How to do a website audit: twenty checks you can run on your phone

Short answer first. A website audit is a structured read of whether your site works for the person already on it: what a stranger understands, whether they believe you serve their town, whether they can reach you, whether the site looks looked after, and how long they waited. Every check below runs on your own phone.

Twenty checks, four under each question, so you can stop after any group and still have done something. Write down each fail. The count at the end sorts you into one of three answers, and one is keep what you have.

Website audit vs SEO audit vs accessibility audit

These three get sold interchangeably and answer different questions. Buying the wrong one is how a business ends up with a report it cannot act on.

That order also saves money. A page nobody acts on does not need more visitors yet.

What a website audit should include

Whoever runs it, these six parts are what make it an audit rather than a list of observations.

• Your first screen quoted, not described. A report that says "improve your messaging" has audited nothing.
• A real comparison: your page next to the two or three businesses a customer would find instead of you, on the same phone.
• A speed figure from a source you can check yourself. Google publishes the measurement for any public address, so a vendor number is never the only option.
• Every claim checkable in thirty seconds without trusting whoever wrote it.
• The fixes in priority order, with a rough time against each, so you can do the first three this week.
• A keep-it path that can genuinely be the answer. An audit that always ends in a rebuild is not one.

Set the test up properly first

Your own phone, on mobile data with the office wifi off, in a private window so nothing loads from cache. For the first check you also need somebody who does not know your business. Skip any of that and the site gives you a kinder answer than your customers get.

Question one: can a stranger tell what you sell?

The five second test, and the one most sites fail. All four checks happen in the first screen, before anybody scrolls.

1. The first sentence, read by someone else. Screenshot your home page without scrolling, send it to somebody who does not know your business, and ask what you sell and where. If they pause, the top of the page is decorating rather than working. A welcome line, a tagline or your own name in that slot is the most common failure on this list.
2. The words a customer would type. Read your headline out loud, then say the words you would use at 9pm with water coming through a ceiling. "Emergency drain repair" is one. "Integrated property solutions" is not.
3. One thing to do, not six. Count the actions competing before you scroll: call, book, request a quote, chat bubble, newsletter, social icons. More than two and you have asked a stranger to decide how to decide.
4. The photo at the top. A stock handshake or a generic glass tower tells a visitor nothing and quietly suggests you might not be real. Your own van, your shop, your team, or work you finished last month does the opposite. Ten minutes, and the highest return change here.

If a stranger cannot say what you sell and where you work from one screenshot, nothing further down the page gets read.

Question two: can they tell you serve their town?

Distance feeds local search results and it decides whether somebody calls. A customer settles it in two seconds, from words rather than from a map.

1. Find your town in the page. In your phone browser, use Find in Page and type the town you most want work from. If the only match is the footer, somebody in Brampton reading quickly has no reason to believe you cover Brampton. The word local on its own matches nothing anybody types.
2. The list, not the radius. A map graphic with a circle on it is a picture, and a hurried reader skips it. Name the municipalities you would send someone to tomorrow. Mississauga tests this well: with no single downtown, Port Credit, Streetsville and Meadowvale behave like separate markets.
3. Neighborhoods, where neighborhoods are how people search. In Toronto, Leslieville and the Junction return different businesses for the same query, and Vaughan splits the same way into Woodbridge, Maple, Concord and Thornhill. Name the three you are in every week.
4. Your town in the page title. Read the tab name, or hit share and look at the title that comes up. That string is your headline in search results, and "Home" wastes the most valuable line you own. A kitchen renovator in Markham gains ground on Toronto-titled pages by putting Markham there and meaning it.

Question three: can they reach you without hunting?

A scanner sees part of this. The rest is counting taps, which needs the device your customer is holding.

1. Tap your own phone number. It should start a call. An iPhone will often dial a plain number on its own, so one that works for you can fail on another phone. It has to be a real telephone link, in the first screen rather than only the footer.
2. Count the taps to reach you. From the home page, count every tap to a form or a booking screen. Two or three is normal. Five means the page is arranged for you rather than for a customer, and every tap is a place to give up.
3. Send yourself a message through your own form, from your phone, then check the inbox you expect it in and the spam folder. Forms fail silently: a host changes something, the receiving address stops working, and a quiet month gets blamed on the economy.
4. Say what happens after hours. Tell a customer at 8pm whether anybody will answer. "Calls answered to 8pm, messages replied to the next morning" beats a bare number, because it removes the reason to keep looking. Your site and your Google Business Profile have to agree.

