Build vs. Buy Is the Wrong Frame: It's Build vs. Ship

The build vs. buy debate is a way of feeling productive while avoiding the actual decision.

I have sat in more of these conversations than I can count. A head of marketing at a 200-person SaaS company wants personalized video on the demo page. Engineering is interested. Someone in the room says "we could build this." Someone else says "or we could use a vendor." Both sides find research. The meeting ends with an action item to evaluate options. Three months later, Q3 planning resets the priority stack and the meeting never happens again. The demo page still has a static screenshot above the form.

Here is the actual question: are you shipping this quarter or not?

Engineering Time Is Not Free

The build-vs-buy framing treats engineering time as close to free, or at least as a variable you are not spending on anything more important. In my experience, that assumption is wrong in every company that has ever made it.

Consider what "build" actually means for personalized video infrastructure: a system that identifies the visitor, selects the right clip, streams it with acceptable LCP, records the engagement event, passes it to your CRM, and connects the watch to a deal outcome. Each piece is solvable. The integration surface is the problem. You are not building a video player. You are building the decision layer, the attribution pipeline, the content management layer, and the CRM connectors. Wiring just the video engagement event to HubSpot, with no personalization layer, takes about 25 minutes with a direct webhook. The full personalization and attribution stack is a different project by an order of magnitude.

According to Wistia's 2025 State of Video Report, 71% of companies now handle video production in-house. What that number does not capture is the operational ceiling most of those teams hit within two quarters: they can produce video, but they cannot instrument it, personalize it, and attribute it simultaneously without significant infrastructure investment.

The teams I have watched try to build this in-house ship a video embed. Sometimes they ship an A/B test. They rarely ship a closed loop between the video, the variant selection, the engagement signal, and the CRM record. That last part is the part that actually justifies the investment.

Ship Is the Metric. Everything Else Is a Prerequisite.

The frame I use is not build vs. buy. It is: what is shipping in the next 90 days?

Personalized, attributed video on your demo page either ships or it does not. If it ships, it starts to compound. Every week it runs, you accumulate data on which variant is moving which segment. The data model that makes personalization attributable takes time to populate. A system that is not live is not accumulating anything.

This is the same logic engineers use when deciding whether to roll a custom auth system or drop in an off-the-shelf provider. It is almost never about capability. Your team can build it. The question is whether building it is the best use of the next 800 engineering hours, and whether the team will still own that codebase in 18 months or will have moved to other priorities.

The buy option is not about capability. It is about optionality. You ship now. You learn what the data says. You iterate on the personalization logic and the clip set based on actual conversion signals, not hypotheses. If the vendor eventually stops serving you, you have 18 months of data to inform what a custom build actually needs to do. That is a much better specification document than a requirements-gathering session.

This is why harloop ships the personalization engine, the attribution stack, and the content studio as a single system. The integration problem that "build" underestimates is the same problem harloop removes from the critical path. These are not three products bolted together. They are the three things that need to work in concert for personalized video to actually move pipeline.

The honest version of the build vs. buy conversation is this: if you could ship personalized, attributed video on your demo page this quarter, would the outcome be worth it? If yes, the build timeline is the constraint, not the buy decision. If no, ask why the outcome changed, because the case for the thing is the same either way.

The demo page is not waiting. Neither is the pipeline.