# DTCUP Intelligence: From Connected Data to a Defensible Decision

A practical method for source scope, stored history, refunds, attribution and decision records, so a useful chart becomes an action the team can explain.

Source: https://dtcup.com/blogs/insights/dtcup-intelligence-connected-data-decisions/

DTCUP / FIELD NOTES / OPERATE

# DTCUP Intelligence: From Connected Data to a Defensible Decision

A practical method for source scope, stored history, refunds, attribution and decision records, so a useful chart becomes an action the team can explain.

Published Oct 15, 202624 min read

 ![Jack Hu](https://dtcup.com/wp-content/plugins/dtcup-knowledge/assets/jack-hu-small.jpg)**Jack Hu**7 years of experience building and operating brand websites, specializing in website development, paid advertising and SEO. Built a brand website from launch to US$10 million in annual sales.



  On this page  ## The Short Answer

Define the commercial question, eligible orders, dates and currency before comparing reports. Check source authorization and history coverage, reconcile refunds and avoid adding overlapping provider credit. Save the observation, interpretation, action and review point so another colleague can defend the decision.

## Key takeaways

- Agree the commercial population and calculation before comparing totals.
- Orders, provider history and browsing touchpoints are different evidence.
- Attribution allocates credit; it does not create additional revenue.
- Preserve assumptions, source coverage, action ownership and review timing.

![Illustrated source tables feed a reconciled evidence ledger and a decision card, with a blue chart used as a conceptual analysis graphic.](https://dtcup.com/wp-content/uploads/2026/10/dtcup-intelligence-product-guides-20261015-1024x576.webp)Monday’s report says sales improved. The advertising platform says the campaign deserves the credit. The store report shows a different total, and an email report appears to have sold several of the same orders again. Five tabs later, the team has more numbers and less agreement. The spreadsheet has become a very expensive debating club.

[DTCUP Intelligence](https://app.dtcup.com/) helps bring connected business and marketing evidence into a workspace where you can investigate those differences and make a more defensible decision. Its usefulness comes from connecting the question, the source, the calculation and the scope. A chart is the visible part. The meaning behind the chart is what the operator needs to trust.

Recent work has strengthened source-data storage, bounded historical imports, keyword and search-term analysis, and email-report history. Those developments support an important operating principle: collect authorized data deliberately, preserve the relevant business facts and let the team query those facts without starting every discussion from an empty browser tab.

This guide explains how I would use that capability in practice. It covers setting a decision, choosing sources, reconciling commercial totals, interpreting attribution and creating a useful handoff. The examples are invented to illustrate the method, not customer performance claims. You will also see where missing permissions, incomplete history and provider definitions should remain visible instead of being converted into the comforting but occasionally fictional number zero.

## Start with the decision the data must support

“Review performance” is too broad to guide a useful analysis. A better starting point is a decision with a time horizon and a constraint. Should we continue the regional campaign next week? Which product needs an inventory review before the next promotion? Has a channel become more expensive after accounting for the orders it is actually associated with?

Name the decision owner. This person chooses the action after considering the evidence and the operating constraints. The analyst may prepare the work, but the person who owns the budget, stock or customer promise should understand what the numbers establish and what remains uncertain.

Define the unit of analysis. You might be examining an order, an order line, a campaign day, a customer journey or a product variant. Those units can appear together in one business question, but they are not interchangeable. A campaign’s reported conversions cannot automatically be treated as a list of unique store orders.

Choose the comparison before looking for a dramatic result. Use complete periods where possible, keep timezones consistent and identify promotions, stockouts or major website changes that affect the interpretation. The question is whether two periods are meaningfully comparable, rather than whether the date picker can display them beside each other.

Write the expected output. A useful request might ask for the selected store’s eligible orders, the agreed sales measure, channel credit under a chosen method, missing-data notes and one recommended next investigation. That gives the work a finish line without pretending the software can make every commercial decision on the owner’s behalf.

Separate a monitoring question from a causal question. Monitoring asks what changed in the observed data. Causality asks what caused the change or what would happen without a particular activity. Connected reports can support the first and inform the second, but a correlation between spend and orders does not by itself establish an incremental effect.

