Lifetimely: Product Journey
Product Journey
At a glance
Product Journey visualizes the most common paths through the first four order positions for repeat customers. Use it to find products, variants, categories, SKUs, order tags, or vendors that frequently appear in successive orders, then investigate cross-sell and lifecycle-message opportunities.
Availability and data freshness
Availability and prerequisites: Customer Behavior is configured as a Pro-tier feature in legacy plan permissions; feature overrides and newer order-limit plans can change access. The current Insights request schema says Amazon beta data is supported only for Repurchase Rate, even though the page can send an Amazon store context.
Freshness and history: The fallback range starts 365 days before today and ends yesterday, using store-timezone day boundaries. A date range saved from this or another report can override that fallback.
Open the report
From Lifetimely, open Customer behaviour > Product Journey (/customer_behaviour/product_journey). The page initially uses Product, graph view, Zoom 50, no report filters, and no hidden values.
Use the report controls
Change the report population and grouping
- Time period: Choose custom dates or Last 6 months, Last 12 months, This year, or a Since 2020 through Since 2025 preset. Presets end yesterday. The selection is saved as the shared report date range and can become the starting range on other reports.
- Product journey by: Group nodes by Product (default), Variant, Product Type, SKU, Order Tag, or Vendor. Changing the grouping clears values selected under Hide noisy data. Product Tag is accepted by the API but is not offered in the current page control.
- Filters: Apply saved cohorts or the filters available for the current store. These can include marketing-touch attributes, tags, product attributes, channel, subscription platform, first-order value or discount, geography, discount code, and survey responses; several Shopify-only groups are hidden for Amazon contexts. Order-based filters let you target the first order, any order, or all orders.
The request also applies store prefilters automatically. Prefilters can remove statuses, channels, tags, POS, draft, cancelled, or zero-value orders before the journey is built, so they can change both order positions and whether a customer qualifies as a repeat customer. They are not changed on this page.
Change only what is displayed
- Zoom: Controls the service's detail limit from 0 to 300; the default is 50. Increasing it requests more rows and makes the graph taller at several thresholds. It does not zoom into an already returned image and can change which lower-volume nodes and links are returned.
- Hide noisy data: Select returned dimension values to remove their nodes and connected links from the graph and rows from the table. This is a local display filter: it does not rerun the calculation or change the CSV export.
- Graph/Table: Switch between the four-stage Sankey graph and a tabular list of returned links.
- Search: Available only in table view. It matches either displayed endpoint, is case-insensitive, and treats plain text as a contains search.
%matches any string and_one character. Search does not change the graph, calculation, or CSV export.
Inspect and export results
- Hover over a node to see its returned share at that order position and, when available, average days from the previous order, average days to the next order, and up to three “Often purchased with” values.
- Hover over a link to see its two endpoints and returned weight. Wider links represent larger weights.
- Click a node with “Often purchased with” data to open the full drawer. Its percentage is calculated as
that returned item's count / sum of counts listed in the drawer × 100; it is not a conversion or attach rate against every eligible order. - The table displays 10 rows per page. It always omits links whose source or destination is Other.
- Export CSV downloads all links returned for the active dates, grouping, filters, and Zoom limit, excluding Other links. It is not limited to the current table page, but it does not apply table search or Hide noisy data. The file contains Order, Product 1, Product 2, Customers, and Average Time Between Orders; the Product labels remain unchanged for non-product groupings.
- Ask AI, when available, receives the returned Product Journey data plus the active dates, grouping, and graph/table mode as report context. Hidden values, Zoom, and applied filters are not separately named in the page-context object, so verify the visible controls when evaluating an answer.
Understand the graph
The graph has columns headed 1st order, 2nd order, 3rd order, and 4th order. A node is one returned dimension value at one order position; the same value can therefore appear in several columns. A link connects two returned nodes and uses the service-provided weight displayed as Customers in the table.
| Graph element | What the page displays | Interpretation limit |
|---|---|---|
| Node | Dimension value at an order position. | A node is stage-specific, not one product total across the report. |
| Avg. time from previous/to next order | Service-provided node timing in days. | This is node-level timing, not timing for one specific outgoing link. |
Caution: Link prominence is not next-purchase probability. The page does not provide the eligible source population needed for
destination customers / source customers, and the detail limit can omit lower-ranked paths.
Understand the table and CSV
Each row represents one returned source-node-to-destination-node link after Other links are removed. Search and hidden-value selections can further reduce the rows visible in the table.
| Column | Definition |
|---|---|
| ORDER | Source and destination order-position numbers parsed from the link, such as 1-2. |
| PRODUCT 1 | Source dimension value. The header still says Product when grouped by Variant, Product Type, SKU, Order Tag, or Vendor. |
| PRODUCT 2 | Destination dimension value, with the same labeling limitation. |
| CUSTOMERS | Service-provided link weight. |
| AVG TIME BETWEEN ORDERS | The source node's returned average days to its next order. Every row with the same source value at the same order position can therefore show the same average; it is not calculated specifically for the PRODUCT 1 → PRODUCT 2 pair. Missing values display 0 in the table but export as a blank CSV cell. |
Worked interpretation
Suppose the returned data contains these first-to-second-order links:
- Coffee → Filters: 80 Customers
- Coffee → Mug: 20 Customers
- the first-order Coffee node has average time to next order of 30 days
The table displays 30 days on both Coffee rows because that timing belongs to the first-order Coffee node. It does not establish that Coffee → Filters specifically averages 30 days. Use the 80 versus 20 weights to prioritize a hypothesis, then validate the customer grain and link-specific timing outside this report before scheduling a day-30 campaign.
Check and reconcile a result
- Confirm the store, Time period, Product journey by, active Filters, and store prefilters.
- Set Zoom high enough to test whether the value was hidden by the detail cutoff, and clear Hide noisy data and table search. Remember that neither display filter affects the export.
- Hover over the source node and link. Distinguish the node's stage percentage and timing from the link's Customers weight.
- Export the active report to inspect every returned non-Other link rather than only the current 10-row page.
- Compare the relevant customers and order positions in the commerce source.
Product Journey should not be expected to match Time between orders: that report groups selected order intervals, while Product Journey returns a detail-limited set of dimension paths and attaches timing to nodes. Repurchase Rate Breakdown uses anchored cohort denominators rather than link weights. Funnel is the better report for measuring a deliberately configured sequence and its conversion/drop-off.
Known limits
- If the selected grouping has no returned values, the graph says data is not available for that journey type; an empty table says No data available in table. Broaden dates or filters, raise Zoom, and confirm prefilters before concluding there are no repeat paths.