Halisi Marketing is a marketing agency serving Israel and the US — two languages, two audiences, two reading directions. When we took the brief in 2024, their site was static, outdated, and missing the one thing a marketing agency needs most: a clear answer to "why you?" We rebuilt it end-to-end in seven days. These are the three cuts that made that possible.
The brief did not say "do it in a week." The timeline emerged from the scope. The founder needed the site live before a pitch campaign. We cut to fit that window and did not apologise for it.
What did the brief actually ask for?
A full-stack rebuild. Four surfaces: home, services, work, contact. Custom homepage animation. Bilingual — Hebrew and English — because Israeli clients read right-to-left and US clients do not.
We could have added a blog, a headless CMS, a testimonials carousel, a case studies grid with filters, social feed embeds. We declined all of it.
The cuts were not painful. They were obvious. Every feature we removed was one the founder had not asked for explicitly. We shipped what the brief required and nothing adjacent to it.
Why was the copy the hardest deliverable?
A marketing agency's homepage is a demonstration, not a description. Every claim about "strategic approach" or "proven process" is indicted by its own presence on the page. The prior site made those claims. Prospects came to it and found a marketing firm that could not market itself.
We rewrote the copy before we wireframed a single page. The brief said "rebuild the site." We heard "fix the message first, then build the container around it."
The rewrite took one day. The wireframes took half a day after that. Sequence matters.
What did the bilingual build actually add?
Less than expected, with one exception.
Hebrew is right-to-left. The layout had to mirror — navigation, button alignment, typography, grid anchoring. We used CSS direction: rtl with logical properties (margin-inline-start instead of margin-left) so a single component tree renders correctly in both languages without duplicated markup. One codebase, two direction states, no conditional render paths.
The exception: the homepage animation. We had designed it for a left-to-right reading axis. In RTL, elements that sweep in from the right read as sweeping out. We rebuilt the entrance direction in the final two days of the build — which, in a seven-day timeline, meant late nights.
If we do this again, the animation concept goes through a one-hour RTL review before a frame is produced.
What did we cut, and would we cut it again?
Three features did not make it into the launch build:
- CMS for the work section. The founder wanted to self-publish new projects without a developer. Adding a headless CMS to a seven-day build doubles the architecture surface. We told them: ship static, see how often you actually update it, add the CMS if the answer is "monthly." It has been over a year. The CMS was never requested.
- Testimonials section. No client testimonials were ready at launch. A placeholder section reads as emptiness — which is a worse signal for a marketing firm than having no section at all. We cut it with a handover note: add it when you have three, each with a photo and a name.
- Blog. The founder mentioned it in the brief. We asked how often they planned to publish. The answer was vague. A marketing agency's blog with zero posts is a negative signal. We shipped without it.
Every cut held. Not one was reversed in the year after launch.
What shipped on day seven?
A full-stack site in Hebrew and English. Custom homepage animation with the RTL entrance corrected. Copy rewritten across four pages. Contact form. Zero CMS. Zero blog. Zero placeholder sections.
The site went live before the pitch campaign. The founder did not have to explain the old site to a new prospect.
What would we do differently?
One thing: animation RTL review at the concept stage, not after the frames are built. We caught the problem and fixed it — but catching it late in a constrained timeline meant two nights of pressure work that a one-hour early review would have prevented.
The rest of the build holds. The cuts were right. The sequence — copy, then wireframes, then build — was right.
The discipline that made a seven-day rebuild possible is the same one that runs every engagement we take: scope what was asked for, cut everything adjacent, and ship. The one-page scope brief we use with every client is where that conversation starts — and why a week-long build does not become a month-long one.
Written 2026-07-19 by Naman Barkiya.