I also ask what would change our mind. If the team plans to reduce a campaign, what evidence would support keeping it? If it plans to increase demand, what stock or service limit would block the increase? This prevents analysis from becoming a search for a chart that endorses a decision already made over breakfast.

Keep the question small enough to resolve. One decision about a defined campaign, period and market can produce a useful result. An attempt to explain the entire business in one report often produces a larger vocabulary and a smaller amount of clarity.

[![A growth decision is framed by an owner, analysis unit, comparable period and evidence that could change the proposed action.](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-1.webp)](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-1.webp)A precise question tells the analyst which facts and comparisons matter.## Connect sources with a clear data contract

A connection establishes access to a particular source under particular permissions. It does not establish that every desirable field is available or that historical coverage is complete. Before using a report, check the connected account, workspace, granted access and the last successful update relevant to the data you need.

Think of a data contract as a plain-language agreement about what a number means. It states the source, the account, the unit, the date basis, the currency and any important exclusions. You do not need to use the phrase in a team meeting if it makes everyone reach for coffee. You do need the information it describes.

For store orders, identify which store and which eligible order states are included. For an advertising report, identify the advertising account and the provider’s reporting basis. For email performance, identify the campaign or flow, the selected commercial event and the date definition. These details determine whether two values can be compared or combined.

DTCUP Intelligence provides connection and analysis surfaces with their own roles. Use the available connector that matches the source and question. Verify that the intended account is connected before importing or querying data. A connection from a different company should not become a convenient substitute for a missing one.

Check authorization separately from interface availability. A page may offer an integration path while the provider application still needs configuration or the merchant needs to grant additional access. The report should reflect the access actually granted. A customer field missing because of permission is unavailable, not evidence that the customer never existed.

Choose one authoritative definition for each core commercial measure. This does not mean ignoring other sources. It means using them for the questions they can answer. Store transactions can ground order totals, while a channel’s own report can explain what that channel claims under its own method.

Keep source freshness visible. An old inventory observation and a current order report describe different moments. A campaign report that has not finished updating may change later. State the available timing so a colleague can understand whether the report is suitable for an immediate decision or a retrospective review.

Finally, define what a failed update means operationally. Who checks the connection? Who decides whether a report can proceed with a limited source? Which customer or campaign decisions should wait? This converts a technical status into a sensible business response instead of an alarm that everyone assumes belongs to somebody else.

## Establish commercial truth before discussing channel credit

Before arguing about which channel deserves an order, establish which orders and value you are discussing. An attribution view distributes credit under a method. It does not repair inconsistent order eligibility, currency or refund treatment upstream.

Start with a defined store-order set and a defined sales basis. Ordered merchandise value, refunded merchandise and a net merchandise measure answer related but different questions. Shipping, tax, discounts and fees may have their own treatment. Write the chosen definition in the report so the team can reproduce the calculation.

Consider an invented example. A selected order set contains US$12,000 of merchandise sales before US$2,000 of related refunds under the chosen definition. A net merchandise measure would be US$10,000. With US$3,000 of advertising spend, the first value divided by spend is 4.0, while the second is about 3.33. Those ratios differ because the numerator differs.

Neither ratio establishes profit. Product cost, fulfillment, payment fees, service costs and other expenses may still matter. Keep a return-on-ad-spend calculation labeled as such. Giving it a more impressive name does not make the warehouse or customer service team volunteer to work for free.

Handle refund timing explicitly. A refund processed this week may belong to an order placed earlier. A report based on transaction dates can therefore differ from a report that attaches later refunds to an order cohort. A cohort is simply a group, such as orders placed in a particular week. Choose the basis appropriate to the decision and avoid quietly switching between them.

Do not add values across currencies without a defined conversion basis. A sum of dollars, euros and pounds is a pile of numbers, not a commercial total. If a reliable conversion is unavailable, keep the currencies separate and explain the limitation. This is especially important when a business starts operating in several markets.

