Illustrative example. At an early-November meeting, everyone brings their own September numbers. Marketing shows 18,400 website sessions and 131 form submissions. Sales reports 114 new CRM records, including 100 valid leads and 62 qualified leads, plus 20 paid orders. Accounting reports $410,000 in payments received. Every number is correct, and each comes from its own system. But together they don’t answer the question “Should we increase the ad budget?” It’s unclear why there are more form submissions than CRM records, whether those 20 orders actually came from September leads, and whether the $410,000 came from those leads.
A compact report becomes useful when every metric has an agreed-upon definition, source, period, and course of action when the numbers move. There’s no need to stitch separate systems into one neat funnel of “impressions → sessions → leads → payments” if the events in those systems aren’t connected. That kind of funnel looks precise, but the rates between its steps mean nothing.
The customer journey from search query to payment, and where money gets lost along the way, is covered in “Where businesses lose money between search and sale.” This article covers the recurring report itself: a template, a metrics glossary, and the rules for reading it.
Which decisions the report should support
An owner’s report is a tool for decisions you make again and again. If a metric doesn’t support any decision, it belongs in a team-level operational report. We suggest building the report around four types of decisions:
| Decision | Question the report answers |
|---|---|
| Customer acquisition spend | How much does a lead cost, and should channel budgets change? |
| Lead handling | Has lead quality slipped, and did every lead reach the sales team? |
| Measurement quality | Can we trust this period’s numbers, or do we need to reconcile them first? |
| Website priorities | Is there a signal to pass to the SEO specialist or developers? |
The third row often gets skipped, yet it determines whether you can make the first two decisions: if one in five leads has an unknown source, your channel breakdown is incomplete.
It’s worth stating the report’s limits up front. It doesn’t calculate profit; that requires cost of goods sold (COGS), other expenses, and accounting rules. It doesn’t prove that a channel caused a sale: the attributed source shows where the customer came from under the chosen rules, not why they bought. And it doesn’t replace a profit and loss statement (P&L), a return on marketing investment (ROMI) analysis, an attribution model, or a dashboard.
Which metrics belong on one page
A basic report needs seven metrics covering six areas: visits, leads, lead quality, payments, spend, and data quality.
Core metrics glossary
| Metric | Definition for this sample template | Source | Decision or check |
|---|---|---|---|
| Visits | Website sessions in the selected period; not Search Console clicks | GA4 | Check changes by source and device |
| Valid leads | Unique leads, excluding duplicates and spam under the rules you’ve set | CRM | Check volume and reasons for exclusion |
| Qualified leads | The subset of valid leads that meets your business criteria | CRM | Check quality and reasons for rejection |
| Paid orders | Orders by payment date; for cohort calculations, only orders linked to the cohort’s leads | CRM and payment records | Check status, link to the lead, delay, and cancellations |
| Payments received | Payments collected, with a stated rule for refunds; not the same as recognized revenue, and not profit | Accounting system | Check receipts and refunds |
| Ad spend | Spend on the selected ad channels for the period, with currency and scope defined up front | Ad platforms or accounting | Calculate cost per lead (CPL) with a comparable denominator |
| Share of leads with an unknown source | The share of valid leads whose source isn’t identified under your rule; the formula is in the derived metrics | CRM | Check how source data is captured and passed through forms to the CRM |
Leads are split into valid and qualified because volume can grow while quality falls. The share of unknown sources shows how complete your channel breakdown is.
You can add search impressions, clicks, and average position as a separate diagnostic section, but don’t treat them as steps in the same funnel as sessions and payments: these are different events with different counting rules. How to read that data is covered in “Google Search Console: what business owners should check in the reports.”
Derived metrics
| Metric | Numerator | Denominator | Conditions |
|---|---|---|---|
| Channel CPL | Ad spend for the channel | Valid leads from that channel | Same period and currency; not calculated when there are no leads |
| Blended CPL | Ad spend across all ad channels | All valid leads, including those that came in without ads or with an unknown source | Same period; labeled as blended because it shifts with leads from other sources |
| Share of qualified leads | Qualified leads | Valid leads | One lead cohort; criteria defined in advance |
| Share of leads with a paid order | Leads in the cohort with a linked paid order | All valid leads in the cohort | Cohort by lead creation date; cutoff date stated |
| Share of leads with an unknown source | Valid leads with no identified source | All valid leads in the same cohort | The rule for “source identified” is defined in advance |
CPL is the cost of a lead, not a measure of profitability. Add ROAS (return on ad spend: ad-attributed revenue divided by ad spend) only when you have reliably attributed revenue; ROAS isn’t profit either.
