Measure First, Then Build

Key takeaways:

• Businesses that audited their analytics before committing to a redesign consistently outperformed those that started with design work, often at a fraction of the cost.
• A $6,000 investment in measurement infrastructure revealed that a planned $14,000 rebuild was unnecessary — targeted optimisation of three pages delivered comparable results.
• The single biggest predictor of website ROI is not budget size or design quality. It is whether anyone was actually watching what the site was doing before money was spent.

The clients who spent less got more

That sounds like a headline designed to get clicks. It is not. It is a pattern I have watched repeat across 25 years of building websites, and it became particularly clear this year as I looked back at the projects we completed at Chillybin.

The businesses that got the best returns were not the ones with the largest budgets. They were the ones who came to us with questions about what their existing site was doing before they asked us to replace it.

The ones who came in saying “we need it to look more modern” or “our competitors just redesigned, so we should too” — those projects were harder to call a success, even when the end product looked great. Because looking great and performing well are two different things, and only one of them pays for itself.

What does “measuring first” actually mean?

It means before any design brief is written, before any developer quotes a project, someone pulls up the analytics and asks the uncomfortable questions.

Where is your traffic coming from? Which pages are people actually visiting? Where are they leaving? If you have forms, what percentage of people who see them fill them out? If you sell something, what does your conversion funnel look like and where does it break?

Most businesses either do not have this data, or they have it and nobody is reading it. I have been in discovery calls where a client has had Google Analytics installed for four years and cannot tell me their three most-visited pages. That is not a technology problem. That is a priorities problem.

The fix is straightforward. Before we commit to any significant build or redesign, we run a proper analytics audit. We look at traffic sources, behaviour flow, exit rates on key pages, Core Web Vitals scores, mobile versus desktop split, and any conversion tracking that exists. If conversion tracking does not exist, we set it up first. That process takes time and costs money. It is also the most valuable thing we do on any project.

Why do so many businesses skip the measurement step?

Because measurement is invisible and a new website is not.

A redesigned homepage is something you can show your board. A spreadsheet showing that 68% of your visitors are leaving your pricing page after twelve seconds is something you have to explain. One feels like progress. The other feels like homework.

There is also a version of this where the agency pushes it. Back in the early days of the industry, the pitch was always about aesthetics. “Your site looks outdated.” “Your competitors have video backgrounds.” “You need to be on the latest version of WordPress.” Some of that is legitimate. A lot of it is just selling the thing that is easiest to sell.

I have been on both sides of this. In the early 2000s, I was building sites where the brief was essentially “make it look like this other site but with our logo.” We did not have the data infrastructure to know whether any of it was working. We built, we launched, we moved on. The industry has changed. The tools are extraordinary now. There is no excuse for going in blind.

What happens when you measure before you build?

You find out things you did not expect, and some of those things change the entire brief.

A firm I worked with earlier this year had budgeted for a full website rebuild. Fourteen thousand dollars. They had already decided it was happening. The brief was written. They were ready to sign.

Before we started, we spent six thousand dollars on measurement infrastructure — setting up proper event tracking, running a full analytics audit on twelve months of data, and mapping the actual user journeys people were taking through the site.

What the data showed was that 80% of their converting traffic was coming through three pages. Not the homepage. Not the about page. Three service pages that had been written by a staff member two years earlier and had never been touched since. The rest of the site — the parts they wanted to redesign — was essentially decorative. People visited it, but they did not convert from it.

The recommendation we made was not what they expected. We told them the full rebuild was not justified by the data. Instead, we proposed optimising those three high-performing pages — improving load time, restructuring the content hierarchy, tightening the calls to action, and fixing a form that had a broken field nobody had noticed. Total cost: around four thousand eight hundred dollars.

Three months later, their leads were up. Not because they had a new website. Because they had fixed what was actually broken, based on evidence rather than assumption.

That is not a unique story. It is the same story I have seen play out with different numbers across many different clients. The budget figure changes. The pattern does not.

Does this mean you should never do a full redesign?

No. Sometimes a full redesign is exactly the right call. Old sites on unsupported infrastructure, codebases that cannot be maintained, platforms that are fundamentally incompatible with what the business needs to do — these are real problems that measurement will not solve.

What measurement does is tell you whether you need it. And if you do need it, measurement tells you what the new site actually has to achieve.

The worst redesigns I have seen are the ones where the brief was purely aesthetic. “We want it to feel more premium.” “We want something clean and modern.” These are not briefs. They are vibes. A vibe-driven redesign produces a site that wins design awards and loses business, and everyone is confused about why.

The best redesigns start with a clear understanding of what the current site is failing to do. “Forty-seven percent of mobile users are dropping off the enquiry form on the final step.” “Our best-performing content is buried three clicks from the homepage.” “Page load time on our service pages is 8.4 seconds on a 4G connection.” Those are briefs. Those are things you can design and build a solution to.

How much should you invest in measurement before a build?

Enough to actually know something.

That sounds vague, so here is a more specific frame. If you are considering a $15,000 website project, spending $2,000 to $5,000 on a proper audit first is not an overhead — it is insurance. It either confirms the spend is justified and tells you exactly what to build, or it finds a cheaper path to the same outcome.

The businesses that skip this step are essentially making a significant investment based on intuition and competitive anxiety. Sometimes they get lucky. More often, they build something that looks better but performs about the same, and they are back in the same conversation eighteen months later.

The tools available now make this easier than it has ever been. Google Analytics 4, Search Console, Hotjar, proper event tracking — none of this requires a massive technical investment to get useful data. What it requires is someone who knows what they are looking at, and enough patience to wait for the data to tell you something before you start spending money on a solution.

What about Google algorithm updates and platform changes?

They matter less than most agencies will tell you they do.

This year we had multiple Google algorithm updates and continued refinement of Core Web Vitals thresholds. Every time one of these hits, there is a wave of content telling businesses they need to act immediately. Some of that advice is good. A lot of it is a reason to sell an audit or a rebuild.

The businesses that weathered algorithm changes without drama this year were the ones who had clean, well-structured sites with fast load times and content that actually matched search intent. That is not new. That has been true since I started doing this. The specific thresholds and signals shift, but the underlying logic of what Google is trying to reward has been consistent for a very long time.

Chasing algorithm updates is a distraction. Building a site that genuinely serves the people who visit it is not. These two things often overlap, which is why the fundamentals keep winning regardless of what changes in any given year.

The measurement-first approach serves you here too. If you are watching what your site is doing, you notice when something changes. A drop in organic traffic shows up in your data before it shows up as lost revenue. You have time to investigate and respond. If you are not watching, you find out about the problem from your accountant.

The version of this I have watched fail, repeatedly, over 25 years is the one where money moves before questions are asked. Big budget, clear vision, no data. The new site launches, everyone takes a photo of themselves in front of the screen, and six months later the enquiries have not moved.

The version that works is less exciting to announce. It starts with pulling up a dashboard and asking what is actually happening. It produces a brief that is specific rather than vague. It sometimes results in a smaller project than originally planned. It almost always produces a better return.

Measure first. Then build.