How to See Which Content Drives Pipeline When Your Tools Don't Talk

Why can't you tell which blog posts bring in deals?

You cannot tell which blog posts bring in deals because the deal and the first visit live in two tools that never share data. Your CRM knows the deal size and close date. Your analytics knows the page and the traffic. Nothing links one row in the first tool to one row in the second.

Here is what that gap costs:

• Budget goes to the loud channel. Without proof, content loses to whatever channel has a dashboard.
• Friday becomes export day. Someone downloads CSV files from each tool and matches rows by hand.
• Each report is a one-off. The person who built it moves on, and nobody can rebuild it the same way.
• The CEO question stays open. "Which posts made us money?" gets a guess instead of a number.

I saw this on my own agency site. A 12-month Search Console review showed about 68% of its impressions came from developer searches that never turn into clients. Meanwhile, one section of profession pages converted at eight times the site average. I had been counting traffic. The useful question was which traffic.

The data is rarely missing. It is just never in one table.

How to connect content to revenue by hand, step by step

Connecting content to revenue by hand takes six steps: tag your links, store the first page on each contact, export deals, export pages, join the two on the URL, and add up the money. You can do all of it in a spreadsheet. Do it once by hand so you know what a clean result looks like.

1. Define revenue. Use the amount on deals marked closed-won in your CRM. Ignore open deals for now, because they may never close.
2. Tag every link you share. UTM codes are short labels added to a link, such as utm_source=newsletter. They tell your analytics where a click came from. Google explains each one in its guide to campaign URLs with UTM parameters.
3. Store the first page on each contact. Make sure your CRM saves the first page a person landed on, plus its UTM values, when they fill in a form. If that field is empty, attribution is impossible for that contact.
4. Export closed deals with their contacts. You want one row per deal with the deal amount, close date and contact email. The HubSpot API docs show how deals and contacts link if you later automate this.
5. Export your pages. Pull a list of your pages with their traffic from GA4, and their top keywords from your SEO tool if you have one. Ahrefs sells API access on top plans; the Ahrefs developer docs list what is included.
6. Clean the URLs. Make every URL lowercase, remove https://, www. and any trailing slash, and strip the UTM part. Two versions of the same page will not match otherwise. This step causes most broken reports.
7. Join the tables. A join means matching rows from two lists that share a value, here the page URL. In a spreadsheet, use XLOOKUP. In Python, the pandas merge function does the same thing in one line.
8. Pick one attribution model. First touch gives the deal to the first page. Last touch gives it to the last one. Google's guide to attribution models explains the options. Choose one and keep it every month.
9. Add up revenue per page. Sum the deal amounts by page. Count how many rows did not match, too. If a third of your deals have no page, fix step 3 before you trust the totals.

Check your tracking before you trust any of it. On October 10, 2026, I found that the analytics on my agency site had collected nothing for a while. The script loaded, but its commands were in the wrong shape, and a browser check counted 0 collect requests. A join on empty data gives you a clean-looking table of zeros.

The first time takes a while. The real problem is week two, when you have to do it all again by hand and nothing has changed except the date.

I wrote about which marketing jobs are worth automating, and which are not, in Claude Code skills for marketing. Attribution sits firmly in the "automate it" pile, because the steps never change.

What click-based attribution cannot see

Some buyers read your post in a Slack thread, a forwarded email or a private message, then type your address in later. Analytics records that as a direct visit, so no page gets credit. Two habits help:

• Ask on the form. Add an optional "How did you hear about us?" field and store the answer on the contact. Compare it with the tracked first page each month.
• Report two numbers. Keep "sourced" revenue, where content was the first touch, apart from "assisted" revenue, where a buyer read a page later in the deal. Mixing them makes content look better or worse than it is.

Read the totals as a trend over months, not as exact money per post.

The faster way: Python scripts that join HubSpot deals to GA4 pages

The faster way is to let a script pull the data from each tool's API and run the same join on a schedule. MarketFlow AI is 11 systems of Python scripts. The one built for the content-to-deal question is revenue_attribution.py in the revenue intelligence folder. It reads closed-won deals and the pages each buyer viewed from HubSpot, and page traffic from GA4.

The bar is low but real: you need to run a script in a terminal and put API keys in an environment file. I like this approach because no-code tools kept failing at the one step that matters, the join across whole tables. Ahrefs and Gong come in through a separate client report script. Apollo is only used by the outbound scripts, to search for leads, not for attribution.

How to use it, step by step

1. Buy and download. Gumroad sends the files right after payment.
2. Install Python 3.9 or newer, then run pip install -r requirements.txt inside each folder you plan to use. Each folder has its own requirements file.
3. Run the attribution script once with no keys. It prints a warning and builds a demo report from built-in sample data. That shows you the format before you connect anything.
4. Add your keys. Set a HubSpot private app token, your GA4 property ID and a GA4 service account file as environment variables.
5. Run the report. Use python revenue_attribution.py --report and pick a model with --model, such as first-touch. Add --json and --output to save the result as a JSON file.
6. Schedule it weekly with cron or Windows Task Scheduler, so the report is ready on Monday.
7. Check for gaps. Run it with --gaps to see which funnel stages have no content working for them.

Two cautions from reading the code. If a key is wrong, the script warns you and falls back to the sample data, so read the warnings before you trust a total. It also matches HubSpot page addresses to GA4 page paths exactly, with no cleanup, so do step 6 of the manual method first.

If the report shows that content brings deals but outbound does not, the outbound side needs its own fix. B2B Growth Machine covers targeting and cold email setup.

What the output looks like

The report prints in your terminal. It opens with a summary: total revenue, number of deals, average deal size, how many pieces of content got credit and average touchpoints per deal. Then comes a table of revenue by content type, such as blog, case study and landing page, with sessions, piece count and average revenue per piece. Each page gets a row with its credited revenue, sessions and deals touched.

How much time it saves

I have not timed a full run against my own CRM, so I will not give you a number. The bigger win is trust. When the same script runs every week, a change in the numbers means something changed in the business, not in how someone built the spreadsheet.