{"id":23747,"date":"2026-09-02T15:55:50","date_gmt":"2026-09-02T15:55:50","guid":{"rendered":"https:\/\/scannn.com\/a-guide-for-marketers-in-the-vibe-code-era\/"},"modified":"2026-09-02T15:55:50","modified_gmt":"2026-09-02T15:55:50","slug":"a-guide-for-marketers-in-the-vibe-code-era","status":"publish","type":"post","link":"https:\/\/scannn.com\/lv\/a-guide-for-marketers-in-the-vibe-code-era\/","title":{"rendered":"a guide for marketers in the vibe code era."},"content":{"rendered":"\n
\n

The exponential number of new vendors popping up on the annual marketing technology landscape map is getting out of hand:<\/span><\/p>\n

\n
<\/p>\n
<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

And yet, we\u2019re all shipping internal tooling in parallel<\/span>.<\/span><\/em> AI adoption was likely a top-down priority from the CEO, Founder, or CMO this year, but how do we ensure we don\u2019t create a tech debt problem, spend tokens\/time on vibe coded tools only one team member uses, and also keep the productivity bar high?<\/span><\/p>\n

Experimenting and solving your most annoying problems with AI is always a good thing. And getting the whole team AI-pilled through this method is great, but there\u2019s so many instances where a project simply shouldn\u2019t be reached for bc it\u2019d be better driven with an engineer or outsourced and bought.<\/span><\/p>\n

\n

At Sentry, everyone in marketing is operating with their own skills, their own tools, their own hosted projects. A handful of these skills and tools are duplicative of what other people have or software that already exists in the stack. We\u2019re now starting to converge on what should be shared, how it should be, and what we\u2019re going to keep within our existing SaaS contracts. – Matt<\/span><\/em><\/p>\n<\/blockquote>\n

We are going to walk through our takes on what should be built vs. bought and to start, we created a framework for deciding <\/span>who<\/span><\/strong> should build internal tooling. Then, we\u2019ll list out common categories of marketing technology and share each of our POVs on whether it\u2019s worth building or buying.<\/span><\/p>\n

Creating an internal system for building can help drastically improve likelihood that tool will be maintained and used widely.<\/span><\/p>\n

\n

At Railway, we have an internal MCP that\u2019s borderline god within the organization. It holds a shared set of skills, tools, guardrails, and the like so anyone using it can create internal tooling safely. Beyond safety, it holds guidelines for things like brand design and architecture specs so our internal tooling is pleasant to look at. This setup is possible because: we have a highly technical team; we spend a lot of effort perfecting our tooling; we\u2019re still under 50 employees. – Sarah<\/span><\/em><\/p>\n<\/blockquote>\n

But knowing when to reach for eng help and knowing when to DIY could save you many hours and wasted resources\/tokens. So we came up with a general framework:<\/span><\/p>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

Occasionally a customer-facing vibe-coded tool could be <\/span>fine<\/span><\/em>. Maybe it\u2019s a survey tool or template creator that lives on your website. but customer facing tools always need more scrutiny.<\/span><\/p>\n

Ok with that out of the way, let\u2019s get into each decision on whether to build or buy a tool for marketing. We both went through 8 scenarios and separately flagged if we\u2019d build or buy that marketing software or system.<\/span><\/p>\n

    \n
  1. \n

    CRM<\/span><\/p>\n

      \n
    1. \n

      Sarah – <\/span>buy.<\/span><\/strong> an automation-forward CRM (like Attio) so you can use it as a datastore and build reporting on top of it. Salesforce may be a necessary evil at scale, as much as I hate to admit it.<\/span><\/p>\n<\/li>\n

    2. \n

      Matt – <\/span>buy.<\/span><\/strong> Salesforce, HubSpot, etc. They have so many integrations that make life easier and the customer data becomes actionable. Google ads conversion passthrough, sales tools like Gong integrate, ops workflows are automated, reporting on opportunities is out of the box, etc. Don\u2019t build this.<\/span><\/p>\n<\/li>\n<\/ol>\n<\/li>\n

    3. \n

      CMS<\/span><\/p>\n

        \n
      1. \n

        Matt – <\/span>build<\/span><\/strong>. This is one marketing technology I see dying off over the next two years. Everyone wants to edit their website in code now that LLMs like Claude make this 10x easier to update pages, spin up new ones, utilize skills as templates, etc. One great engineer working to create guardrails, scalable systems, etc. will enable the rest of the team to build it all out and move faster. I wrote more on our website migration out of a CMS <\/span>here.<\/span><\/a><\/p>\n<\/li>\n

      2. \n

        Sarah – <\/span>build<\/span><\/strong> but only if you have good design AI guardrails, are ready to vibe code your website, and have an eng-forward GTM\/marketing team. Without those 3 things, your website is going to look like AI slop.<\/span><\/p>\n<\/li>\n<\/ol>\n<\/li>\n

