A purchase can appear in Google Analytics, Shopify and a campaign report without becoming three separate sales. That distinction sits at the centre of the update Adsmurai announced on September 30 for its Marketing Platform, AMP: one widget can now combine metrics from different sources. The change reduces an awkward separation between advertising expenditure and commercial activity. It also makes the definitions behind those numbers more consequential when a shared chart starts influencing the next budget decision.
The Barcelona-origin business places the change inside a larger package covering ecommerce and CRM connections, calculated metrics and AI-assisted analysis. Its announcement identifies Shopify, VTEX and HubSpot among the integrations and proposes comparing paid-media investment with GA4 revenue, ecommerce revenue or CRM leads. These are different commercial comparisons. One brings spending closer to an order; another relates it to an enquiry that may take time to become a sale, or may never become one at all.
A shared dashboard becomes a shared visualisation
According to Adsmurai's explanation, a dashboard could already contain information from multiple sources, but each widget previously had to use metrics from a single source. The change takes place inside the visualisation. A team can examine expenditure and sales together instead of moving between separate blocks. This matters in a weekly review where the task is to explain a change in commercial performance, rather than list what each advertising platform reports about itself.
The provider attaches a specific condition to the feature: common dimensions, such as date or country, can be used when the fields are correctly related. AMP's identification of compatible metrics, filters and groupings does not establish universal accounting reconciliation. A country can describe the market targeted by an advertisement or the commercial location of its buyer. When those meanings differ, matching field names does not make the underlying comparison economically equivalent or resolve where a sale belongs.
Responsibility for formulas also becomes more explicit. Organisation administrators can manage their metrics in Data Manager and create Calculated Metrics using information from different sources. This offers an alternative to relying entirely on an advertising platform's reporting vocabulary. It also moves a definition that might previously have lived in an individual's spreadsheet into the shared reporting environment. A formula used by several decision makers becomes an organisational choice about what counts, rather than merely a convenient calculation.
Cost per lead illustrates the distinction. The arithmetic divides monetary expenditure by a population of contacts. Its meaning changes when the CRM separates enquiries, duplicate submissions, accepted opportunities and customers. A widget cannot decide which population constitutes the commercially relevant denominator. AMP provides the place for the comparison; the business still determines which event deserves to be described as a result, and whether that definition remains stable across the periods that the team is comparing.
Revenue in Analytics has a definition
Google's metrics documentation distinguishes purchase revenue from item revenue. Purchase revenue subtracts refunds; item revenue also subtracts refunds and excludes tax and shipping. Two monetary columns can consequently describe the same order and contain different amounts without necessarily indicating a broken connector. A bare revenue label does not explain the difference. The definition tells the reader which economic components have entered the comparison and which have deliberately remained outside it.
Refunds create a second problem involving time. An order recorded during a promotion can be returned after the period used to assess its advertising has closed. The initial report and a later review can answer different legitimate questions: what orders were recorded then, and what revenue survived afterwards. A dashboard connecting business and marketing information becomes more informative when it makes that movement visible. Treating an early sales close as a final commercial outcome hides the subsequent change in the money retained.
Recorded revenue remains exposed to later returns
Share of annual sales large US retailers expected to be returned.
Tap or hover for details. All values are also available in the data table.
Collapse notes
NRF/Happy Returns: expectations reported by large US retailers with revenue over $500 million; 249 respondents in 2024 and 358 in 2025. These are not final audited returns, a Spanish sample or Adsmurai results; cohorts differ.
Explore the data table
| Category | Sales expected to be returned |
|---|---|
| 2024 | 16.9% |
| 2025 | 15.8% |
NRF and Happy Returns, reported by Statista, estimate annual exposure to returns of 15.8% for 2025 from retailers’ stated expectations. It is not a percentage to deduct from a Spanish store’s revenue. Here it explains why AMP’s comparison needs to retain each order’s state: initially recorded revenue may cease to match the money retained after the return window.
The order's identity matters more than neighbouring curves
Google documents a technical mechanism that helps explain this issue: repeated purchase events can be deduplicated using a unique transaction identifier. Its transaction-ID guide specifies that this deduplication applies to web streams, not app streams. Reusing an identifier for different orders can improperly reduce the count, while an empty string is not a valid unique order identity. These conditions affect the underlying observation before it arrives in the reporting interface that will display the result.
