the short answer
Marketing and sales numbers disagree because each tool counts differently, uses its own attribution window and gets copied around in manual exports. The fix is to define each metric once, make the CRM the record of revenue, bring every source into one warehouse and report from a single dashboard. Then give every definition an owner so it stays agreed.
Why do marketing and sales numbers never match?
Each tool answers a slightly different question. The ad platform counts conversions it believes it caused, Google Analytics counts sessions and events on your site, and the CRM counts contacts and deals that someone entered. Ask all three how many leads you had last month and you will get three honest, different answers.
Attribution windows make it worse. An ad platform may credit a sale to an ad clicked weeks earlier, while your analytics tool credits the last channel before the purchase. Both are counting the same sale, so add them up and you have counted it twice.
Then there are manual exports. Someone downloads a CSV, filters it, pastes it into a spreadsheet and adds a formula. A month later nobody remembers which filter was used, and the spreadsheet has quietly become its own version of the truth.
Define each metric once
Before touching any tool, write down what each important number means. What counts as a lead? Does a newsletter sign-up count, or only a demo request? When does a lead become qualified, and who decides? Is revenue counted when the deal is signed, invoiced or paid?
Keep these definitions in one short document that marketing, sales and finance have all agreed to. It does not need to be long. A table with the metric name, a plain definition, the source system and the owner is enough to end most arguments before they start.
- Lead: a named contact who asked to hear from you, with the form or channel recorded.
- Qualified lead: a lead that meets criteria sales has agreed to, marked by a field in the CRM.
- Opportunity: a qualified lead with a deal created and a value estimated.
- Revenue: closed-won deal value, or invoiced amount, as agreed with finance.
- Cost per qualified lead: spend from the ad platforms divided by qualified leads from the CRM, for the same period.
Get the tracking right at the source
Google Analytics 4 should track the events that match your definitions: form submissions, bookings, purchases, and the key steps before them. Name events consistently and pass the same identifiers (such as a lead ID or click ID) through to the CRM, so a website visit can later be joined to a deal.
Browser-based tracking misses some activity because of ad blockers, consent choices and privacy features. Server-side tracking, where your own server sends conversion events to analytics and ad platforms, can make the data more complete and gives you more control over what is shared. It still needs to respect consent, so set it up with your consent banner, not around it.
Most importantly, make the CRM the record of revenue. Ad platforms and analytics tools can estimate what led to a sale, but only the CRM knows which deals closed and for how much. Send those outcomes back to the ad platforms as offline conversions, so they optimise for customers rather than form fills.
Do you need a data warehouse?
Once you have more than a few sources, a warehouse is usually the simplest way to get one view. A warehouse such as BigQuery holds raw data from your ad platforms, analytics, CRM and billing tool in one place, refreshed on a schedule by automated connectors rather than manual exports.
In the warehouse you apply your agreed definitions once, as shared logic that every report reads from. If the definition of a qualified lead changes, you change it in one place and every dashboard updates. That is the practical meaning of one source of truth.
A BI tool such as Looker Studio or Power BI then sits on top. Build one main dashboard that answers the questions leadership actually asks: what did we spend, what did it bring in, and what changed since last week. Add detail pages for each team, but keep them reading from the same tables.
Automate the reports, then add AI summaries
Once the dashboard is trusted, stop rebuilding the weekly report by hand. Schedule it to be sent to the people who need it, with the same layout every week so changes are easy to spot.
AI summaries can help here, as long as they work from the same governed tables. A short written note that says which numbers moved, by how much, and which campaigns or segments drove the change saves people from having to read every chart. Keep the summary tied to the data underneath it, and have someone sanity-check it until you trust how it reads the numbers.
Who should own a metric definition?
Every definition needs a single named owner, usually the person whose decisions depend on it most. Sales might own the definition of a qualified lead, finance might own revenue, and marketing might own channel and campaign naming.
Owners approve changes, and changes are logged with a date, so a jump in a chart can be traced to a change in definition rather than a change in the business. Review the definitions every quarter, and retire any report or spreadsheet that does not read from the warehouse.
- One document listing every metric, its definition, source and owner.
- A change log for definitions and tracking, with dates.
- Access rules for who can edit the shared logic and who can only view.
- A quarterly review where owners confirm their definitions still hold.
Where to start this month
Start with the argument you have most often. If marketing and sales disagree on lead numbers, define a lead and a qualified lead first, fix the CRM fields to match, and build one chart that both teams accept. That single agreed number does more for trust than a large dashboard nobody believes.
From there, connect the next source, apply the next definition and retire the next spreadsheet. If you would rather have help with the warehouse and dashboard build, WhiteThread sets these up, but the definitions document is worth writing yourself either way.