Measurement · Attribution · CRM evidence

Reconciling Platform ROAS vs. CRM Pipeline Reality

A practical method for testing whether advertising-platform conversions connect to website actions, usable CRM records, qualified opportunities, and closed revenue.

An advertising platform can report a healthy return on ad spend while the CRM shows little qualified pipeline. That does not automatically mean the platform is wrong, the agency is hiding something, or the CRM is correct. It means the systems are answering different questions with different rules.

The ad platform sees the interactions and conversion actions available to it. The website sees sessions and events. The CRM sees the contacts and deals that were captured, associated, updated, and maintained. Finance sees recognized revenue. A reliable performance review does not select one system as the winner. It reconciles the chain and identifies where the definitions, identifiers, or records stop matching.

Practical workflowFollow the evidence from ad interaction to revenue

Compare each handoff in sequence. Mark the connection verified, assumed, or not tracked before interpreting the final ROAS number.

  1. 01

    Advertising interaction

    Campaign, ad group, click or view identifiers, and cost.

    Evidence to checkWhich paid interactions occurred?

  2. 02

    Platform conversion

    Conversion action, timestamp, value, and attribution setting.

    Evidence to checkWhat event received credit?

  3. 03

    Website event

    Form submission, call, meeting, purchase, or event ID.

    Evidence to checkDid the intended action actually complete?

  4. 04

    CRM contact

    Contact ID, create date, record source, and traffic-source properties.

    Evidence to checkWas a usable record created or updated?

  5. 05

    Qualification

    Lifecycle stage, owner, disposition, and qualification evidence.

    Evidence to checkDid sales accept or reject the inquiry, and why?

  6. 06

    Opportunity

    Deal ID, amount, stage, and associated contact and company.

    Evidence to checkDid the inquiry become pipeline?

  7. 07

    Revenue

    Closed-won date, amount, and finance status where available.

    Evidence to checkWhat business outcome can be verified?

The totals do not have to match one to one. The objective is to explain every material difference and identify the evidence needed to resolve it.

Start by defining what the platform calls a conversion

Before discussing ROAS, list every conversion action included in the advertising report. A conversion may be a completed lead form, a phone call, a scheduled meeting, a purchase, an imported offline event, or a page or button event configured to stand in for one of those outcomes.

For each action, document:

  • the event that triggers it;
  • whether one person can trigger it more than once;
  • whether it is a primary or secondary action;
  • the attribution model and conversion window;
  • whether a value is assigned, imported, or estimated;
  • whether enhanced or offline conversion data is involved; and
  • the date of the last verified test.

Google explains that a conversion window determines how long after an ad interaction a conversion can be recorded. Its attribution reports also use lookback windows and may present totals differently from standard campaign reports. Those settings affect which interactions receive credit and when the result appears. They should be written down before two reports are compared.

Official references

ROAS is only as reliable as the value entering the platform

The formula is simple:

Reported ROAS = attributed conversion value ÷ advertising cost

The difficult part is the numerator.

For an eCommerce transaction, the platform may receive an order value directly from the checkout. That value still needs checks for duplicate events, refunds, taxes, shipping, currency, and consent-related gaps.

For a B2B lead-generation program, the platform may receive a fixed value assigned to every form submission or a later value imported from the CRM. A fixed lead value is an operating assumption, not closed revenue. An imported value is stronger only if the contact, opportunity, and revenue records are correctly associated with the original advertising interaction.

This distinction should be visible in reporting. Label the number according to what it represents: platform-assigned lead value, pipeline value, closed-won deal value, or recognized revenue. Do not combine those labels under one generic “revenue” heading.

Build one reconciliation table from click to revenue

Use a common date range, time zone, and business definition, then reconcile the evidence in sequence.