      3. \n

        Audience data platform<\/span><\/p>\n

          \n
        1. \n

          Sarah – <\/span>build<\/span><\/strong> full stop. Any audience tool I\u2019ve used is great at first, but hits a wall after a certain amount of customization. Your ICP is your company\u2019s secret sauce and getting it right is a competitive advantage.<\/span><\/p>\n<\/li>\n

        2. \n

          Matt – <\/span>depends<\/span><\/strong>. One of the tools I built in an afternoon allows you to define your ICP, then it will create thousands of youtube video URLs that you can use as targeting in google ads for youtube ads. But tools like Clay (for B2B) exist to create ICP accounts, enrich, and find identities. So I\u2019d build what you can, then use cheap, proven tools for the rest.<\/span><\/p>\n<\/li>\n<\/ol>\n<\/li>\n

        3. \n

          BI tools<\/span><\/p>\n

            \n
          1. \n

            Sarah – <\/span>buy<\/span><\/strong> because it\u2019s a PITA to build. Just ask the Omni team how long it took them to implement all the quirks that come with editing charts, or the Hex team how long it took to get their AI query agent right. There is no competitive advantage to having your own BI tool, and the requirements are fairly standard across different organizations.<\/span><\/p>\n<\/li>\n

          2. \n

            Matt – <\/span>buy<\/span><\/strong> because you need a data eng team to correctly make a scalable database and logic behind the scenes. The business will run on that, but build your own dashboards for blind spots like social listening data, or AI visibility data. Anytime you log into a separate tool for reporting, ask yourself if you could just build something to automate that.<\/span><\/p>\n<\/li>\n<\/ol>\n<\/li>\n

          3. \n

            Emailing system<\/span><\/p>\n

              \n
            1. \n

              Sarah – <\/span>buy<\/span><\/strong>. <\/span>Customer.io<\/span><\/a> is the answer for any company pre-series C, and for most even after. Managing domain verification, email batch sends, DMARC policies, etc is also a PITA. Have a tool that manages the technicals, allows you to add your own design, and use it to your heart\u2019s desire.<\/span><\/p>\n<\/li>\n

            2. \n

              Matt – <\/span>build<\/span><\/strong>. I\u2019m a big fan of what Resend is doing and if I had to start from scratch, I\u2019d start there. I use it on my side projects and it works great for asking a coding agent \u201cspin up a retention email sequence for once a customer hits 6 months of usage. Use my design system.\u201d and it just works. Having your email templates and sequences all in code is a dream for updating. Sentry already has a well thought out and integrated system via Inflection, so I wouldn\u2019t rip it out just to vibe code something.<\/span><\/p>\n<\/li>\n<\/ol>\n<\/li>\n

            3. \n

              AEO \/ SEO tracking<\/span><\/p>\n

                \n
              1. \n

                Sarah – <\/span>buy<\/span><\/strong> tools like Ahrefs and Gauge, which are built to run tests at scale and scrape tools via integrations. Both have the advantage of using aggregate data to make better suggestions, use it to your advantage.<\/span><\/p>\n<\/li>\n

              2. \n

                Matt – <\/span>buy<\/span><\/strong>. Build your own tools\/skills\/dashboards\/views based on the tools you buy. Ahrefs\/Profound are what we use which help me automate AI visibility and SEO reporting as well as give context to my agents to update our site.<\/span><\/p>\n<\/li>\n<\/ol>\n<\/li>\n<\/ol>\n

                  \n
                1. \n

                  AB testing platform<\/span><\/p>\n

                    \n
                  1. \n

                    Sarah – <\/span>don\u2019t do<\/span><\/strong> except for when you reach a huge scale. Trust your convictions and run isolated tests on other platforms (like Google ads is a great place to test messaging).<\/span><\/p>\n<\/li>\n

                  2. \n

                    Matt – <\/span>build<\/span><\/strong> a simple directional tool if you have your website in code (before\/after analysis). Don\u2019t buy until you reach a scale that matters. <\/span>Statistical significance calculators<\/span><\/a> exist so you could test the amount of sessions and conversions you get to see if a tool would even be able to make good decisions for you.<\/span><\/p>\n<\/li>\n<\/ol>\n<\/li>\n

                  3. \n

                    Attribution system<\/span><\/p>\n

                      \n
                    1. \n

                      Sarah – <\/span>build<\/span><\/strong> – a <\/span>waterfall attribution algorithm<\/span><\/a> will get you really far. Sure, things like MMM exist, but unless you reach huge scale, you won\u2019t have the numbers to make the investment worth it. Run timed tests, fail fast, and iterate.<\/span><\/p>\n<\/li>\n

                    2. \n

                      Matt – <\/span>build<\/span><\/strong>. I worked with my BI team to write the SQL logic that we still use. I talk about ways to measure top of funnel <\/span>here<\/span><\/a> based on how we spend 70% of our budget on awareness efforts.<\/span><\/p>\n<\/li>\n<\/ol>\n<\/li>\n<\/ol>\n

                      Interestingly enough, we both did this exercise separately and we disagreed on very little.<\/span><\/p>\n