gematria.today
A research site we built and run. Every change to how it is found is logged with the reasoning and the expectation attached.
Proof, not promises
Every studio says the same things about quality and care. Almost none of it can be checked before you have paid. These are the parts of our work you can look at first, on this site, without asking us for anything.
We publish the prediction before the result
The easy way to look good at search work is to show the wins afterward and stay quiet about the rest. So we write the prediction down first (what we found, what we changed, and what we expect to happen) and publish it before the outcome is knowable. When it does not work, that entry stays up.
-
Shipped
The change goes live and the entry goes up with it: what we found, what we changed, and what we expect to happen.
-
While it is unknowable
The prediction sits on the page untouched, with the result that would prove it wrong written down beside it.
Measuring
-
Once it is knowable
The outcome is added to the entry it belongs to, whichever way it went. Nothing is quietly deleted.
Confirmed Wrong
stellarreach.app
Our own sales software. Same treatment, a different kind of site: an application rather than pages.
Why it is on our own sites
Because a client's traffic is theirs, not marketing material. We take the risk on properties we own, and you get to read the working.
Neither log is a highlight reel. Entries that did nothing are the useful ones, and they are the reason the rest is worth anything.
A design system, published on your own site
Most sites are styled a page at a time, which is invisible until the day somebody needs to change one and cannot tell which decisions were deliberate. Every Voyager build gets a page listing its colors, type, spacing and components (read live out of the stylesheet, so it cannot quietly go out of date) and it is yours to hand to whoever works on the site next, including someone who is not us.
Contrast · four of this site's twenty-one token pairs, worst first
| Color pair | Ratio | Verdict |
|---|---|---|
--accent-violet on --bg-3 |
3.95:1 | large text only |
--accent-cool on --bg-3 |
4.63:1 | AA |
--muted on --bg-2 |
10.84:1 | AAA |
--text on --bg-1 |
18.40:1 | AAA |
Look at ours first
The same page for this site, measured off our own stylesheet as it stands today, including the parts that are not finished.
Contrast, measured
Every text and background pairing checked against the accessibility standard, with the ratio shown. Colors that fail are listed rather than left for somebody to discover.
It grades itself
The page counts how many one-off values the stylesheet still carries and says whether that is a system or a mess. Ours says both, in different places.
A picture of every change we make
An invoice says an hour was spent. It does not say what moved. On Ascent and Mission Control, your monthly summary is generated by photographing every page at four screen widths before and after the work and comparing them pixel by pixel, so you get the before, the after, and every pixel that moved marked in color.
Nothing is taken on trust
If a page is not in the report, no pixel on it moved. That is a measurement, not a claim.
It catches our mistakes too
The same comparison is what stops a tidy-up from quietly breaking a layout somewhere nobody was looking. It is a safety net first and a report second.
Quiet months look quiet
Some months nothing visible changes: updates, patches, backups. The report says so rather than dressing it up.
We build and run our own software
StellarReach is our sales platform: multi-tenant, on PostgreSQL and serverless functions, with more than 7,000 tests. It is not a brochure site, and running it is how the habits on this page were formed. You do not write a test suite that size because it is fashionable.
The case study
How it is designed, built and operated, in more detail than a portfolio tile.
Products, not mockups
Two of the things in our portfolio are live products we own and operate. You can open them and use them.
Small studio, real engineering
One person in Austin. The tooling exists because there is nobody else to catch a mistake.
What none of this proves
It is worth saying plainly. A published search log proves we do not hide the misses; it does not promise your rankings will move, and anyone who promises that is guessing. A design system proves the site is built so it can be changed; it does not make the design right for your business. That comes from the conversation, not the tooling. A change report proves what moved; it cannot tell you whether it was worth doing.
What we think these are worth: they make it possible to check us as we go, rather than at the end when the money is spent. That is the whole claim.
See what we would do with your site
The free scorecard is the same check we run on every site we look at: how easily people find you, how it holds up on a phone, and what a visitor notices first. It comes back in a business day, and there is nothing to cancel afterward.