There is exactly one number on our work pages. Coursim relaunched with revenue up 146%, and it is printed with a line underneath saying that Coursim reported it. The other two projects have no result on them anywhere.
That is not modesty and it is not an oversight. It is a rule, and it is enforced in the data model rather than left to whoever is writing the page that week.
The rule
- A result is published only if it is verifiable from outside, or attributed to the person who reported it.
- A figure we cannot check ourselves is published with the name of whoever can.
- Where neither holds, the section is omitted entirely — not softened, not padded, not replaced with a sentence about our approach.
In practice that means the outcome field on a project is optional, and it is paired with a mandatory source field. You cannot add the first without the second. It is a small piece of typing and it does the work that good intentions do not: it makes the honest version the easy one, and the marketing version a thing you would have to go out of your way to write.
Why the attribution changes what the sentence means
“Revenue up 146%” asserted by us is a marketing claim. The same number credited to the client is a quotation, and the reader can tell which they are being given.
This is the whole argument. A prospect reading a portfolio has no way to audit any figure on it, and they know that, so they discount all of them. Every studio site says conversions improved. The discount is applied uniformly, to true claims as well as invented ones, which means the honest studio is paying for the dishonest one.
Naming the source is the only move available that survives the discount. It does not ask to be believed. It says who said it, and it lets the reader do what they were going to do anyway, which is decide how much weight a client's own number deserves.
What we deleted to keep the rule
Three screens from the ALRT case study. They were exports from the design file with the data bindings unresolved — placeholder tokens where a name and a timestamp should have been. So the one project on the site with no live URL to point at was also the one showing screens that read as unfinished.
They are gone rather than retouched. A case study can stand on its reasoning, and retouching them would have meant publishing a picture of something that never looked like that. The project keeps its problem statement and its two real design decisions, which is what we actually have.
We also removed a fabricated street address from the site's structured data — a placeholder that had been marked as an example and then quietly become the machine-readable answer to where this company is. Along with it went the coordinates derived from it, and a set of opening hours nobody had ever verified.
The cost, stated plainly
One testimonial and one number across three projects is thin, and we know how it reads next to a page of logos and percentages. The answer to thin proof is to go and get more proof, which means asking clients for things on the record. It is not to write the version we wish were true. A portfolio is a document about trustworthiness before it is a document about work, and the fastest way to fail it is to be caught being generous with yourself.
Gal Etrog · SILICA