A field report on the small-business web in Kitchener–Waterloo.

On this report

We build things for people who build things.

Candid Creative is a small studio in Kitchener–Waterloo. We make websites and software for the trades, the contractors, and the manufacturers in Waterloo Region — the operations that move steel, lumber, and parts through the local economy and need a web presence that works as hard as a job site does. We publish what we learn while building. This page is a brief on the work, the research behind it, and what we have found.

In 2025, most local agencies were still selling page builders. In 2026, the gap between a well-built site and a typical one is large enough to measure from the parking lot.

Every number on the site is sourced and dated. Where the evidence is thin, we say so. The full method is in sources and methodology.

43%
Share of WordPress sites that pass Google's mobile speed test — the test Google uses to decide who shows up in search. Fewer than half make it.
Source — HTTP Archive CrUX Tech Report (field data, multi-site) · Jun 2025
81
Mobile speed score for this site, candidcreative.ca, when Google runs the same test. Desktop is 100. Mobile is the number Google uses to rank, and the WordPress field figure to the left is mobile too — honest apples-to-apples.
Source — Google PageSpeed Insights (lab data, candidcreative.ca) · May 2026
523
Days, on average, for a website's search traffic to recover after a botched move from one platform to another. Almost a year and a half.
Source — Search Engine Journal, n=892 · Jan 2025
91%
Share of WordPress security holes discovered in 2026 that lived inside add-on plugins. Only six were inside WordPress itself.
Source — Patchstack, State of WordPress Security 2026 · 2026
Chapter one
01
The stack

The tools that build websites have changed faster in the last three years than in the previous ten. Most local agencies have not changed with them.

The agency you hire is almost certainly still selling the same WordPress-and-page-builder setup it was selling in 2019. Google measures every public website on the same speed test. The setup most agencies are still selling fails that test on more than half of all real visitor sessions. Yours probably included.

We build with newer tools that send the visitor's phone or laptop much less code, on hosting the business owns a copy of — its own domain in its own name, its own files in its own account, its own customer data in its own database. Page builders push two to five megabytes of code onto every page before the actual content arrives. We do not use them.

Read about how we build →

Chapter two
02
The writing

The generated answers at the top of Google search now appear on about half of all searches. Brands quoted inside those answers earn measurably more clicks. The ones not quoted lose them.

Getting quoted is no longer about keywords. Researchers at Princeton and the Allen Institute for AI measured what the answer-generators actually pull from a page. Pages that quote real sources got pulled 41% more often. Pages with concrete numbers, 31% more. Pages that name where their information came from, 28% more. Pages that try to game the old way — cramming the same keyword in over and over — performed worse than pages that did nothing at all.

Our writing follows the same rules we apply to client work. Every number on the page names where it came from and when. Mistakes get logged on a public corrections page when we make them. The writing is the marketing.

Read the writing →

Chapter three
03
The work

The studio has been at this since 1996. Most of the work has never been published.

Case studies will land here as projects warrant it — with the client's permission, with the real names of the tools we used, and with real before-and-after numbers measured against the same speed test Google uses to rank the site. No cherry-picked benchmarks from a controlled lab; the numbers are taken from how the site behaves on real visitors' phones over weeks of real use. Until those numbers are in, the writing is the proof.

About one in three of our audits ends with a recommendation not to migrate. We say that out loud because the audit is the relationship; the rebuild is downstream of that.

See the case study slate →

Core Web Vitals are best understood as a gate, not a signal of excellence. In an AI-led search landscape, this clarity matters.
Dan Taylor · Search Engine Land · 13 January 2026
Largest publicly disclosed AI-citation study, n=107,352 pages.
Reading order
Start here
This report draws from
the Candid knowledge base
and the public research record.
Request an audit →
$1,250 · about a week · report yours
Candid Creative
Kitchener–Waterloo
Est. 1996 · 30 years

Candid Creative