'"My nephew built it": when a family-built site is fine and when it is not'

A family-built website is fine when the phone number dials, the form arrives, the padlock shows, Google's lab test does not flinch, the link preview works, and you hold the logins, and it is not fine when any one of those is false, regardless of who built it or what they charged. The nephew is not the problem. The nephew is not the solution either. The six checks are, and they take less time than the family dinner where the site came up.

To be clear about the nephew: plenty of them are good. Some of the sharpest builders we know started as somebody's nephew. The checks below do not ask who built the site. They ask whether it works.

The six checks, from your phone

Do these on your own phone, on cellular data, standing in the parking lot if that is where you are.

1. Press the phone number. If the dialer opens with your number in it, pass. If the page scrolls, or nothing happens, or a contact page opens, fail. A phone number is a link only when it is written as one; MDN's reference describes the behavior plainly: "Cellular devices autodial the number" when the link is a tel: link (Source: MDN). Plain text is plain text.

2. Send yourself a message through the form. Fill it in, send it, then open the inbox it is supposed to reach and wait two minutes. Check spam. If it is there, pass. If the thank-you page showed and nothing arrived, fail. Many forms send through a relay service that must be switched on once; one such service says the first use "will trigger an email requesting confirmation" (Source: FormSubmit). If the confirmation went to the nephew and he was in a chemistry final, the form has been a wall ever since.

3. Look at the address bar. Type your address with http:// in front. If it jumps to https:// and shows a padlock, pass. If it shows "Not secure," fail. Chrome's own explanation of that label: "The site doesn't use a private connection. Someone may be able to view and change the information you send and get through this site" (Source: Google Chrome Help).

4. Run Google's lab test. Go to pagespeed.web.dev, enter your home page, leave Mobile selected. Google describes what the test does: "Lighthouse simulates the page load conditions of a mid-tier device (Moto G4) device on a mobile network for mobile" (Source: Google, About PageSpeed Insights). It is a lab test on a simulated slow phone, not your phone and not your customers' phones. Read the Largest Contentful Paint number. Google's line, in its words: "A good LCP value is 2.5 seconds or less" (Source: web.dev, Largest Contentful Paint). Write the number down; it is the one you will quote to whoever fixes it.

5. Text yourself the link. Send your home page address to yourself in a text message and look at the preview. Your name, a line of text and a picture: pass. A gray box with the bare address: fail. Nobody is harmed by a gray box, but nobody is helped by one either.

6. Find the logins. Who holds the domain registration, in whose name, and when does it renew? Who holds the hosting login? Who can edit the site? If every answer is "the nephew," and the nephew is now in Denver, the site is fine right up to the day it is not. The cheapest fix on this list is a shared document with three logins in it, written today.

When it is fine

Six passes. The site loads, the number dials, the form arrives, the padlock shows, the preview is pretty, and the renewal email comes to you. Leave it alone. Thank the nephew. Send him a gift card. Do not let anyone talk you into a rebuild because the design is three years old; a working site with an older look is a working site.

When it is not

One fail on checks one, two or three. Those are the ones where a customer tries to reach you and cannot. A site that does not dial, does not deliver the form, or warns the visitor off is not a website. It is a brochure with a typo in the phone number.

A fail on check four is slower damage. We do not have a number for what visitors do while they wait; nobody measured it on sites like yours, and we will not pretend otherwise. We can say that Google's lab test is the yardstick Google publishes, and a site over its line is over the line.

A fail on check six is the quiet one. Nothing is wrong today. The domain renews in March, the card on file expired in January, and the first sign is a customer saying your site is "some kind of parking page now."

The myth, both directions

The myth runs both ways. "It was free, so it must be fine" skips the checks. "It was free, so it must be bad" also skips the checks. An expensive site can fail check two as easily as a free one. Price tells you what somebody paid. The six checks tell you what the site does.

What to do about it

Run the six. Write down the results with the date. If everything passed, you are done until next quarter. If something failed, you now have a precise list to hand to whoever fixes it, nephew or otherwise, instead of "the website seems off."

If you would rather we ran them, the audit is free and covers the number, the padlock, the lab test and the preview image, and the Website Fix is $700: images, dead links, the tap-to-call number and mobile layout, HTTPS forced, and a full re-run of our gate with a before-and-after report. Works on most platforms; if yours blocks an item we tell you before you pay. Details at /services#fix.