StageEvidence to compareMain question
Advertising interactionCampaign, ad group, click or view identifiers, costWhich paid interactions occurred?
Platform conversionConversion action, timestamp, value, attribution settingWhat event received credit?
Website eventForm submission, call, meeting, purchase, event IDDid the intended action actually complete?
CRM contactContact ID, create date, record source, traffic-source propertiesWas a usable record created or updated?
QualificationLifecycle stage, owner, disposition, qualification evidenceDid sales accept or reject the inquiry, and why?
OpportunityDeal ID, amount, stage, associated contact and companyDid the inquiry become pipeline?
RevenueClosed-won date, amount, finance status where availableWhat business outcome can be verified?

The rows will not always match one to one. A returning contact may submit twice but remain one CRM record. Several contacts may be associated with one opportunity. One person may convert through more than one tracked action. The objective is not to force equal totals. It is to explain every material difference.

Separate attribution differences from pipeline failures

Reporting-rule differences

  • different attribution models;
  • different conversion or lookback windows;
  • platform reporting by ad-interaction date while another report uses conversion date;
  • cross-device or modeled conversions;
  • time-zone differences; and
  • late offline conversion uploads.

Operational failures

  • a form event fires without a successful submission;
  • a submission reaches email but does not create or update the CRM record;
  • source parameters are lost during a redirect or embedded-form handoff;
  • duplicate contacts split the history;
  • a contact has no owner or follow-up status;
  • a deal is created without the originating contact association; or
  • closed revenue is never returned to the reporting system.

Treat these as different findings. Changing the attribution model will not fix a broken form-to-CRM handoff. Repairing the handoff will not make two platforms use the same credit-allocation rules.

Check source, record source, and deal associations separately

HubSpot distinguishes traffic-source properties from record-source properties. Original and latest traffic source describe how a contact first or most recently connected with the business. Record source describes how the CRM record was created. Those fields can answer different questions and should not be substituted for one another.

HubSpot also notes that ad metrics calculated inside HubSpot depend on tracking and association requirements. Revenue reporting depends on the deal, its stage and close date, and its association with the attributed contact. An ad click can be genuine while the revenue link remains unprovable because the CRM association is missing.

Official references

Use three statuses instead of pretending every number is equally reliable

  • Verified: a record, identifier, and tested process support the connection.
  • Assumed: the connection is plausible but has not been tested or matched.
  • Not tracked: the system does not retain the evidence needed to make the connection.

This prevents a dashboard from presenting an assigned lead value with the same confidence as closed-won revenue. It also gives the operating team a practical repair list.

Questions leadership should ask before accepting reported ROAS

  1. Which conversion actions are included in this ROAS number?
  2. Is the conversion value transactional, imported from the CRM, or assigned as an assumption?
  3. What attribution model, date basis, time zone, and conversion window are being used?
  4. Can each priority conversion be matched to a real website action or inquiry?
  5. Can the inquiry be matched to one CRM contact with a source, owner, and disposition?
  6. Is the contact associated with the opportunity and closed-won record used in revenue reporting?
  7. How are duplicate submissions, duplicate contacts, refunds, cancellations, and reopened deals handled?
  8. Which gaps are reporting differences, and which are broken operating handoffs?

The useful outcome is not one perfect number

A reconciliation should produce four things:

  1. a shared definition of conversion, qualified pipeline, and revenue;
  2. a documented mapping between platform events and CRM records;
  3. a list of unexplained gaps with an owner and verification date; and
  4. a reporting view that separates verified revenue from modeled or assigned value.

The goal is not to make the CRM total equal the platform total. The goal is to know why they differ, which number is appropriate for each decision, and what evidence would make the next report more reliable.

For a broader review method, see Independent PPC and SEO Evaluation: Is Your Agency Reporting Telling You the Whole Story? For the operational handoff after a website inquiry, see What Should Happen After a B2B Contact Form Is Submitted?

The goal is not to force the CRM total to equal the platform total. The goal is to know why they differ and which evidence supports each business decision.

Written and reviewed byJoseph George

Founder, Smart Marketing Lab LLC

Published · Last reviewed