Reconcile counts as well as value. A value total can appear plausible even when duplicate or ineligible orders are present. Check the order count, the selection rules and any excluded records. If a source is partial, retain that coverage warning beside the result.

Once the commercial base is clear, channel analysis becomes more productive. The team can ask how a defined set of orders relates to observed marketing activity. It is a much better discussion than comparing several platforms’ favorite revenue numbers and hoping that one of them will eventually confess.

[![A commercial ledger separates eligible orders, merchandise sales, refunds and advertising spend before calculating a labeled ratio.](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-2.webp)](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-2.webp)Agree on the commercial base before interpreting the credit assigned to channels.## Use stored history without assuming perfect history

Repeatedly asking an external platform for the same historical data can make reporting slow and fragile. Provider limits, pagination and historical access rules all affect the result. A more useful workflow imports authorized data in bounded jobs, preserves the relevant business facts and lets the analysis view query the stored result.

DTCUP Intelligence’s recent history work follows that direction. Supported source histories, keyword and search-term retrieval and Klaviyo campaign or flow reporting can be collected under their relevant limits and definitions. The practical advantage is a clearer separation between obtaining evidence and reading an analysis.

Start with a bounded date range. Check the source account, requested period and expected reporting unit before launching a historical import. A smaller complete range is often more useful for an initial reconciliation than a very large request whose coverage is unclear.

Distinguish manual backfill from scheduled updates. Backfill means collecting available earlier data. A daily update keeps an authorized source current according to the configured schedule. These are separate operating choices. Turning off a schedule does not necessarily mean an already-running manual import has been cancelled.

Review job completion and coverage before using the result. A request being accepted is progress; a completed import with the intended dates is stronger evidence. Where the system reports an incomplete range or a provider limitation, keep that information with the report. Page speed should not come from hiding unfinished collection.

Keep raw responses, stored business facts and saved analyses conceptually separate. Raw material may have a short retention window. A published daily fact can be preserved under different rules. A saved report may preserve the result of a particular analysis. These objects serve different purposes and should not be treated as one undifferentiated history drawer.

Customer journeys require their own evidence. Retaining an order does not prove that every earlier browsing touchpoint is still available. A touchpoint is an observed interaction, such as a visit or click. If those interactions were never captured or are no longer retained, the old journey cannot be reconstructed by confidence alone.

Check current retention and source coverage when a decision depends on older history. A missing journey does not automatically mean a customer arrived without marketing influence. It may mean that the required evidence is unavailable. Keep the distinction visible so a retention setting does not become an accidental marketing conclusion.

Give update ownership to a person. Decide who reviews failed imports, who can authorize a schedule and who confirms that the chosen purchase event remains appropriate. Reliable history is an operating habit supported by software, not a background process that can be forgotten until the next quarterly surprise.

[![Authorized source data passes through a bounded import into stored daily facts, with separate paths for scheduled updates and saved analyses.](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-3.webp)](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-3.webp)Data collection, stored facts and saved analysis have different jobs and should retain their own status.## Read totals and provider metrics with their definitions

A table can display one page of rows while its total describes the complete filtered set. That is useful when clearly labeled. It becomes confusing when a reader expects the visible rows to add up to the full result. Check whether each total covers the displayed page, the filtered result or another stated scope.

Use the same filters for the comparison and the export. A campaign or search-term analysis should retain its selected dates, account and relevant inclusion rules. If the CSV describes a different set from the screen, it becomes difficult for a colleague to reproduce the finding.

Calculate ratios from the appropriate totals. If two campaigns have very different spend, their individual return ratios should not be averaged casually. Add the compatible value and spend first, then divide. The same principle applies to many rates: the denominator matters, especially when one row is much larger than another.

Unique people require particular care. A person may open several emails on several days. Adding daily unique counts can count the same person repeatedly. When the source does not establish a deduplicated period total, keep the period’s unique count unavailable instead of presenting a sum with a misleading label.

Klaviyo reporting adds useful evidence about campaigns and flows under its own event and attribution definitions. Read the date basis and selected commercial event before comparing that value with store orders. A provider-attributed email value and a store’s net merchandise sales are not automatically the same measure.

