Your dashboard says the campaign generated leads. Your inbox says something different. Or the inquiries arrived, but no one can tell which campaign brought them in.

A report is useful only when the actions it describes match what actually happened. If your results from the CAI Assessment identify Measurement & Attribution as a primary constraint, start by tracing one customer journey instead of building another dashboard.

This guide will help you verify what you measure, connect an inquiry to its source where possible, and distinguish evidence from assumptions. You do not need perfect attribution to make better decisions. You need to know which numbers you can trust and where the trail goes cold.

Choose your starting point: Pick one important inquiry path, such as a quote request or consultation form. Use a clearly labeled test submission with permission. If possible, also select a few recent real inquiries whose outcomes you can inspect in your CRM.

Follow the evidence to the outcome

Each milestone is different. A recorded event alone does not prove receipt or a customer outcome.

  1. Source

    The campaign, channel or referral where the inquiry originated

  2. Landing page

    The destination that received the visitor

  3. Inquiry action

    The action the customer actually completed

  4. Received lead

    The usable inquiry the business actually received

  5. Qualified conversation

    A conversation the business can serve

  6. Customer outcome

    The business outcome recorded afterward

Check 1 of 5

Have you defined the actions that matter?

A page view, button click, form start, submitted inquiry, qualified lead, and new customer are different events. Counting them all as “conversions” can make a report look impressive while obscuring the outcome the business cares about.

Imagine a roofing company running a campaign for repair estimates. The report shows 40 clicks on “Request an Estimate,” but only 12 forms were submitted. Of those, the team received 10, and six were requests it could actually serve. These are different questions, and each number needs a clear name.

  • What exactly must happen for it to count?
  • Where is that action recorded?
  • Is it a customer action, a successfully received inquiry, or a business judgment about lead quality?
  • Could the same person or submission be counted twice?
Test it
Write down the actions your reports currently call a lead or conversion. For each one, ask:
Write down what happened
Create a short definition for each meaningful action. Mark any metric whose trigger or meaning you cannot explain as unknown .
If this check fails
Agree on a small measurement vocabulary before editing tags or reports. For example: inquiry started, inquiry successfully submitted, inquiry received, qualified inquiry, and customer. Keep those milestones distinct.

Google Analytics recommends generate_lead for a submitted form or request for information and also supports later lead-stage events. The event name is useful only when its implementation reflects the action you intend to count: Google Analytics recommended events.

Check 2 of 5

Does the reported event match a real customer action?

A tracking tag can fire when someone clicks a button, even if the form fails. A thank-you page can be viewed twice. A single-page application can miss navigation events or send duplicates if its page-view setup is inconsistent.

If you do not have access to the analytics setup, ask whoever manages it to show you the test event and explain what triggered it. Do not accept “the tag fired” as proof that a usable inquiry arrived.

Test it
Complete the chosen inquiry path once. Follow it from the landing page through the confirmation. Check what your analytics tools record at each step, then compare the reported submission with the inquiry that actually reaches the business.
Write down what happened
Record which actions were observed, which were missing, and whether any appeared more than once. Note the page and time of the test so the events can be located.
If this check fails
Fix the event trigger at the point where it diverges from the real action. A lead event should represent the intended completed action, not merely interest in starting it. Then repeat the test.

For sites that change pages without a full reload, page views need particular care. Google provides separate guidance for measuring single-page applications: Google’s single-page application measurement guide.

Check 3 of 5

Can you identify where the inquiry came from?

A useful source trail might show that someone arrived from an ad, an organic search result, an email, a referral, or another campaign. Missing or inconsistent campaign information can turn those visits into an unhelpful “unknown” or misleading source.

Ask two separate questions:

1. Can the website report identify the visit’s source or campaign?

2. Can the business connect that information to the received inquiry when it needs to evaluate lead quality?

Test it
Use a tagged test link from one campaign or channel you control. Follow it to the website and submit the labeled inquiry. Then look at the analytics record and the information available to the person handling the lead.
Write down what happened
Record the link used, the source shown in analytics, and any source information retained with the inquiry. Do not put a customer’s name, email address, phone number, or form message into analytics URLs or event data.

