Marketing demanded IT add website feature that was already working

Marketing demanded IT add website feature that was already working

The email landed in the IT ticketing system at 4:47 PM on a Friday. Subject line: URGENT: Need "Filter by Category" on Product Pages by Monday Launch. Body: three paragraphs of marketing-speak about "user journey optimization" and "conversion funnel acceleration," cc'd to the VP of Sales, the CMO, and somehow the CEO.

The feature had shipped in 2019.

It survived two redesigns, a CMS migration, and the great "let's rebuild everything in React" catastrophe of 2021. It lived quietly in the product template, documented in the component library, tested in CI/CD, accessible via a single props toggle. The frontend team had even added keyboard navigation and ARIA labels last quarter during an accessibility sprint nobody asked for but everyone needed.

Nobody in marketing knew it existed.

The institutional amnesia tax

This isn't a communication failure. It's a structural one. Marketing operates on campaign cycles — quarterly themes, seasonal pushes, product launches tied to trade shows. IT operates on product cycles — sprints, releases, technical debt paydown, platform stability. The two rhythms rarely sync, and the gap between them is where institutional knowledge goes to die.

Every organization pays an amnesia tax. The cost compounds when turnover hits. The marketer who championed the 2019 filter feature left for a Series B startup in 2020. The product manager who specced it moved to infrastructure in 2021. The engineer who built it? They’re now a staff engineer at a FAANG, and their commit history is the only documentation that matters.

New marketing hires inherit a website they didn't build, running on a stack they don't understand, supported by a team they rarely talk to. They see gaps that aren't there. They request features that already exist. They commission agencies to build what's already in production, because nobody thought to give them a map.

The shadow requirements document

Ask any senior engineer about the "shadow requirements document" and they'll laugh — bitterly. It's the unwritten list of everything the system already does that stakeholders keep asking for. Dark mode toggle? Shipped 2020. CSV export? Behind the admin flag since 2018. SSO via SAML? Configured, tested, waiting on legal to approve the vendor addendum.

Marketing doesn't read changelogs. They shouldn't have to. But someone — product, engineering leadership, a technical PM — should be maintaining a living, searchable catalog of user-facing capabilities. Not a Confluence page last updated in 2019. Not a Notion database nobody has permissions for. A real catalog. Integrated into the design system. Linked from the component library. Owned by a human being whose performance review includes "stakeholders didn't ask for existing features this quarter."

That role doesn't exist at most companies. So marketing files tickets. IT groans. Engineers waste cycles verifying, documenting, screenshotting, proving the negative. The feature gets "re-built" — worse, because the rebuild inherits none of the battle scars, edge cases, or accessibility fixes of the original.

The rebuild trap

Here's the dirty secret: the rebuild is often easier than the excavation. Digging through three years of git history, tracing feature flags through deprecated services, explaining to a skeptical marketing director that yes, the filter works, it just requires the data-category attribute on the product card — that takes political capital. Rebuilding takes sprint points. Sprint points are visible. Political capital is not.

So the rebuild happens. The new version launches. It lacks the "clear all filters" button the 2019 team added after user testing. It breaks on Safari because nobody tested the custom select component on iOS. It ships without the analytics events the old version fired, so marketing can't even measure if anyone uses it.

Six months later, a new marketing hire files a ticket: URGENT: Need "Clear All Filters" Button by Q3 Launch.

Breaking the cycle

Fixing this requires admitting that documentation is product work, not busywork. It means staffing a "platform advocate" or "internal developer relations" role — someone who translates engineering capability into stakeholder language, proactively, before the Friday 4:47 PM email arrives.

It means marketing gets read access to the component library's Storybook. It means product demos happen at sprint reviews, not just launch retrospectives. It means the CMS exposes a "what components exist" API that the marketing team's internal tools can consume.

Most of all, it means leadership stops treating IT as a vending machine. You don't get features by inserting a ticket and waiting for the snack to drop. You get features by understanding what's already on the shelf.

The filter-by-category feature still works. It handled Black Friday traffic in 2022, 2023, and 2024 without a hiccup. Marketing used it for the holiday campaign — after IT sent a Loom video showing where to toggle it on.

The video has 47 views. The ticket was closed "wontfix — already exists."

Next quarter, a new campaign manager will ask for it again. The amnesia tax is due. And IT will pay it — again — because nobody budgeted for the accountant.