Search terms and keywords also need their own interpretation. A keyword can be a targeting object, while a search term describes a query reported by the platform. Available search-term rows may be subject to the provider’s visibility rules. A complete import of available rows does not establish knowledge of every query a person typed.

Use the report to identify a question you can investigate. A rise in cost for a relevant query may lead to a landing-page review. A flow with changing commercial performance may lead to a message or audience review. Keep the observed metric attached to the proposed investigation so the next person understands the reason.

When the source cannot answer a question, choose another suitable source or a smaller question. This is more productive than making an empty field look complete. A trustworthy analytics workflow leaves space for “we do not know yet,” then assigns a practical way to find out.

## Choose attribution for the question you are asking

Attribution is a way of assigning credit to observed interactions associated with an outcome. It is useful because customers often encounter several channels before ordering. It is also easy to misuse, especially when a team treats one credit method as the complete explanation of why a customer bought.

Choose the method before interpreting the ranking. A first-observed-touch view emphasizes the start of the recorded journey. A last-touch view emphasizes the final eligible interaction. A linear view distributes credit across eligible interactions under its rules. Paid-only and all-channel approaches answer different questions because they consider different sets of interactions.

Use a short invented journey to test your understanding. A customer sees a paid social message, later searches for the brand, clicks an email and places an order. A method focused on the final eligible touch can assign credit very differently from one that spreads credit across the recorded path. The order has not changed. The question asked of the journey has.

Compare methods as a sensitivity check. If a channel looks important only under one narrow assumption, investigate the reason. If a channel consistently appears near the start of observed journeys, that suggests a different role from a channel that frequently appears near the end. It does not automatically tell you the budget each deserves.

Keep unknown or unobserved interactions visible. A recorded path may be incomplete because of permissions, device changes, tracking limitations or retention. A model can distribute credit among the evidence available; it cannot turn missing interactions into observed facts.

Do not add credited revenue from several methods together. Each method is an alternative view of the defined outcome set. Adding the views can count the same orders repeatedly. Compare them side by side with the method clearly named, then select the view appropriate to the decision.

Separate provider claims from your cross-channel view. A platform can credit an order under its own lookback and interaction rules while another platform credits the same order too. That does not mean the store received two payments. The overlapping claims are evidence to interpret, not a new form of revenue multiplication.

For a fuller explanation of model choice, use [the attribution-model decision guide](https://dtcup.com/blogs/insights/choose-attribution-model-channel-decision/). In the Intelligence workflow, the practical discipline is to retain the selected method, the eligible interaction rules, the order set and the limitations beside the conclusion.

[![One customer journey is viewed through first-touch, last-touch and distributed-credit methods without adding their alternative totals together.](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-4.webp)](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-4.webp)Alternative credit methods illuminate different parts of the same observed journey.## Turn a channel question into a bounded investigation

Once the source and credit method are clear, investigate a specific change. Suppose a campaign’s purchase cost rises while its traffic remains similar. Several explanations are possible: a different audience, a weaker page, stock availability, a changed offer, a tracking problem or a difference in the orders included.

Start by verifying the measurement. Confirm the dates, account, purchase definition and source coverage. Check whether a recent import is complete and whether the comparison uses compatible values. This prevents the team from redesigning an advertisement to solve what was actually a reporting mismatch.

Next, narrow the commercial pattern. Does the change affect one product, one market, one campaign or the whole store? A localized change suggests a different investigation from a broad store-level decline. Use the available detail rather than assuming the largest visible percentage identifies the root cause.

Read product availability and customer friction alongside marketing data. A campaign can send interested visitors to a product that is unavailable or poorly explained. [DTCUP Site Analyzer](https://analyzer.dtcup.com/) can help inspect the relevant page, while [DTCUP CRM](https://crm.dtcup.com/) can help the team review the customer questions and service context it has actually recorded.

Keep the handoff explicit. An analytics finding should name the page or product, the observed change and the question to investigate. It should not simply say “conversion is down, please fix.” That instruction provides pressure while outsourcing all the useful thinking.