If this check fails: Standardize campaign naming and link tagging, then repair how source information passes through the inquiry path where appropriate. Test the full path again. Google’s analytics policies prohibit sending personally identifiable information such as names and email addresses to Google Analytics: Google Analytics guidance on avoiding personal information.

If the people arriving are simply a poor fit for the offer, use the Traffic & Campaigns check to examine targeting and message fit. Better tagging will make that issue visible; it will not correct the audience by itself.

Check 4 of 5

Can you reconcile the dashboard with received inquiries?

Analytics may show a submitted request while the CRM has no record. The CRM may contain phone calls and referrals that never passed through a website form. Spam and test submissions may appear alongside legitimate inquiries.

These differences do not automatically mean one system is wrong. They do mean the numbers need definitions and a reconciliation process.

Start by inspecting individual examples behind a mismatch. Was an event counted twice? Did a submission fail to create a record? Did a phone inquiry enter the CRM without a web event?

  • Reported website inquiries
  • Inquiries received by the team or CRM
  • Clearly labeled tests and spam
  • Inquiries judged relevant to the business
Test it
Choose a recent period and compare:
Write down what happened
Record what each system counts and the specific examples that explain the difference. Do not force the totals to match by changing definitions after the fact.

If this check fails: Repair the earliest broken handoff. If the submission never reaches the team, work through the Lead Capture & Automation check. If it reaches the team but its status or outcome is never updated, use the CRM & Customer Lifecycle check.

Check 5 of 5

Can you connect marketing activity to a business outcome?

A channel that produces many inquiries may produce few qualified conversations. Another may produce fewer inquiries but better-fit customers. Looking only at clicks or form submissions can hide that difference.

  • Where the person appeared to come from
  • What they requested
  • Whether the business could serve them
  • Whether a conversation or appointment occurred
  • The current outcome
Test it
Take a small sample of recent inquiries from the service you chose. For each one, identify what you can verify:
Write down what happened
Mark each connection as verified , inferred , or unknown . For example, a CRM record with a retained campaign source may support a stronger conclusion than someone’s later recollection of seeing an ad.
If this check fails
Decide which outcome the team needs to record consistently. Add a simple way to update lead quality and final outcome in the CRM. Then evaluate channels using both inquiry volume and the quality of opportunities they create.

What can you actually conclude?

Verified, inferred or unknown

  1. Verified

    A real action, received inquiry or outcome supported by an inspectable record.

    What this means

    State what the evidence directly supports.

  2. Inferred

    Credit assigned using the observed journey or an attribution model.

    What this means

    Acknowledge interactions and influences the systems may not observe.

  3. Unknown

    A source, handoff or outcome that cannot be traced.

    What this means

    Record the gap instead of treating an assumption as fact.

Attribution means assigning credit for an outcome to one or more interactions. That can be useful, but it is a model of the journey—not a complete record of everything that influenced a customer.

A homeowner might find a business in search, read a review, ask a neighbor, and later call from a saved phone number. Your systems may observe some of those steps and miss others. Different attribution models can assign credit differently to the same observed path: Google Analytics’ attribution explanation.

Use the strongest conclusion your evidence supports:

1. Events do not reflect real actions: Fix the measurement implementation before comparing campaigns.

2. Sources disappear before the inquiry: Repair campaign tagging and the source handoff.

3. Analytics and received inquiries disagree: Reconcile specific records and identify the broken step.

4. You know where inquiries came from but not whether they were useful: Record qualification and outcome consistently.

5. The chain is reliable enough to compare: Use the evidence to decide what to improve, test, continue, or pause—while acknowledging journeys you cannot fully observe.

Do not present an attributed revenue figure as an exact account of what would have happened without a channel. And do not stop making decisions simply because attribution is imperfect. State what is known, what is inferred, and what remains unknown.

Start with one trustworthy journey

The first goal is not an elaborate dashboard. It is a trace you can explain:

This person came from this source, reached this page, completed this action, became a received inquiry, and had this business outcome.

Once that path works for one test and holds up across real examples, you can expand it to other forms, calls, campaigns, and stages.

Measurement & Attribution gives the CAI a feedback loop. It helps you see whether Traffic & Campaigns bring relevant people in, whether the inquiry path works, and which outcomes deserve more attention. A clear measurement chain turns activity into evidence for the next decision.