How to agree on definitions
Arguments about numbers most often start when different things go by the same name. Marketing counts every form submission as a lead; sales counts only a CRM record created after talking to the customer. The report can show both numbers side by side, as long as they have different names and a note explaining why they differ.
Agree on these in writing:
- Lead: which channels count (form, phone call, text message or live chat, email) and at what point a lead counts, for example once a CRM record is created. A form submission event in analytics is a website signal, not a lead as the report defines it.
- Duplicate: a repeat inquiry from the same contact for the same request within a set time window, such as 30 days; matched by phone, email, or both.
- Spam and junk leads: bots, sales pitches, job applicants, vendors; also decide who records the reason.
- Qualified lead: your business criteria (service area, service type, deal size), plus who marks leads as qualified and when.
- Sale and payment: a signed contract, an invoice, or a received payment; how partial payments and deposits are handled. Counting payments is more reliable for the report because the accounting system confirms them.
- Refunds: subtracted in the month of the refund or from the original order; pick one rule.
- System of record when numbers conflict: the CRM for leads, the accounting system for payments, and ad platforms or accounting for spend.
Google Analytics 4 distinguishes between events, key events, and conversions. A key event is an event you’ve marked as important to your business, such as a form submission. Google now reserves “conversion” for ad measurement: you create one from a key event for use in Google Ads. In the report, use “conversion rate” only for a business ratio with a stated numerator and denominator.
Keep definitions in a single document with the date of every change, and flag the change in the report for the month when a rule changed.
How to combine data from different systems
The report draws on at least four systems: website analytics, the CRM, ad platforms, and the accounting system. Before combining the data, agree on the rules.
| Rule | What to decide | Sample decision |
|---|---|---|
| Period boundaries | Calendar month, dates inclusive | September 1–30, inclusive, for all systems |
| Time zone | The time zone for all dates | Eastern Time in GA4 and ad platforms. Search Console uses Pacific Time for standard date ranges, so search data is reported separately |
| Currency | Reporting currency and exchange rate | US dollars; spend in other currencies converted at the exchange rate on the charge date |
| Taxes and fees | Whether sales tax and platform fees are included in spend | With or without tax, as agreed with accounting; the same rule every month |
| Record matching | Fields used to link records across systems | Lead to order by CRM record ID; order to payment by invoice number |
| Unknown values | Records with no source or no link | Keep them in an “unknown” group; don’t distribute them proportionally |
| Missing data | How to mark a metric with no data | “No data” or “payments entered through September 25 only,” not 0 |
An unknown source stays unknown: if you distribute those leads proportionally across channels, a tracking error gets buried in the channel numbers. And missing data isn’t zero: zero means there were no payments, and it can lead to a wrong decision.
Reconciling data across systems
Compare apples to apples: match form submission events in analytics only against CRM records that came from forms, not against all leads. The gap can go either way. There are more events when someone submits a form twice, the event fires again, a bot fills out the form, or the submission never reaches the CRM. There are fewer events when a visitor opts out of tracking or the analytics code doesn’t fire.
In the scenario at the start of this article, 131 form events correspond to 90 form records in the CRM (the other 24 of the 114 records came from calls, texts, and chat). The ratio by itself proves nothing; what matters is whether it was the same before. If the gap appeared after a certain date, check what changed then: the form, the analytics code, or the website-to-CRM integration.
To compare individual submissions, the form can generate a unique submission ID and pass it to both the analytics event and the CRM record. Phone numbers, email addresses, and other contact details must not be sent to analytics: Google Analytics policies prohibit it. Matching by contact details is possible only within your own systems, for example between the website’s form log and the CRM.
How to keep sales periods from getting mixed up
Time passes between a lead and a payment: a day for a simple product, months for a complex service. So “September sales” can mean one of two things:
- the calendar view: all payments received in September, regardless of when the lead came in; this is how the accounting system counts, and it’s the right view for tracking cash;
- the cohort view: leads created in September and what happened to them by a certain date; this shows the results from a specific batch of leads and how they relate to the cost of acquiring them.
Illustrative example: a lead cohort
Illustrative example. The figures are hypothetical, for illustration only.