Choose a repair or test that matches the evidence. If the page’s delivery promise is unclear, review the explanation and measure the relevant customer task. If a data source is incomplete, repair collection and reconcile before changing spend. If the evidence points to a market-specific mismatch, investigate that market rather than editing the entire website.

Consider stock and service capacity before increasing demand. [DTCUP Demand](https://app.dtcup.com/demand) supports a planning discussion around the available product history and explicit assumptions. A promising channel does not establish that the business can fulfill a larger volume on the dates customers expect.

Record what the analysis supports today and what needs another check. This can be a recommendation to maintain the current setting while collecting evidence. It can be a small approved test. It can be a clear technical repair. A useful analysis ends in an owned next step, not a compulsory budget change.

## An illustrative weekly growth review

Imagine a small brand reviewing a regional launch after two complete weeks. This is a teaching example with invented numbers. The team wants to decide whether to maintain its current campaign scope for another week or pause expansion while it fixes a customer-facing problem.

The owner first defines the selected store, currency, timezone and eligible order set. They choose a merchandise-sales measure and keep refunds visible under the stated basis. Advertising spend comes from the selected connected account. Email performance remains labeled with its provider definitions rather than being added to the store total.

The analysis shows that the store’s defined order value improved, but the increase is concentrated in one product. A different product receives substantial campaign traffic and a growing number of delivery questions. The team does not assume that the lower-returning campaign is the sole cause. It asks whether the destination experience is creating uncertainty.

They check source coverage and confirm the campaign and order periods are comparable. A search-term view suggests that many available queries relate to immediate availability. The finding is a clue about visitor intent, with the provider’s reporting limitations retained. It is not a claim that every visitor used those exact words.

The page review then finds that the delivery explanation is difficult to locate and refers to an estimated dispatch date without explaining the next delivery step. Support records contain similar questions. Together, those observations provide a more specific hypothesis than “the campaign is bad.”

The team chooses a bounded action: clarify the approved delivery explanation on that product page, verify the regional destination links and keep the campaign’s current scope while collecting an appropriate comparison period. The decision owner records the reason and the conditions that would justify a further change.

A colleague uses [the DTCUP Browser task method](https://dtcup.com/blogs/insights/dtcup-browser-reliable-business-tasks/) to prepare the authorized page update and check the saved result. The customer-service owner updates the approved knowledge in CRM. These are explicit handoffs, with a person responsible for confirming each receiving system.

At the next review, the team checks whether relevant delivery questions and the customer task have improved. It also checks order eligibility, stock changes and any promotion that might affect interpretation. The result may support the hypothesis, contradict it or remain inconclusive. Each is a legitimate outcome.

The value of Intelligence in this process is the organized evidence chain: a defined business question, compatible sources, a localized pattern and an owned action. The software does not guarantee the campaign’s success. It helps the team avoid spending another week arguing about totals that were never describing the same thing.

## Save an analysis another person can defend

A useful analysis should survive a handoff. The next person needs the question, period, source scope, calculation basis, selected method, findings and unresolved limitations. They should also be able to identify the action and its owner without decoding a collection of screenshots.

Use the product’s explicit save or export workflow where available, and name the analysis around the decision. A name such as “regional launch delivery question review” is more useful than “final report version seven actually final.” The report’s purpose should remain recognizable after the original browser session has gone away.

Distinguish a saved result from a live view. A live view can change as sources update. A saved analysis represents a defined state under its supported behavior. Record enough source timing and coverage to explain that difference, especially when later refunds or provider updates can alter the numbers.

Keep assumptions beside the conclusion. If the comparison depends on a stable price, a chosen refund basis or an incomplete historical period, say so. Assumptions are not embarrassing footnotes. They explain when the recommendation should be reconsidered.

Separate observation, interpretation and action. An observation describes the data. An interpretation offers a possible explanation. An action proposes what to do next. This makes disagreement productive: a colleague can challenge the interpretation while accepting the observation, or agree with the diagnosis while suggesting a different action.

