The site was selling
the wrong business.
Uassist.ME runs bilingual virtual assistants and configurable BPO through nearshore teams in Latin America. They wanted recognition as a specialised BPO provider, and their site was built to sell virtual assistants, so the side of the business they were growing into had almost no room on it. We rebuilt from scratch in two weeks and handed back a stack they own outright.
- Client
- Uassist.ME
- Industry
- Nearshore BPO and virtual assistant services
- Scope
- Migration, headless CMS, technical SEO, infrastructure
- Engagement
- February to June 2026
- Build window
- Two weeks. Live 14 April 2026
- Stack
- Next.js, Sanity, Cloudflare, HubSpot API
Four problems,
three of them invisible.
The engagement began as marketing support. It became clear quickly that the site itself was the constraint, so the rebuild moved to the front of the queue. Roughly 23 pages on Webflow, and four things wrong underneath them.
URLs sitting in Search Console against a few hundred real pages. Google was crawling a site roughly forty times bigger than the one that actually existed.
- The forms did not work reliably. HubSpot ran through Webflow embeds and took ten to twenty seconds to appear, when it appeared at all. A contact page and a lead capture page failing silently is the most expensive kind of failure a B2B site can have.
- Performance was poor. Load times ran into multiple seconds and PageSpeed sat around 60.
- Search signals were broken or missing. Organization and FAQ markup were absent entirely, and the Review markup that did exist had been returning errors to Google for months.
- The index was full of noise. Junk paths, parameter variations, duplicates and deprecated routes, spread across eleven distinct categories of indexing problem and all inherited from previous setups.
What we changed,
in four moves.
- 01
Left Webflow on purpose
Webflow is a good tool right up until you need it to do something it was not designed for. Uassist.ME needed several of those things at once, so we moved them to Next.js and Sanity.
- 02
Gave the BPO side room to grow
The old site was built to sell virtual assistants, so that is what it sold. Twenty three pages became more than 45, restructured around service lines rather than one flat offering.
- 03
Fixed what was failing silently
The forms were the expensive one. Everything else on this list was invisible from the front end and just as costly.
- 04
Migrated without a gap
This is the part that goes wrong most often, so we planned it properly. Traffic transferred cleanly at cutover and the client woke up to a working site.
The rest of this is the method.
four moves are above. What sits under them is the actual work. The page-level structure, the order we fixed things in, and the checks we run before we agree to rebuild anything. Give us an email and it opens, and we will send you the pre-rebuild checklist we ran on this project so you can run it on your own site.
The numbers,
and where each comes from.
- 2 weeksbuild window, live 14 AprilOn schedule
- 23 → 45+production pagesBuild inventory
- 60 → 90sPageSpeed scorePageSpeed Insights
- PassedCore Web Vitals, mobile and desktopCrUX field data · 28 days
- 4×organic trafficGA4 · Feb to Jun 2026
- Page 5 → 2average search position, 45.5 to 18.2GSC · 3 months vs 12-month baseline
One deprecated third party API that the client chose to keep still holds the PageSpeed score down. That is a business decision rather than a build constraint, and we would rather say so than quote a number that needs an asterisk.
One view instead of six tabs
We built them a client analytics portal on DooDock. Meta, Instagram, LinkedIn, Google Ads, Search Console and GA4 land in one place, with PDF reports generated off the same data. It went in as a value add. It is the piece they mention first.
The team has been able to capture our business identity and differentiators and reflect that on the new website.
Your site probably has
its own version of this.
Every project here started with one short call and something worth fixing. Yours can too.