Website performance audit: repair, or rebuild?

TL;DR A website performance audit answers the question you cannot put off: can the site you run today be repaired, or does it need rebuilding, and what does each path cost? One week, $2,500, written down, and yours whether or not we do the work.

The report is not what you are buying. You are buying the moment the argument inside your own building ends, because the answer came with a price on it.

What this includes

An audit week is spent on the site you actually run, not on the one a new supplier would like to sell you. If you have asked for a technical audit somewhere else and been handed a deck, this is the opposite: we open your real pages, on a phone, on the connection your customers have.

• The pages that carry most of your traffic, timed on a phone rather than on a desk machine
• Where visitors leave and never come back: the checkout steps, the search box, the page that makes somebody wait before showing a price
• A read of the code behind those pages, so you know whether the delay is one repair or a habit of the whole site
• Whether search engines can read each page, and which pages describe something other than what they sell
• One line per fix, with a price and the effect we expect, listed in the order we would work
• Whether a visitor who cannot use a mouse can still buy from you, measured against the accessibility rules that apply to public sites
• What the delays cost you in orders per month, built from your own figures rather than from an industry average

Everything is written so that it can be acted on by whoever does the work: your team, your developer, or ours. The audit is one engagement inside our consulting practice, and when the week shows that the real question is which platform rather than which repair, that is a different decision, measured the same way. Most of what we find repeats patterns we keep meeting in the Miami storefront work.

What a website performance audit changes in your week

A slow page on a phone loses the sale before the visitor ever sees a price. The same page on a laptop looks perfectly fine, and that is why the problem survives for years: nobody inside the business ever experiences what the customer experiences.

An audit replaces opinion with a number per delay: this page, this step, these hours, and what it is worth in orders a month. We use the speed measures Google publishes as the shared language, because they are the ones your developer and your marketing supplier already argue in, and because they are measured on real phones rather than on a good connection.

The distance between repairing and rebuilding is money, not pride. A site whose delays sit in a handful of page templates is a repair list, and a repair list is almost always cheaper than a rebuild. A rebuild wins when the weight is everywhere: when the shop, the content and the checkout were assembled by three different people and no longer agree with each other. One week is long enough to tell those two situations apart, and short enough that you have not committed a budget before you know which one you are in. Where the two answers are close, both are priced in full and the trade is written out, so the decision is yours and not a recommendation you have to accept on trust.

What you take away is a decision in two columns a partner or a board can look at: what the repairs cost per milestone, what a rebuild would cost, and what the delays charge you in the meantime. If the answer is repairs, the list is the scope, and the numbers on it can be checked by anybody you trust.

The week also has a licence to be uninteresting. If your site is fast enough and the losses sit somewhere else, in how the phone gets answered or in what your price looks like beside a competitor's, we say so and you keep the measurement. A supplier who only ever finds work for their own team is not measuring anything.

Because repairs change what your customers see, work on the visible storefront is scheduled outside November through April, when Miami is quiet and a change is less expensive to make. The audit week itself runs whenever you are ready, and it is spent with the people who answer for the numbers, in English or in Spanish, on site if that is where the subject gets decided.

How it works

1. Audit week, one week, $2,500. We open your real pages on a phone, walk a buying session from search to payment, and write down what each delay costs you.
2. Findings, that same week. The ranked list, a price against every line, and the order we would work in, handed over whether or not you hire us afterwards.
3. Repairs, quoted per milestone from $5,000. We do the work, or your team does, from the same list. The audit fee comes off the build.
4. Re-measurement, included. The same pages, the same phone profile, with both numbers side by side so the gain is not a claim you have to take on faith.

The week is deliberately short. An audit that runs two months has become a project of its own, and by the time it lands the season has turned. The list is written to be started on the Monday after it arrives, with or without us.

When the answer turns out to be a rebuild rather than a repair, the audit week is still the cheapest way to find out. The list then becomes the scope of the rebuild itself, and you have bought the plan before the build rather than during it.

What it costs

Prices below are where each phase starts; your scoped quote fixes the exact scope, the price and the timeline.

One week, $2,500, written up and handed over at the end of it. The result is yours whether or not you hire us afterwards. Repairs are quoted per milestone from $5,000, and care after launch is from $1,500 a month. How each phase is priced is published, and it is the same grid you are reading here.

What moves the figure is visible before you commit: how many page templates carry the delays, whether the checkout is involved, and whether the work sits with your team or with us. A shop with one slow template is a shorter week than a site where every page waits on the same heavy element, and a page that is slow on a phone but fine on a desk is a different job again from one that never loads at all.

A rebuild is a different number, and a bigger one. The point of paying for the audit week is that you buy the answer to that question for the price of the audit, instead of discovering it halfway through a build you already started. If the site turns out to be fast enough, and the losses sit elsewhere, we say so and you keep the measurement.

What you are not paying for: a licence, a slide deck, or a monthly report that repeats the same list with a new date on it. Every line you receive carries a price, and you choose which lines get done.

Proof

Food By Box carried 2,500 products when we measured it, and its Google speed rating came out at 95 after the rebuild. Monthly visits went from 20,000 to 500,000 over 24 months, and every address that had brought traffic kept it. On a phone, Tatiana Miath's shop moved from just under five seconds to barely over one, and nothing it ranked for was lost. The numbers behind these projects are published page by page, including the parts that took longer than we planned.

Those are the numbers we can show. What we will not do is fill this section with an industry average about visitors who leave after a few seconds, borrowed from somebody else's study and applied to your shop. When a figure is missing we say it is missing, which is also how the audit itself is written: what was measured, and what could not be.

Michael Agnassia writes the audit and reads the repairs before they ship, after 15 years in e-commerce. Who signs the audit and checks the work is on our about page. The studio is at 78 SW 7th Street in Brickell, Miami, and we work in English and Spanish, on site across South Florida.

Questions

What does a website performance audit give us that the free speed test Google publishes does not?

The free test scores one page out of 100 and stops there. The audit separates your templates, says which item is worth doing first, puts a price against each one, and expresses the result in orders per month rather than in points.

Do we need a full code audit for a shop our size?

You get the part you need. On most shops the delays live in a few templates, so a code audit covers those pages rather than every file on the site. Paying for a reading of code that never ships anything is a waste of the week.

How do you work out what a delay costs us in orders?

From your own figures: orders, average order value, and how many visitors reach the payment step. We measure the page again after the repairs and show you both numbers, so the effect is a comparison rather than a promise.

Should the audit come before a rebuild?

Before, every time. Some stores need a handful of repairs rather than a new site, and the difference between those two answers runs into five figures. The audit is what tells you which one you are looking at.

Will the list be usable by our own developer?

It is written for whoever does the work: the page, the file, the reason, and the effect we expect. Your team can run it without us, and the price per line is the same whether the work is done in house or here.

What happens if the audit finds nothing worth fixing?

Then you have bought a measurement, not a project, and we tell you in the same week. That happens more often than a supplier's marketing suggests, and it is a result we are willing to be judged on.

How much of our time does the audit week take from us?

Two short sessions: one at the start to walk the buying journey with the person who knows it, one at the end to hand over the list. Everything else is our work, on your live pages, without asking your team to fill in a questionnaire about their own site.

Next

Start with the audit week: one week, $2,500, and a written list you keep, with a price against every line. If the site itself turns out to be the ceiling, the next step is the rebuild rather than the repairs. Choose a time that suits you and bring the site exactly as it is; nothing needs tidying before we look at it.