Your Systems Don't Talk to Each Other
Data lives in silos. Manual exports. Copy-paste workflows.
Monday morning, three different numbers
I was in a pipeline review with a client last year. Marketing reported 312 MQLs for the month. Sales said they’d received 187 leads. The CEO’s dashboard showed 241.
Same company. Same month. Three systems, three truths.
Nobody was wrong, exactly. The CRM counted one way, the marketing platform another, and the BI tool had its own logic. But the meeting turned into a 40-minute argument about whose number was real instead of a conversation about what to do next.
The cost nobody calculates
I’ve seen a marketing coordinator spend every Monday morning exporting CSVs from the ad platforms, reformatting columns, and uploading them into HubSpot. Four hours a week. Over 200 hours a year of a smart person doing robot work.
That’s the visible cost. The invisible one is worse.
When getting a report requires pulling from four systems and hoping the numbers align, people stop checking. Decisions get made on gut feel. Leads slip through because the handoff between tools is a manual step someone forgot on a busy Friday.
The most expensive integration is the one you’re doing manually every week without realizing it’s an integration problem.
What good actually looks like
When systems are properly connected, something happens in one place and the others know immediately. Not after tonight’s batch job. Not after someone remembers to export. Now.
Every piece of data has one home. Other systems reference it, they don’t duplicate it. That’s how you end up with one number in the pipeline review instead of three.
And the automations actually work. A lead hits a score threshold, the right sequence fires, the sales rep gets notified, the activity logs in the CRM. No human triggering required. No “can you send me that spreadsheet” Slack messages.
The patterns I see most
The CRM-to-marketing-automation connection is the classic one. HubSpot to Marketo, Salesforce to Pardot, whatever the combination. It’s almost always half-done. Contacts sync, but lifecycle stages don’t. Or they sync with a 24-hour delay that makes lead routing useless.
Ad platforms to CRM is the second. LinkedIn, Google, Meta — they all generate leads, but attribution dies somewhere between the click and the closed deal. Fixing this single connection often changes how a company thinks about channel spend.
Then there’s the website itself. Form submissions, page views, product usage events — all of it should flow to the systems that need it. Most of the time, it flows to Google Analytics and stops there.
I use whatever gets the job done. Native integrations when they work, Zapier or Make for simpler automations, custom API work when the standard tools fall short. I don’t push a particular stack.
How I work on these
I don’t follow a numbered playbook. But there’s a shape to it.
I start by mapping what exists. What talks to what, what’s manual, where data gets stuck or corrupted. This usually surfaces the real problem, which is rarely what the client thought it was going in.
Then I design the target state and build toward it one connection at a time. Incremental, tested, documented. Because the worst thing you can do with integrations is build five at once and then try to figure out which one broke your data.
Documentation matters here more than anywhere else. Six months from now, someone will need to understand why lead source maps this particular way. If it’s not written down, it might as well not exist.
Integration work can be a fixed project for a defined scope, or part of an embedded engagement for ongoing systems work. If your Monday mornings involve more CSV exports than strategy, let’s talk about it.