One Page, Four Numbers, Zero Agreement
If you run a content-heavy website — especially one built around calculators, data tools, or reference tables — you already know the quiet dread of the consistency audit. You scroll through a page you've published, and somewhere between the interactive widget and the FAQ section, the same number appears twice with two different values. Neither one is obviously wrong. Both came from research. And yet there they are, a few hundred pixels apart, contradicting each other in front of every visitor who reads carefully enough to notice.
That's exactly the situation that surfaced on costto.build, a small site of home-improvement cost calculators. Each page lets a visitor pick a project and receive a low, average, and high estimate broken down by materials and labor. Simple enough in concept — but each page was displaying the same number in four distinct places, and for a meaningful stretch of time, those four places didn't agree.
The story of how that happened, why it was essentially inevitable under the original content structure, and how one rule made it impossible to repeat is worth unpacking in detail. Because this isn't really a story about one site's internal bug. It's a story about a mistake that content-driven websites make constantly, often without realizing it until a reader calls it out in a comment or a review.
The Four Places a Single Number Can Live
Take a room-addition calculator page as the example. A visitor landing on that page encounters the same core cost figure in four separate contexts:
- An interactive widget powered by a
calculate()function that returns low, average, and high estimates based on inputs like square footage and finish level. - A static reference table above the fold showing cost ranges by project size — the kind of quick-glance data that satisfies visitors who aren't ready to use the widget yet.
- Worked examples written in prose — sentences like "a 20×20 in-law suite runs about X" that ground the abstract numbers in a real-world scenario.
- An FAQ section that answers questions like "what does a room addition cost?" with its own cost bands.
When those four elements were built initially, each one was written by hand. The widget had a formula. The table had numbers drawn from research. The prose examples had figures reasoned out during the writing process. The FAQ had estimates that felt consistent with everything else at the time of drafting. Close enough that nothing looked wrong — until a consistency pass revealed the gap.
A $5,500 Gap Hidden in Plain Sight
The discovery was specific and stark: the in-law suite worked example in the body copy stated a cost of $62,300. The interactive widget, given the exact same square footage and finish level as inputs, returned $56,779. Same project type, same specifications, same page. A difference of roughly $5,500 sitting on the page simultaneously, with no indication to the reader which figure to trust.
What made this particularly insidious was that the numbers weren't random or careless. Both came from genuine research and real reasoning. The discrepancy existed not because of a typo or an error in isolation, but because the two figures had been built on different hidden assumptions that no one had ever documented explicitly.
Why It Was Always Going to Happen
The prose example had been written with a specific assumption baked in — say, a brand-new HVAC unit as part of the scope. The calculator's formula might have assumed HVAC as an optional add-on or used a slightly different regional cost baseline. Neither assumption was wrong on its own. They were just different, and because they were never written down or reconciled, the gap between them had no reason to close over time. If anything, as the page was updated and refined across multiple sessions, the gap would only widen.
This is the core of the problem: when the same number lives in multiple places and each place derives that number independently, you don't have one source of truth. You have four competing sources of truth, each one quietly diverging from the others every time any single element is updated or revised. The widget gets a formula tweak. The table doesn't. The prose example stays frozen at the original figure. The FAQ reflects a middle ground that nobody consciously chose.
For a home improvement cost site, this kind of inconsistency carries real consequences. Visitors use these numbers to plan budgets, get contractor quotes, and make decisions about whether a project is financially viable. A $5,500 discrepancy on a room addition isn't a rounding error — it's the difference between a project that fits a family's budget and one that doesn't.
The One Rule That Made It Impossible to Repeat
The fix wasn't a more careful proofreading process, or a spreadsheet to track which number was "official," or even a better editorial workflow. All of those approaches still leave multiple independent sources of truth in place — they just add friction to maintaining them. The fix was architectural: every number that appears on a page must be derived from the same underlying source. The widget's formula becomes the canonical definition of cost. The reference table is generated from that formula, not typed by hand. The prose examples use output values from the same calculation with documented inputs. The FAQ pulls from the same formula logic.
When the formula changes, everything changes with it automatically. There is no version of the page where the worked example can disagree with the calculator, because the worked example no longer has an independent existence. It is the calculator, rendered in a different form.
What This Means for Any Content-Driven Site
The lesson scales well beyond cost calculators. Any website that presents data, statistics, estimates, or quantitative claims across multiple content formats — body copy, tables, FAQs, callout boxes, schema markup — is running the same structural risk. Each format feels like a different presentation of the same information. But unless those formats share a single upstream source, they are separate sources of truth, and separate sources of truth drift apart.
The question worth asking of any page that carries numbers is not "are these numbers correct?" but "do these numbers have a single origin?" If the answer is no — if the table was typed separately from the article, if the FAQ was written by a different contributor than the body copy, if the calculator was built by a developer who never saw the prose — then the page has a consistency bug waiting to be discovered. It may not be discovered today. It may not surface for months. But the gap is already there, growing quietly every time any one of the four sources gets touched without the others being updated to match.
Running a consistency audit is how you find the bug that already exists. Building a single source of truth is how you stop creating new ones.