Our current records support one strong quantitative finding: 987 of 2,140 lead records—46.1%—were explicitly scored with “No website” as of August 2, 2026. That is a finding about the businesses in our prospecting pipeline, not a statistic for all Canadian small businesses. The dataset is concentrated in the industries and places where we look for businesses that may need a first website or a rebuild, so selection bias is built in.
The records also document a substantial prototype workflow, but they do not support the claim that we audited 56 distinct businesses for every problem in this article. The prototype registry contains 56 rows, 44 marked completed. Those completed rows map to 28 distinct recorded lead IDs; other rows are variants or have no lead ID. We use that work to explain the checks we make, not to invent prevalence numbers the files cannot prove.
That distinction matters. “What we notice” includes a measured pipeline result, a production history, and editorial observations from reviewing small-business sites. Only the first has a denominator suitable for a percentage today. This report separates those evidence levels so an owner can see what the data proves, what it merely suggests, and what to check on their own site.
What is in the dataset
The lead file is an operating list, not a survey. It contains 2,140 records gathered through the lead-discovery workflow. Each record may include a business name, location, industry, public contact details, website fields, an internal score, and the reasoning used for that score. Some records have not passed through every enrichment or review step.
The prototype file is a build registry. It records 56 prototype rows with generation status, links, variants, scores where available, and lead identifiers where the workflow has one. Forty-four rows are marked completed, ten pending review, one pending, and one failed. Across the completed rows, 28 distinct non-empty lead IDs are present. Fourteen of the 56 total rows have no lead ID, and several businesses have multiple variants.
The table below is the boundary we use for this report.
| Record | What it supports | What it does not support |
|---|---|---|
| 2,140 lead rows | The size and composition of our prospecting pipeline | A representative estimate for every Canadian small business |
| 987 “No website” scoring flags | The share of our lead rows explicitly given that flag | Mobile, HTTPS, contact-form, or conversion failures for those same 987 businesses |
| 56 prototype rows | The amount and status of work in the prototype registry | 56 distinct audited businesses |
| 44 completed prototype rows | Completed production records | A claim that every original site lacked the same features |
| 28 distinct lead IDs among completed rows | A conservative count of identifiable completed lead relationships | A complete deduplicated sample when some rows have no lead ID |
| Internal lead scores | How our workflow prioritizes follow-up | An objective website-quality grade or a public market benchmark |
We calculated these figures directly from data/leads.json and data/prototypes.json on August 2, 2026. A future update should state a new date and rerun the counts instead of copying these numbers forward.
The measured finding: many pipeline records have no website
The score reasoning explicitly contains “No website” for 987 of the 2,140 lead rows. Dividing 987 by 2,140 gives 46.1%. That is the most common directly countable website problem in the current file.
The wording is intentionally narrow. We are not combining parked domains, broken pages, social-only profiles, or slow sites into that number. The flag says “No website,” so that is what we report. We also do not treat a blank website_status or a missing URL as proof by itself, because those values can reflect an incomplete discovery step rather than a verified absence.
The pipeline is designed to find potential website work, so the percentage should be expected to run higher than a neutral national sample. It is most useful as an operational finding: among the businesses Seaway Sites currently discovers and scores, a missing website is common enough that first-site guidance matters as much as redesign guidance.
For an owner without a website, the next decision is not whether to buy every feature. It is what the smallest credible site needs to do: explain the offer, establish the service area, show trustworthy contact details, and create a working path to a call, booking, or quote. The website cost guide covers the available build paths and helps with the who-should-build-it decision.
What the no-website leads have in common
Looking only at the 987 leads flagged with “No website,” three patterns are strong enough to report, and they are the patterns the rest of this report refers back to.
First, the no-website leads are concentrated in a small number of local-service industries. The largest twelve industry buckets account for the majority of the 987 records: auto repair (146), hair salon (141), accounting firm (71), electrician (63), plumber (61), lawyer (61), bakery (57), restaurant (48), dentist (48), barber (44), cleaning service (39), and contractor (32). Three of those (cleaning service, plumber, and restaurant) are over-represented in the no-website set relative to their share of all 2,140 leads; auto repair and hair salon dominate absolute count and are close to their overall share. None of these buckets, on their own, prove a national claim. They describe the shape of our prospecting pipeline in Eastern Ontario and a small Quebec extension.
Second, the no-website flag almost never appears alone. Of the 987 leads, 985 (99.8%) also lack a public email, 984 (99.7%) are scored as “Unclear services,” and 740 (75.0%) carry all three additional flags. The combination is the rule, not the exception. For an owner reading this, the implication is that the businesses in this set are not just missing a website; they are missing the public surface area that a small site is meant to provide.