In September, 100 valid leads were created. As of October 30, the observation date you’ve set, 20 paid orders are linked to them, one per lead, totaling $200,000; there are no refunds. Ad spend to acquire the cohort is $30,000: under this example’s rule, all September ad spend is assigned to the cohort.
| Calculation | Formula | Result |
|---|---|---|
| Share of leads with a paid order | 20 / 100 × 100% | 20% |
| CPL for the cohort | $30,000 / 100 | $300 |
| Ad spend per paid order in the cohort | $30,000 / 20 | $1,500 |
| Average payment per order in the cohort | $200,000 / 20 | $10,000 |
What follows from this:
- Don’t divide calendar payments by the cohort’s leads. September receipts include payments for leads from July and August but exclude payments for September leads that arrive in October. Divided by 100 leads, that figure would come out higher or lower than the cohort’s actual result and would tell you nothing about the cohort.
- $200,000 isn’t profit: the example leaves out COGS, fulfillment costs, taxes, and revenue recognition rules.
- Assigning all of the month’s spend to the cohort is a simplifying assumption. Some people may have seen an ad in August, and some leads may have come in without ads; in that case $300 is a blended CPL, and the report should state the rule.
Cutoff date
While some deals are still open, a cohort’s results depend on the day you count them. You can’t compare the September cohort, checked on October 30, with the August cohort, checked on November 30, without a caveat: the August cohort has had about two months longer to convert. Compare cohorts with the same observation period, such as 30 days after the end of the month they were created in, or state the cutoff date next to every number.
The practical consequence: cohort results always lag. It helps to publish the report in two stages: September metrics in early October, then an update with the September cohort’s results after October 30.
How to read the report: an illustrative example
Row definitions are in the glossary above; sources are listed below the table.
Illustrative example. The figures are hypothetical, for illustration only.
| Metric | September | August |
|---|---|---|
| Website sessions | 18,400 | 17,900 |
| New CRM records before cleanup | 114 | 105 |
| Excluded as duplicates and spam¹ | 14 | 9 |
| Valid leads | 100 | 96 |
| Qualified leads | 62 (62%) | 65 (67.7%) |
| Leads with an unknown source | 18 (18%) | 7 (7.3%) |
| Ad spend | $30,000 | $28,800 |
| Blended CPL² | $300 | $300 |
| Cohort: leads with a payment by day 30³ | 20 (20%) | 19 (19.8%) |
| Cohort: payments received by day 30³ | $200,000 | $186,000 |
| All payments received in the calendar month⁴ | $410,000 | $395,000 |
| Search impressions and clicks⁵ | reported separately | reported separately |
- The duplicate rule may have differed between August and September; see Decision 1.
- See the derived metrics table.
- September cohort cutoff: October 30; August cohort cutoff: September 30; no refunds.
- Includes payments for leads from previous months.
- Search Console, Pacific Time; not combined with the rows in this table.
Sources: GA4 for sessions; the CRM for records, exclusions, valid and qualified leads, and unknown sources; ad platforms and accounting for ad spend; the CRM and the accounting system for cohort rows; the accounting system for calendar payments. September data is complete as of October 30.
Arithmetic check: 114 − 14 = 100; 105 − 9 = 96; 65 / 96 ≈ 67.7%; 7 / 96 ≈ 7.3%; $30,000 / 100 = $300; $28,800 / 96 = $300; 19 / 96 ≈ 19.8%.
What you can decide from this report
- Based on the currently reported figures, the share of leads with an unknown source rose from 7.3% to 18%. That’s a reason to check source tracking and whether the counting rules stayed the same, and it’s the first row to look at. The September channel breakdown for leads is incomplete. Calculate the share of unknown sources among the cohort’s paid orders separately; it isn’t necessarily the same as among leads. Don’t reallocate budget between channels until the source data is reconciled.
- More duplicates and spam were excluded: 9 → 14. Possible causes include bots, a new form without spam protection, or a different approach to flagging duplicates.
- Blended CPL didn’t change: $300. It also depends on leads that came in without ads, so it doesn’t show the cost per lead for individual channels. This row gives no reason to change the overall budget, but it doesn’t confirm that the channels are performing well either. The comparison holds only if the duplicate rule was the same in both months.
- Cohorts with the same observation period: 20% vs. 19.8%. Again assuming the duplicate rule didn’t change, there’s no sign that a smaller share of leads is converting to paid orders.
