsers’ profiles and the facts of a visit still in progress.