We run a structured audit on every small-business website we seriously consider working on. Same rubric, same checks, scored the same way. Thirty-seven contractor website audits are currently backed by reproducible records in the project across roofing, plumbing, electrical, HVAC, landscaping, pool service, pest control, garage doors, and remodeling.

Auditing that many sites consistently produces something more useful than opinion. Below is what the committed corpus actually shows, ranked by how often each measured defect appeared. The percentages are calculated from those 37 records, not estimated.

The findings, by frequency

Across 37 audited sites:

  • 76% — logo usability problems. The logo fails at small sizes, on dark backgrounds, or as a favicon.
  • 49% — dated logo design. The visual treatment places the brand firmly in a previous era.
  • 41% — stale copyright year. The footer advertises a year that has passed.
  • 38% — dated or DIY builder signature. The site was assembled on a platform or pattern that has visibly aged.
  • 35% — no click-to-call. The phone number is plain text, not a tappable link.
  • 32% — template leftovers. Placeholder text or stock content the original builder never replaced.
  • 19% — homepage under 200 visible words.
  • 14% — no usable favicon.
  • 8% — dead hash-only navigation links.
  • 8% — table-based layout. A technique abandoned long ago for page structure.
  • 5% — pinch zoom disabled. An accessibility failure.
  • 5% — no viewport meta tag. Mobile browsers never receive responsive viewport instructions.
  • 3% — still served over plain HTTP. No certificate, and browsers say so.
  • 3% — deprecated plugin technology.
  • 3% — broken image requests.
  • 3% — no clear call to action.
Horizontal bar chart showing the five most common findings in SiteMilk’s 37-site audit corpus
The five most common findings in the current 37-site reproducible audit corpus.

The first surprise: most sites were repairable, not catastrophes

We expected a landscape of disasters. That is not what the scores show.

Rated by total need score, the 37 sites split into 2 HOT, 12 WARM, 11 COOL, and 12 LOW. The median score was 23 out of a possible 100. Twenty-three sites landed in the two lower-need bands, compared with fourteen in HOT or WARM.

If someone audits your website and tells you everything is broken, ask to see the scoring. Ask what they measured, and what the result was for the sites they did not sell a rebuild to.

The second surprise: the logo was the most common problem

We did not expect this. The single most frequent finding across the 37 reproducible audits was not a mobile failure or a speed problem. It was logo usability: 28 sites, or 76%, had a logo that broke down at small sizes, disappeared on some backgrounds, or lacked a clean version for constrained contexts.

It is easy to dismiss this as cosmetic. It is not, for a specific reason: the logo is the one asset that appears in contexts the business does not fully control. Google Business Profile, the map result, the browser tab, a shared link preview, an invoice, or a truck door can all expose weaknesses that are hidden in a wide desktop header.

Combined with 18 sites, or 49%, showing dated logo design, brand assets were the largest category of finding in the current corpus. That reframed how we sequence work.

The most consequential defect: no click-to-call

Logo problems were most common. Missing click-to-call, found on 13 sites or 35%, is the functional defect we would usually fix first.

The customer has to memorize ten digits, leave the site, open the dialer, and retype them. Some do. Others return to the search results, where a competitor's number dials on the first tap.

The fix is a tel: link and takes a developer minutes. We have not found many defects with a worse ratio of cost-to-fix versus opportunity lost. Check this one on your own site before considering anything larger.

Homepages that say almost nothing

We flag any homepage with under 200 visible words of body content. Seven of the 37 fell below that line, or 19%.

One plumbing company's homepage carried about 180 visible words. That business offers residential and commercial plumbing, water heater service, leak detection, drain repair, video sewer inspection, and general contracting. Almost none of that was on the page. A visitor could not gauge the depth of the operation, and neither could a search engine.

The cause is rarely laziness. It is usually a design-led build where the template favored large images and short headline text, and nobody pushed back on behalf of the content. The site looks clean and communicates little.

Three of 37 sites, or 8%, had dead hash-only links, so this was not a widespread problem. It is worth including because of how severe it is where it occurs.

On one site, fourteen navigation links did nothing: placeholder anchors from a template that were never connected to real pages. Menu items labelled "Financing," "Service Areas," and "Warranty" simply jumped the page to the top.

The visitor's interpretation is not "this link is broken." It is "this business is careless." For a trade where the customer may let a stranger into their home and hand over several thousand dollars, perceived carelessness is expensive.

What we did not measure

Being straight about the limits of the dataset:

Our rubric does not currently measure page load speed or whether contact forms actually deliver. We think both matter. A silently broken form is especially dangerous because the visitor sees a confirmation while the business may receive nothing. But we have not measured either condition systematically across this corpus, so we are not quoting a number.

Also worth stating: this is a corpus of 37 reproducible audit records, selected because the businesses were prospects rather than sampled randomly from a market. Treat these as findings from a specific operating dataset, not as national statistics.

We would rather publish a smaller honest number than a larger impressive one.

What to do with this

Sort anything you find on your own site into two piles.

Defects — a phone number that will not dial, links that go nowhere, no certificate, or template placeholder text still on the page — should be repaired promptly regardless of larger plans. They are usually cheap and can interfere with work today.

Dated elements — thin content, a logo that only works in one context, or a layout that fights the phone — are redesign questions, worth solving properly rather than patching indefinitely.

The useful distinction is that the first pile can often be cleared for very little. Based on this dataset, plenty of owners who assume they need to start over may actually need a focused set of repairs.

See the direction before the rebuild

SiteMilk offers a free before-and-after homepage mockup so you can judge the modernization direction before committing anything. Start in the homepage contact section or email milkman@sitemilk.com.

Same brand. Modern build. Fresh pour.

Frequently asked questions

What is the most common problem on contractor websites?

In the reproducible corpus of 37 audits, logo usability was the most common finding at 28 sites, or 76%. The most consequential functional defect was a phone number presented as plain text rather than a tappable link, found on 13 sites, or 35%.

How many words should a contractor homepage have?

We use 200 visible body words as a minimum screening threshold, not a target. Below it, a page rarely establishes what the business does, where it works, and why it should be trusted. Seven of the 37 audited sites fell below that line. The goal is to describe the services clearly, not to pad the page.

Are most small-business websites badly broken?

No. Of the 37 reproducible audits, the split was 2 HOT, 12 WARM, 11 COOL, and 12 LOW, with a median score of 23 out of 100. Twenty-three sites sat in the lower-need COOL or LOW bands. A credible audit should distinguish focused repairs from a genuine rebuild case.