Most brand drift does not happen because somebody ignored the brand guide. It happens because nobody ever held the guide up against the work.
Try this before you commission anything. Open the folder where your last twenty published pieces live. The proposal, the conference one pager, the rate sheet, the deck a colleague rebuilt because he could not find the original file, the three social graphics, the PDF that went to a printer in February. Put them all on one screen at thumbnail size.
Now count. How many distinct blues. How many typefaces. How many versions of the logo, including the one with the tagline nobody uses anymore. How many different ways the phone number is formatted.
That count is your brand. The document saying you have one blue is a wish.
A guide describes intentions. The files record decisions.
Every brand guide I have opened was written at the beginning, when the brand was an idea and the work did not exist yet. It is a statement of intent, and intent is fine as far as it goes.
The deliverables are a different kind of object. Each one records a decision somebody actually made under time pressure, usually without the guide open, often in a tool that did not have the right font installed. The invoice template took its color from whatever the accounting software offered. The conference handout took its type from what the printer could run. Those are the real decisions, and they are the ones your customers have already seen.
When the document and the files disagree, the files win. They have been in front of people. The document has not.
I learned this on my own business, which is the embarrassing way to learn it. I had a documented palette and a documented type pairing, both written down, both confidently wrong. I rebuilt a template from them and it looked nothing like my own work. The palette I actually use had never been written down anywhere. It only existed in the files.
Why counting beats redesigning
The instinct, when a brand feels inconsistent, is to start over. New logo, new palette, new guide. That is the most expensive response available and it usually does not hold, because the thing that produced the drift is still in place. Nothing reconciles the document against the output.
Counting is cheap, and it tells you which problem you actually have.
Two blues and one stray font is not a brand problem. It is a file problem, and it is an afternoon of work.
Six blues, four typefaces and three logo versions still may not call for a redesign. What that usually calls for is a decision about which one is right, and then the discipline to make the other five disappear from every folder anyone can reach.
And if the strongest looking piece in the folder is the one that matches the guide least, that is the most useful finding available to you. Your instincts already chose. They just chose somewhere other than the document.
Hand people a finished thing, not a list of values
Here is the part most brand systems get wrong, and it is a small mechanical thing with an outsized effect.
A list of hex codes and font names is not usable by the person under pressure at four in the afternoon. It asks them to design, and designing is the part they were trying to avoid.
What works is one canonical finished piece. A real document, built correctly, that anyone can open and copy. Not a mood board and not a specimen sheet. A finished proposal, a finished one pager, a finished report, with the type sizes, the margins, the header treatment and the way a photograph gets handled all decided already and visible on the page.
Then the instruction stops being "use our brand" and becomes "start from this file." Those are very different asks. People do the second one.
I keep exactly one of those for my own studio, and every new piece starts there. When the system changes, that file changes the same day, and everything built after it inherits the change. The written notes point at the file. They never restate what is in it, because restating it is how the two drift apart in the first place.
What a spec has to contain to be worth writing
Four things, and it is genuinely this short.
The exact values, in a form somebody can copy rather than retype. A hex code that has to be squinted at inside a PDF will get approximated, and approximated is how a second blue is born.
The rules that say no. What the logo never sits on. What the display face is never used for. Negative rules do more work than positive ones, because they catch the cases nobody anticipated.
The one canonical build, named, with the place it actually lives.
And who decides. A brand with no named owner gets amended by whoever is in the biggest hurry.
The step that stops it happening again
Put the count on a schedule. Once a quarter, pull the last twenty pieces and look at them together. Fifteen minutes. The entire value of it is that nobody has to remember to notice.
Drift is never dramatic. It does not arrive as one bad decision. It arrives as twenty small reasonable ones, each made alone, each defensible on its own, and the only place they are visibly wrong is next to each other.
So put them next to each other.
If you want that audit run, the canonical file built, and the whole thing kept from sliding again, that is the gap I close. Start here: hazelpipercreative.com/contact