A business we spoke with last year was spending a little over $12,000 a month on Google Ads. The traffic was fine. The keywords were fine. The site converted visitors into enquiries at 0.8%, which meant roughly nine hundred and ninety-two people out of every thousand arrived, looked and left without doing anything at all.
Their agency's recommendation was a larger ad budget.
This is the reason website optimisation has become a phrase worth being careful with. Four different vendors will sell it to you and mean four different things, and the one who gets there first tends to define the problem in the shape of whatever they happen to sell. The speed specialist finds a speed problem, the SEO agency finds a ranking problem, and the agency running your ads finds a budget problem. All three can be right about the detail and wrong about what to do first.
The order is what decides whether any of it returns money.
What the phrase actually covers
Website optimisation is an umbrella over four distinct disciplines, and they do not compete so much as depend on each other.
Conversion is the work of turning the people already on your site into enquiries or orders. Performance is how fast the thing loads and responds. Search is whether the right people arrive at all. Content and messaging is whether what they read once they get there makes the case.
Most engagements sold in New Zealand under the optimisation banner are one of those four wearing the name of all four. That is fine as long as everyone knows which one they bought.
The order that works
Here is the sequence we use, and the reasoning behind it matters more than the list.
Know what the site is for, commercially. Not the brand goal. The number. Enquiries per month at a defined quality, or revenue per session, or average order value. Sites drift into underperformance because nobody wrote down what performance meant, which is a problem we have written about at length in the real cost of a bad website. Optimisation without a target is redecoration.
Get measurement you believe. Before changing anything, confirm the analytics are counting what you think they count. Duplicate tags, missing conversion events, form submissions that fire on validation errors, bot traffic inflating sessions. We have opened accounts where the recorded conversion rate was wrong by a factor of three in both directions. Every decision downstream of bad measurement is a guess wearing a chart.
Fix the leak before opening the tap. Conversion first, traffic second. A site converting at 0.8% that reaches 1.6% has doubled its return on every dollar of media spend already committed, retroactively, without a single new visitor. Buying more traffic into a leaking funnel is the most expensive way to discover the funnel leaks.
Then performance. Speed compounds with everything above it and it feeds search, so it comes fourth rather than first, though it is the one people reach for first because it produces a satisfying score.
Then search. Once the site converts and loads properly, ranking it is worth doing, because the traffic it earns now goes somewhere useful.
Run that order backwards, which is the industry default, and you spend the most money at the point where it does the least.
Performance, and the numbers that matter
Google measures three things and publishes the thresholds, so this part carries less opinion than the rest.
Largest Contentful Paint should come in under 2.5 seconds. Interaction to Next Paint, which replaced First Input Delay in 2024 and measures how quickly the page responds when someone taps something, should be under 200 milliseconds. Cumulative Layout Shift, the metric that catches a page rearranging itself while somebody is trying to press a button, should sit under 0.1. Anything past 4 seconds, 500 milliseconds or 0.25 is graded poor.
The detail most reports skip is that Google grades on the 75th percentile of real user data. Three quarters of your actual visits have to be good. A lab test on a fast connection in an office in Grey Lynn tells you very little about someone loading the site on a phone in Kaitaia, and the second person is the one being counted.
For a New Zealand business, the practical causes are boringly consistent. Images uploaded at camera resolution. Scripts from tag managers, chat widgets and tracking pixels loading before the content. Fonts blocking the first paint. On commerce specifically, app bloat is the largest single cause, which we have covered in Shopify speed optimisation and in how too many apps cost your store revenue.
Conversion, where the money actually is
The arithmetic here is unglamorous and it is the strongest argument in the whole discipline.
Take a business doing $200,000 a month online at a 1.2% conversion rate. Lift that to 1.8%, which is a realistic outcome for a site that has never had the work done properly, and you are at $300,000 on identical traffic. No extra media spend, no new audience, the same visitors treated better.
When we rebuilt Clean Collective across Australia and New Zealand, conversion rate rose 103% and sales rose 28%. Those two numbers moving at different rates is what conversion work looks like in practice: the rate improves faster than revenue because the traffic mix shifts as well.
Where the gains come from is rarely exotic. Clarity about what the business does within the first few seconds, which we set out in five seconds to trust. Fewer form fields. Pricing that exists on the page rather than sitting behind an enquiry form, product imagery that answers the question a buyer is actually asking, and checkout or enquiry flows with the friction taken out of them. UX and UI for conversion goes deeper on the mechanics, and why your website isn't converting covers the diagnosis.
Search, and what it can and cannot fix
Search is the part most often bought first and it is the part most dependent on everything else being right.
A site that loads slowly, converts poorly and says nothing distinctive will not be rescued by ranking. It will simply be found by more people who then leave, which is a more expensive version of the same problem. Plenty of what gets diagnosed as an SEO failure is a website failure with a search-shaped symptom, a point worth reading alongside why your site isn't ranking.
Search earns its place once the fundamentals hold. Then the technical work, the site structure, the content that answers real questions and the internal linking all compound rather than amplifying a weak page.
This is also where the website design and SEO conversation tends to go wrong. They are treated as sequential purchases from separate suppliers, when the structural decisions that determine whether a site can rank are made during the build rather than afterwards. Retro-fitting is always more expensive than deciding correctly at the outset.
What to measure once it is running
Four numbers, reviewed monthly, will tell you more than any dashboard with forty.
Conversion rate by channel, because an average hides the one channel that is failing. Revenue or qualified enquiries per session, which is the only figure that survives a traffic mix change. Core Web Vitals at the 75th percentile of field data. And the assisted conversion picture, so the channel doing the introducing does not lose the credit to the channel doing the closing.
Anything beyond that is interesting rather than actionable.
When optimisation is the wrong answer
Sometimes the right recommendation is to stop optimising.
If the site is built on an architecture that fights every change, if the content model cannot express what the business now sells, or if the brand has moved on and the site has not, incremental work becomes expensive maintenance on an asset that will be replaced anyway. The tell is when each improvement takes longer than the last and the wins get smaller.
At that point the money is better spent on a rebuild with the commercial target defined at the start. Optimisation raises the ceiling of what a site can do. It cannot move the ceiling itself.
Common questions
What does website optimisation cost in New Zealand?
A one-off audit with a prioritised action list sits in the low thousands. Ongoing conversion work, done properly with testing and measurement, runs as a monthly engagement in the thousands per month depending on traffic volume, because you need enough sessions to reach significance. Speed remediation is a fixed piece of work rather than a retainer.
How long before optimisation shows results?
Speed improvements are visible immediately in field data within about four weeks, as Google's data set refreshes. Conversion changes need enough traffic to be confident the movement is real, which for most New Zealand businesses means four to eight weeks per meaningful test. Search takes longest, three to six months before structural work shows in rankings.
Do I need SEO or conversion work first?
Conversion, almost always. Improving the rate at which existing visitors convert makes every other channel more valuable retroactively, including the search traffic you have not earned yet.
Is a Google PageSpeed score of 100 the goal?
No. The lab score is a diagnostic rather than a target. Field data at the 75th percentile is what Google actually uses, and a site scoring 70 in the lab with good real-world numbers is in better shape than one scoring 95 that is slow on a phone in Invercargill.
Skyrocket builds Webflow and Shopify websites for established New Zealand and Australian businesses, and the same thinking runs through a website audit as through a build. If you want a straight read on which of the four disciplines your site actually needs, get in touch.