Record the review window. A small page repair may need a technical check immediately and a later customer-behavior review. A commercial test may need a longer period with attention to attribution delay, stock and promotions. Avoid declaring success because the first convenient day looks encouraging.

Keep access appropriate to the analysis. Customer-level details, commercial costs and connected-account information can have different sensitivities. Share the useful conclusion with the people who need it, while preserving the relevant workspace and role boundaries. A common login does not mean every product or colleague receives every permission automatically.

When the discussion happens in a meeting, [DTCUP Meet](https://meet.dtcup.com/) can help preserve the reasoning and follow-up. Review the final decision and assign the action explicitly. An excellent transcript still needs a clear statement of who will do what next.

[![A decision pack contains the question, source scope, calculation method, findings, assumptions, owner and review date.](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-5.webp)](https://dtcup.com/wp-content/plugins/dtcup-product-guides/images/intelligence-5.webp)Save the meaning of the analysis, not just a picture of its chart.## Introduce the workflow without building a reporting maze

Start with one store, one commercially important question and the smallest useful set of authorized sources. A team can learn a great deal from reconciling orders, spend and one campaign question properly. Connecting every possible source before agreeing on definitions often creates more reconciliation work than insight.

Give the setup a clear owner. This person checks accounts, source permissions, import coverage and the chosen purchase measure. The business owner confirms the decision and its constraints. The two roles can belong to one person in a small team, but the responsibilities should remain clear.

Establish a small metric dictionary. Include order count, the selected sales basis, refunds, spend and any decision-specific rate. For each metric, write the source, date basis, unit and exclusions. Keep the dictionary practical; it should help people resolve a question during a review, not become a document that only the person who wrote it understands.

Run one complete analysis cycle. Collect or verify the data, reconcile the base totals, inspect the relevant detail, choose a bounded action and schedule a review. Measure how long the cycle takes and where confusion appears. That gives you a sensible improvement list for the next cycle.

Add sources when they answer a recurring question. Email history may help explain campaign and flow behavior. Search-term detail may help diagnose intent. Product and planning context may reveal a constraint on growth. Each addition should have a defined use and an owner who understands its limits.

Keep history maintenance separate from report consumption. An analyst should know whether the required source has updated, while the person managing imports should know which gaps affect decisions. Clear ownership prevents a useful warehouse of facts from becoming a cupboard of unexplained timestamps.

Review recurring disagreements as a system problem. If the same revenue question appears every week, improve the definition and reporting labels. If a connection repeatedly fails, address its operating process. If actions are never reviewed, strengthen the decision handoff. More charts may not be the missing ingredient.

The wider [DTCUP international growth workflow](https://dtcup.com/blogs/insights/dtcup-apps-international-growth-workflow/) places Intelligence beside page inspection, customer work, planning and execution. The goal is a business that learns from evidence and can act on it, rather than a reporting system that becomes a separate hobby.

## When a number looks wrong, follow one order

A useful reconciliation starts with a small, understandable example. Imagine an illustrative store with ten eligible orders totaling $1,000 before a $100 refund. Under the team’s selected after-refund basis, the commercial value is $900. The advertising account reports $700 of attributed purchase value and the email provider reports $400. These are invented numbers for explaining the method, not results from a DTCUP customer.

The first mistake would be to add the two provider claims and announce $1,100 in sales. The same order can receive credit in both systems. The second mistake would be to call either provider wrong simply because its total differs from $900. Each report may answer a different question, use a different date basis or include different eligible purchases. The purpose of the exercise is to identify the difference before choosing the appropriate measure.

Pick one order that appears in the disputed period. Check its order time, currency, eligibility and refund state in the commercial source. Then inspect the available evidence that links it to a campaign or journey. If the provider does not expose a reliable order-level match, record that limitation. Do not manufacture a match because two totals happen to be close; resemblance is an excellent basis for a family photo and a poor basis for a financial reconciliation.