- The share of qualified leads fell from 67.7% to 62%. With about a hundred leads, a difference this size could easily be random, or it could reflect a change in criteria. It calls for a look at rejection reasons, not a budget decision.
What the report doesn’t show:
- the breakdown of payments by channel, which isn’t included in the example and needs a separate completeness check: an unknown source for 18% of leads doesn’t mean the same gap exists among paid orders;
- whether the month was profitable, because COGS and other expenses are missing;
- a “conversion rate” for September leads based on calendar payments, because the $410,000 includes money from earlier months.
Two decisions for the business owner
Decision 1. Agree on definitions. Approve the duplicate rule and the criteria for a qualified lead, add them to the glossary with a date, recalculate August and September under the same rules, and recheck every conclusion that depends on the number of valid leads, including the share of unknown sources. Who: the head of sales and the marketing manager; the business owner approves.
Decision 2. Fix source tracking before changing the budget. Find out where the 18 leads without a source came from: whether UTM parameters from new forms and campaigns reach the CRM, whether CRM fields changed, and whether calls without source tracking were added. Who: the marketing analyst and the developer. The budget decision comes after this check.
What to do when numbers move
When a metric moves noticeably away from its usual level, check the data first and business causes second. Our recommendation: until the data is reconciled, postpone decisions that rely on questionable metrics, because decisions based on bad data are hard to reverse. Other decisions can go ahead based on verified data.
- Fewer valid leads. Data: forms and CRM integration, duplicate rules, all lead channels. Then: sessions by source, campaign changes, seasonality, website changes. Who checks: the marketing analyst and the marketing manager.
- Lower share of qualified leads. Data: whether the criteria or the people applying them changed. Then: new campaigns and audiences, form fields, rejection reasons in a sample of leads. Who checks: the head of sales and the marketing manager.
- Higher cost per lead. Data: in the numerator, currency, taxes, and new channels; in the denominator, the valid-lead rule, duplicates, and, for blended CPL, the share of leads that came in without ads. Then: cost by channel, bids, and budgets. Who checks: the marketing manager and the marketing analyst.
- Fewer payments. Data: calendar payments or cohort, cutoff date, cancellations and refunds, CRM-to-accounting matching. Then: longer deal cycles, pricing and terms, sales team workload. Who checks: the controller and the head of sales.
- A higher share of unknown sources. Data: UTM parameters in new campaigns, changes to forms and CRM fields, calls without tracking, a sample of leads without a source. Then: new channels without UTM tagging rules. Who checks: the marketing analyst and the developer.
Who updates the report and checks the data
A report without an owner gradually drifts from reality: definitions change, new channels don’t get added, and nobody catches errors.
| Role | Responsibility |
|---|---|
| Report owner (marketing analyst or head of marketing) | Compiles the data, maintains the glossary, flags data status, confirms the report is ready |
| Website data owner | Analytics setup and passing source data through forms to the CRM |
| CRM owner (head of sales or CRM administrator) | Statuses, duplicates, qualification, rejection reasons |
| Marketing manager | Complete ad spend across all channels, currency, and scope |
| Controller or bookkeeper | Payments and refunds, reconciliation with the CRM |
| Business owner | Reads the report, makes decisions, approves changes to definitions |
The report is ready to read when data from every source is loaded for the full period, totals are reconciled, incomplete metrics are flagged, the cohort’s cutoff date is stated, and any definition changes are noted.
How often to update it depends on data volume and your decision cycle. If the budget is reviewed monthly and you get several hundred leads, a monthly report is enough. If you get only a few leads a week, week-to-week changes will mostly reflect random fluctuation. Operational monitoring of lead follow-up can run more frequently, separate from the owner’s report.
Next steps
- The business owner chooses the decisions the report should support and appoints a report owner.
- The report owner builds the glossary from the tables above and agrees on definitions with sales, marketing, and finance.
- The website data owner and the CRM owner check whether lead source data is being captured and record the current share of unknown sources.
- The report owner fills in the template for the last full month; at the first review, the business owner determines which decisions can be made now and which need reconciliation first.
The report is set up when definitions are approved in writing, every row has a source and an owner, calendar payments are kept separate from cohort results, and everyone knows who checks the data, and in what order, when a number moves.
Let’s discuss which website and sales data you need to bring together for your decisions.
