Ugly, slow and broken are three different problems
They get discussed as one thing and they behave nothing alike.
| Problem | What it looks like | What it costs |
|---|---|---|
| Ugly | Dated typography, stock photography, a layout from a decade ago | Some trust, unevenly. Often less than owners assume, particularly in trades |
| Slow | A blank screen for several seconds, layout jumping as things load | The visitors who leave before it renders, and they are disproportionately on phones |
| Unclear | The visitor cannot tell what you sell or where you work | Everyone who was not already sure they wanted you |
| Broken | The form does not send, the number is not tappable, a page errors | Everything, silently, until somebody tests it |
Only the last row is an emergency, and it is also the one that is invisible from a desk. A business can advertise into a form that has not delivered an enquiry in six weeks and see nothing wrong in any report, because a form that fails silently produces the same reporting shape as a market that went quiet.
Slow is the one that has hardened over the last few years. Google Search Central documents the Core Web Vitals — how quickly the main content paints, how soon the page responds, and how much it shifts while loading — and a site that fails those on a phone is losing paid visitors before they have seen anything you wrote.
The ninety-second test, done on your own phone
Not a desktop. Not office wi-fi. Your phone, on mobile data, held the way you actually hold it.
- Open the page an ad would send someone to. Start the clock.
- Without scrolling, can you tell what the business does and roughly where it works?
- Scroll once. Is there a reason to choose this business rather than the next one — licensing, years, a guarantee, a named area?
- Find the way to make contact. Count the taps.
- Complete it. Actually submit, actually let the phone ring.
If you cannot do all five inside ninety seconds on your own site, a stranger with three other tabs open will not either. Most failures show up at step two or step four, and both are cheap to fix without redesigning anything: a clearer first line, and a contact method that is reachable without hunting.
Do the test again with the consent banner showing, because a banner that covers the call button turns your most valuable action into your hardest one. That single interaction has ended more paid campaigns than any bidding decision.
A poor landing experience is not free
There is a common belief that a bad site simply converts worse and the cost per click stays the same. It does not work that way.
Google Ads Help sets out that Ad Rank is determined by several components, and landing page experience is one of them. Practically, that means the page influences whether your ad shows and what each click costs, not just what happens after the click. A poor page is charged for twice: once in the auction and again in the share of visitors who leave.
A weak page does not only convert less. It also buys the same click at a worse price.
There is a second cost that is easier to miss. The page is the only part of this system that improves every channel at once. A better page lifts paid search, organic, the map listing, the link you paste into a text message and the referral somebody was sent by a neighbour. Money spent on the account improves the account. Money spent on the page improves everything downstream of every channel, which is why it is usually the better first dollar when both are candidates.
Accuracy belongs in this section too. The ACCC's guidance on misleading claims under the Australian Consumer Law applies to your website exactly as it applies to your advertising — availability, pricing, licensing and qualifications all have to be true on the page you are paying to send people to.
A weak page does not only convert less. It also buys the same click at a worse price.
When running anyway is the right call
There are three situations where advertising into an imperfect site is the correct decision rather than a compromise.
When the work is urgent. Someone with a burst pipe or a locked door is not evaluating your typography. Urgency compresses the decision to two questions — can you come, and can you come now — and a plain page that answers both outperforms a beautiful one that buries the phone number.
When you need to know what the market is worth before you invest in it. A rebuild is a large commitment made on assumptions. A modest, well-instrumented run of advertising into an ordinary page tells you what people actually search, what an enquiry costs and which service is in demand. That is a cheaper way to buy the brief for the rebuild than a workshop.
When the site is genuinely fine and the owner is simply sick of looking at it. This is more common than any of the above. Familiarity breeds contempt for a page that a first-time visitor finds perfectly clear.
What ties the three together is that in each case you know what you are trading. Running ads into a site you have not tested is not a decision — it is a hope, and it has no ceiling on what it can cost.
The smallest page worth sending paid traffic to
If you are going to run anyway, there is a minimum. Everything below is a floor rather than a target, and none of it requires a redesign.
- What you do and where you do it, visible without scrolling.
- A tappable phone number that is the same number everywhere else you exist.
- A form short enough to complete on a phone with one hand, and one that has been tested this month.
- One concrete reason to believe you: a licence number, a warranty, the years, the area you actually service, named.
- Nothing that covers the contact method — no interstitial, no chat widget in the thumb zone, no banner over the button.
- A page that finishes loading on mobile data before a person gives up.
That is six items and none of them is a rebuild. A single page meeting all six will usually outperform an entire handsome site that meets four, which is worth knowing before anyone quotes you for a redesign you may not yet need. When the page really is the constraint, it is a website problem wearing an advertising costume, and no amount of account work will fix it. When it is not — when the visits arrive, the enquiries arrive, and the enquiries are simply wrong — the constraint has moved to what happens after the form is sent.
Ugly is a discount. Broken is the full price for nothing, and it does not announce itself.
Repair or rebuild, and how to tell which you need
Repair when the structure is sound and the problems are local: a slow image, a form that fails, a missing phone number, a page that does not say where you work. These are hours of work and they are usually the whole gap.
Rebuild when the problems are structural — when there is no page for the service you want to sell, when the site cannot be edited without a developer, when it was built for a business you no longer are, or when it cannot be made to load acceptably on a phone without starting again.
The way to tell them apart is to write down what you would change and then ask how many of those changes require touching the underlying build. If the answer is none, you have a repair job and you should stop calling it a rebuild.
Either way, set the measurement up first. You cannot prove a page change worked if you were not counting before it, and the argument about whether the new site is better is unwinnable without a number from the old one. That is also the only way to separate the three conversion rates in a service business — visits to enquiries, enquiries to quotes, quotes to booked work — and to see which of the three is actually the weak one. Run the paid search layer and the measurement as one decision, not two.
Common questions
Will a bad website get my ads disapproved?
Not usually for being unattractive. It can happen when a page is unreachable, when it does not work on a phone, or when what it claims does not match what the ad promised. Google Ads Help also lists landing page experience as a component of Ad Rank, so a weak page raises your costs long before it triggers any policy issue.
Should I build a landing page or fix the whole site?
A single page, done properly, is the faster way to learn and the cheaper way to test. It is not a substitute for a site that cannot be edited or does not work on a phone. Use the page to find out what the market wants, then let that decide what the rebuild has to contain.
How fast does the page need to be?
Fast enough that a person on mobile data does not leave first. Google Search Central publishes the Core Web Vitals as the measures it uses for loading, responsiveness and visual stability; testing your own page against them on a phone is more useful than any target number quoted in the abstract.
My site converts fine but looks dated. Do I need to redo it?
Not for the advertising. If enquiries arrive and the ninety-second test passes, the design is not what is limiting you. Redesign when the site cannot do something you need it to do, not because it is out of fashion.