Third, the no-website leads span a wide range of internal lead scores. The 987 leads have a mean score of 51.3, a median of 53, and a range of 45 to 108. A high lead score in this pipeline reflects fit with our offer, not website presence, so a top-scoring no-website lead is still a no-website lead. We use the score to rank follow-up, not to characterize the business.
What the prototype registry teaches us—and what it cannot
Prototype work is useful because it forces an idea to become a page. The build has to choose a headline, organize services, place contact information, select proof, and make the primary action clear on a phone. Those decisions surface gaps in the public information available for a business.
The registry, however, is not a structured audit table. It does not have one boolean column for “mobile failure,” another for “expired certificate,” another for “missing phone number,” and another for “unclear call to action.” Without those fields, we cannot truthfully publish percentages for those problems. A sentence such as “most of 56 sites failed mobile testing” would require records that do not exist today.
What we can say is that the registry contains repeated variants, missing lead IDs, and different workflow states. That makes row count a production measure, not a business count. Before we use prototypes for a later prevalence report, we need to deduplicate them by a stable business identifier and record the same audit checks for every included business.
The qualitative problems we check next
Once a business has a website, we repeatedly inspect four areas. These are editorial observations from our workflow, not quantified findings from the current JSON files.
Can a visitor understand the offer?
The homepage should name the service, the customer, and the service area in plain language. “Welcome to our website” and “quality solutions” do not help someone decide whether the business solves their problem. A visitor should not need to open several pages to learn what the company actually does.
The practical check is simple: show the homepage to someone unfamiliar with the business for ten seconds, then ask what the business sells and where it operates. If the answer is wrong or vague, fix the opening before buying more traffic.
Is there a clear next step?
A button labelled “Learn more” describes navigation, not an outcome. A useful call to action says what happens next: call for emergency service, request a quote, book an appointment, or check availability. The phone number or form must then work.
Test the full path on a phone. Tap the number, submit the form, confirm where the message arrives, and read the success or error state. This direct test proves more than assuming a form works because it is visible.
Can the visitor verify the business?
Useful trust information includes the official business name, current contact details, real people or project photos, relevant credentials, service boundaries, and evidence that can be checked. The right proof depends on the industry. A regulated trade may need licensing information; a salon may need stylist work; a restaurant needs current hours and menu details.
Do not invent review counts, years in business, awards, or service areas to make the page feel complete. Missing proof is a content task for the owner, not permission for the website writer to manufacture it.
Does the site work where customers use it?
Open the site on a phone over a normal connection. Check that text fits, buttons are reachable, images do not obscure content, and the main action works without zooming. Confirm HTTPS, then use Google Search Console and the site's analytics to find issues that a visual pass cannot show.
Google's page-experience guidance and mobile-first indexing documentation explain the search-side expectations. They do not promise rankings, and neither do we.
Five checks an owner can run this week
- Search the business name and confirm the correct website appears.
- Open the homepage on a phone and state the offer and service area after ten seconds.
- Tap the primary phone number or submit the main form, then verify the message reaches the right person.
- Compare the business name, address or service area, phone number, and hours with the Google Business Profile.
- Confirm the site uses HTTPS and that its most important pages load without broken images, missing text, or blocked actions.
Record the result as pass, fail, or not applicable. A dated checklist makes the next review comparable. It also prevents a redesign conversation from swallowing a smaller problem that could be fixed directly.
The local SEO checklist (in progress) goes deeper on business information and local discovery. The questions to ask a web designer guide is the next step before hiring someone to address the gaps.
When a refresh is enough
A refresh is usually enough when the platform still works, the business information is recoverable, and the main problems are copy, images, contact paths, or outdated details. Fix the measured failure, then watch the relevant key event.
A rebuild becomes reasonable when the platform is unsupported, the site cannot be updated safely, the information architecture no longer matches the business, or the cost of repairing the existing implementation approaches the cost of replacing it. That decision should follow inspection, not a sales script.
There is no universal conversion-rate target that proves the work succeeded. Track the action the page is supposed to produce—qualified calls, forms, bookings, or purchases—and compare it with the site's own baseline while accounting for seasonality and campaigns.
What the next research version needs
The next version can make broader claims only after the data model improves. Each included business needs a stable identifier, one selected audit date, and explicit fields for website presence, mobile usability, HTTPS, contact-path success, local information, and primary call to action. Variants must be linked to one business rather than counted as new audits.
We should also publish the inclusion rule: which locations and industries were searched, which records were excluded, and how missing data was treated. That will not turn a prospecting sample into a national survey, but it will make comparisons within the pipeline reproducible.
Until then, the honest conclusion is limited and useful: 46.1% of our 2,140 current lead records were explicitly flagged as having no website, while the remaining common problems are checks we perform but have not yet counted consistently enough to report as prevalence statistics.