Next, write the calculation in ordinary language. For this example: “Ten eligible orders, $1,000 in selected order value, minus $100 in refunds, leaves $900.” If $300 of eligible spend belongs to the same currency and period, the resulting revenue-to-spend ratio is three. Calling it profit would be incorrect: product cost, delivery, fees and other expenses have not been deducted. A simple equation with an accurate label beats an impressive acronym attached to the wrong denominator.

Check the table’s scope before investigating a supposed missing record. A visible page may show only part of a filtered result. Compare the stated filtered total, the active account and date filters, and the export scope. If the full export contains the expected record but the first table page does not, the issue may be navigation or sorting rather than ingestion. This small check can save an afternoon of rebuilding a connection that was doing its job.

If the record is absent from stored history, inspect the import boundary. Was that source connected for the relevant period? Did the selected backfill finish? Is the requested day within the provider’s available history? Is the source still updating? Resolve the smallest identifiable gap and read the result back. Repeating a large historical import without understanding the gap can create avoidable work and make the original question harder to trace.

Finally, separate the reporting repair from the commercial decision. Discovering a currency mismatch explains why the report looked strange; it does not establish that a campaign needs a larger budget. Discovering that refunds belong to a later period explains a timing difference; it does not make the refunds disappear. Save the corrected basis, record which earlier conclusion is affected and tell the decision owner what should be reconsidered.

This method also gives disagreements a useful ending. The team can say, “We found a date-scope difference, corrected it and will use this definition next week,” rather than returning to the same argument with a newer screenshot. Keep a short reconciliation note beside the recurring report. It becomes shared working knowledge, and the next analyst can spend their attention on customers and constraints instead of conducting another archaeological expedition through filters.

## Questions about using Intelligence for decisions

### Why can the store, advertising and email reports show different sales?

They can use different order eligibility, refund treatment, currencies, date bases and attribution rules. Establish the definition behind each value before comparing it. Keep the store’s commercial base and each provider’s attributed claim labeled, rather than adding them as if they represented separate orders.

### Does importing history make every old customer journey complete?

No. Orders, daily provider facts and browsing touchpoints are different evidence. A provider may limit historical access, and interactions may never have been collected or may no longer be retained. Check coverage and current retention for the question you need to answer.

### Should I always choose the attribution method showing the highest return?

Choose the method that matches the question and inspect how sensitive the conclusion is to its assumptions. Alternative methods distribute credit differently across the same defined outcome set. A higher credited value under one method does not establish that the business received additional money.

### Why should unavailable data remain blank instead of zero?

Zero means the measured value was zero under a known scope. Unavailable means the report did not establish the value. Permissions, source limits and incomplete updates can all create missing evidence. Preserving that distinction prevents a data-access problem from becoming a false business conclusion.

### Will the product automatically change advertising or supplier orders?

An analysis and its recommendation should remain distinct from an external action. Use explicit authorization and the relevant connected workflow for any change. An internal planning approval does not establish that a supplier purchase order was created, and a report finding does not establish that an advertising budget changed.

### What is the best first analysis for a small international team?

Choose one decision with a defined period, currency and owner. Reconcile its commercial base, inspect one relevant channel or product question and agree on a reviewable next step. The first success is a decision the team can explain and reproduce, rather than the largest possible collection of dashboards.



 CONTINUE / THE TOPIC

## Keep the next question connected.

[NEXT BEST READ**DTCUP Site Analyzer: Find the Website Problems Worth Fixing First**](https://dtcup.com/blogs/insights/dtcup-site-analyzer-practical-website-audit/)[RELATED GUIDE**DTCUP CRM: Build a Customer Record Sales and Support Can Both Use**](https://dtcup.com/blogs/insights/dtcup-crm-shared-customer-record-workflow/)[RELATED GUIDE**DTCUP Meet: Turn International Meetings into Decisions That Survive Monday**](https://dtcup.com/blogs/insights/dtcup-meet-international-meeting-workflow/)[RELATED GUIDE**Going Global with DTCUP: A Six-App Workflow from Market Questions to Repeat Orders**](https://dtcup.com/blogs/insights/dtcup-apps-international-growth-workflow/)

Still have a question?[Ask the DTCUP Community ↗](https://dtcup.com/community/)