Adding GA4 revenue to Shopify revenue as though they were complementary sales would answer the wrong business question if both contained the same orders. Comparing them can instead reveal coverage and discrepancies. This separates a useful calculated metric from an impressive-looking sum. The mathematics can be correct while the business activity is double counted, because the independence of commercial observations does not follow from the independence of the software connectors that supplied the figures to the dashboard.
HubSpot changes the commercial question
Connecting CRM information deserves a different reading from connecting Shopify or VTEX. An order records a transaction; a lead records a potential step in a commercial relationship. Adsmurai includes CRM leads and conversions among its examples, alongside cost per lead linked to each channel. The integration becomes meaningful when the advertising report can follow subsequent commercial development, particularly where the sale requires assessment, contact or negotiation. The intervening work affects both the eventual value and the time needed to understand it.
A contact may write again because it has a second requirement, received no reply or used another form. Those actions do not necessarily represent acquisition of a new customer. If a widget connects expenditure to CRM entries without distinguishing people, requests and opportunities, cost per lead may describe administrative volume rather than commercial expansion. The same information becomes more useful when the analysis preserves the first arrival separately from later activity involving a company already known to the sales organisation.
The date dimension is particularly sensitive here. Advertising expenditure belongs to the point when the placement was purchased; an opportunity may be accepted later and a sale close later still. Organising everything by closing date can show revenue in a low-spending week even though acquisition was financed earlier. Organising it by first contact produces another view, while leaving recent outcomes less mature. These are distinct commercial lenses, each answering a different question about the relationship between acquisition spending and the resulting business.
Commercial data supports different decisions and acquisition tasks
Reported benefits of better marketing data, January 2023. Three selected responses.
Tap or hover for details. All values are also available in the data table.
Collapse notes
Ascend2: 374 decision-makers surveyed online January 16–22, 2023, multiple answers. Statista labels the scope worldwide; countries are not identified in the original. Perceived benefits, not measured gains or AMP outcomes.
Explore the data table
| Category | Decision-makers citing benefit |
|---|---|
| Lead generation | 35% |
| Overall decision-making | 33% |
| Team alignment | 15% |
In Ascend2’s survey, reported by Statista, 35% cited lead generation and 33% overall decision-making among data-quality benefits. The survey does not prove a shared widget improves either. The HubSpot connection makes the distinction concrete: a recorded enquiry can support acquisition, while its progress to an opportunity and sale changes which spending is worth funding.
Context helps explanation without turning coincidence into causality
According to the announcement, Insights Generator incorporates optimisations made in Meta, Google, TikTok and AMP. Automated analysis can therefore consider actions taken during the period when interpreting changes in metrics. This adds operational memory: a budget change need not remain solely in the team's conversation. However, an action preceding a sales increase does not establish that it explains the whole increase. Concurrent changes in pricing, availability or promotion can alter the commercial result while advertising activity changes alongside them.
The ability to attach promotional calendars and business documents is more consequential than producing a fluent summary. A promotion may alter price, basket composition and ordering time. An analysis observing expenditure alone could assign to the media a change that coincided with the offer. Access to the calendar permits a more precise interpretation of what happened, including the possibility that more units were sold alongside a different revenue per order. The document supplies a commercial condition missing from the advertising series itself.
Adsmurai says it has strengthened instructions requiring the agent to separate facts, interpretations and recommendations and to identify missing information. This distinction is well suited to the product's purpose. A fact might be a sales movement; an interpretation might connect it to a promotion; a recommendation might change spending. Preserving those categories prevents uncertainty in the explanation from disappearing when the analysis reaches a final sentence proposing action on a monetary amount. The recommendation remains dependent on the reasoning supporting it.
A pinned insight also belongs to a moment
The provider allows an insight, table or chart to be retained as a static dashboard widget. This turns a temporary conclusion into an object available in later meetings. Its value depends on readers distinguishing the stored interpretation from data that continue updating. A correct explanation of a past promotion may cease to explain the current series while remaining beside it in the same interface. The convenience of retaining a conclusion increases the importance of knowing the period and circumstances to which it belongs.
The consequence for the advertiser is concrete. A shared widget may reveal that a convincing campaign result inside an advertising platform does not produce the same picture in ecommerce or CRM. It may also show that a discrepancy reflects a refund, date or definition and prevent a premature withdrawal of investment. The visualisation deserves attention when it enables that economic discussion. When it merely assembles figures without resolving what they count, the curves move closer together while the uncertainty surrounding the budget remains unchanged.