Question four: does the site look looked after?

Nobody decides this consciously. A neglected-looking site reads as a business that may not answer the phone, fair or not.

1. The copyright year in the footer. If it reads 2023, a visitor takes it as a site nobody has touched in two years, and they are not wrong. A hard-coded year is the usual cause, and it should update itself.
2. The last dated thing you published. A blog or projects page whose newest entry is from 2022 is worse than not having one. Either add to it or take it down.
3. Read your services page as an inspector would. Is anything on it something you stopped doing? Is anything you do now missing? A page listing work you no longer take produces the worst kind of call.
4. Tap every link in the menu and the footer. Look for pages that no longer exist, images that do not load, and a social icon pointing at an account nobody has posted to since 2019. Five minutes, and the one that surprises owners most.

Staleness is the cheapest thing here to fix and the most expensive to leave, because it costs you the customers who never call to tell you.

Question five: how long did they wait?

The one check where your own opinion does not count. You are on office wifi, on a device that already holds a copy of your site.

1. Time it yourself, on mobile data. Wifi off, private window, type your address and count out loud until something readable appears. Do it twice: the second load is faster because the phone kept a copy, so the first number is the honest one.
2. Take Google's measurement rather than a vendor's. Google publishes a free speed test for any public address, reported on a simulated phone over a slow connection. What matters is how long your main content takes to show, and the line it draws is about two and a half seconds.
3. Weigh your images. A photo straight off a modern phone is commonly 2 to 5 MB. Resized to the width it is displayed at and saved in a modern format, it usually lands under 200 KB. Five unresized photos ask a customer on two bars to download a short film first.
4. Count what loads before your content. A cookie banner, a chat bubble, a newsletter pop-up and three tracking scripts are all requested before the page settles. Anything arriving before your headline is competing with it.

What this checklist leaves out

Twenty checks on a phone cover the half of a website audit that decides whether a visitor stays, not the half that decides whether they arrive.

• Where you rank. Nothing on a phone tells you where you sit for a search that matters, and a tool quoting you a position from a quick scan is guessing.
• Schema markup: the structured facts inside the page telling a search engine you are a real local business, where you work and what you sell. You cannot see it on the page, which is why it is usually missing. The local SEO guide covers it.
• Whether Google holds your pages at all. Search for site: followed by your domain and count what comes back. Missing pages are a different problem from badly written ones.
• What your competitors are doing better. The hardest check to run on yourself.

How to tell if your website is bad, honestly

Count the checks you failed. The total is not a grade. It sorts you into one of three answers.

1. Zero to three fails: keep it. Fix what you found, most of which is an afternoon, and put the money into being found rather than being rebuilt.
2. Four to eight: fix, in order. The site is sound and specific things are costing you customers. Most of that list needs no developer.
3. Nine or more: check the three hard fails below before concluding anything. A pile of small problems on a sound site is still a fix list, not a rebuild.

The three hard fails are structural: no phone layout at all, so a visitor pinches and zooms to read a sentence; a platform that no longer gets security updates; and a home page where nobody can tell what you sell after the whole first screen. Patching those is paying twice, and the Toronto cost guide prices the alternative across every band.

The order to fix them in

If you found more than you expected, this sequence moves the most for the least money.

1. The first sentence on the home page. Ten minutes, and the biggest single change.
2. The photo at the top, if it is stock.
3. The phone number: tappable, in the first screen, matching your Google Business Profile.
4. The towns you serve, named above the footer.
5. The contact form, tested from a phone and confirmed to arrive.
6. The images, resized. Usually the entire speed problem.
7. The stale things: the copyright year, the old post, the service you dropped.

Everything after that is search work rather than first impression work, and the local SEO guide picks it up from there.

If you would rather someone else ran it

The list above is yours whether or not you ever speak to me. What is hard alone is the comparison: you cannot read your own home page the way a stranger does, because you already know what you meant. That is what the written website audit is for. It is $49 plus tax, reads every page, shows who is ahead of you on Google, sets your site beside up to three competitors on a phone, and comes back inside three business days with the fixes in order. Not available to businesses billed in Quebec. The FAQ sets out what is in it. If the answer is keep what you have, it says that. If it is start clean, the build scope and the price are published rather than quoted on a call.

Or start with the machine. The free website scan runs the five questions in two to five minutes and quotes your own page back to you, which is the quickest way to know whether the afternoon is worth it.