2 project slots left this quarter. 30-minute call, no deck.
Book a callWe've been building for the web since 2016. In that time the single most valuable thing we've learned is that most projects fail before anyone opens a code editor — during the conversation where nobody agreed what the work was supposed to change.
KatalystAI started in 2016 between an engineer and a marketer who kept getting hired to fix each other's work. The engineer inherited beautiful sites that ranked for nothing. The marketer inherited campaigns pointed at pages that took five seconds to load.
The pattern was always the same: two competent teams, no shared definition of success, and a client paying for the gap between them. We built the agency to close that gap — one team accountable for the traffic and the page it lands on.
Nine years later we're still deliberately small. Four to six engagements at a time, senior people only, and a habit of turning down work we don't think we'd do well. That last part costs us revenue every quarter and it is the reason most of our work arrives by referral.
Not aspirations on a wall — these each cost us something, which is how you can tell they're real.
If a project is scoped wrong, mispriced or aimed at the wrong problem, week one is when that gets said. Week nine is when it becomes expensive.
We run four to six engagements at a time. That ceiling is deliberate — it's the number at which the people who sold the work can still be the people doing it.
We record a baseline before any work starts. It's the only way to distinguish a result from a coincidence, and it occasionally proves us wrong.
Standard tools, documented decisions, no clever abstractions that only make sense to their author. Being replaceable is a feature we ship on purpose.
No pitch team, no delivery team. The names below are who you meet on the call and who does the work.
Principal Engineer
Architecture, performance, platform
Head of Growth
SEO, paid acquisition, attribution
Engineering Lead
Applications, data, integrations
Operations Director
Delivery, VA programme, QA
Conversion Strategy
CRO, research, copy
Design Director
Product design, brand systems
Roughly one in five discovery calls ends with us recommending something smaller, cheaper, or somebody else entirely. It costs us revenue in the short term and it is the single reason our referral rate is what it is.
You get a number and a date before work starts, plus a written list of what's excluded. Change requests are re-quoted in writing. No hourly meter, no quiet overruns discovered at invoice time.
The people in your kickoff call are the people writing the code and the copy. We don't run a pitch team and a delivery team, and we don't staff projects with whoever is on the bench.
Lighthouse and accessibility budgets run in CI from the first commit. A build that regresses past the threshold doesn't merge. That's why our median LCP at launch is 0.9 seconds.
Repository, domain, analytics, ad accounts, CMS. All in your name from day one. Leaving us should cost you nothing but the notice period.
Not impressions, not deliverables shipped, not hours burned. Pipeline, qualified leads and revenue — measured against the baseline we recorded before we started.