The teardown
Send me the domain. I will tell you what is broken.
A teardown is a written audit of one storefront. You get every finding with the evidence that proves it and, where it can be measured, what it is costing you. It is the only thing I will sell you before I have looked at your site.
What you get
Findings in writing, one per entry.
Each finding has the same four parts: what is wrong, how I established it, what it is costing where that can be measured, and what fixing it involves. You can hand the document to anyone. You can act on it without me.
Every claim carries its evidence. If a paid option is invisible on a class of product, you get the rule that hides it and the query showing no order has ever contained it. If a feed advertises the wrong price, you get the emitted link next to the price the page charges. You should never have to take my word for a finding, and you will not be asked to.
If there is nothing in there worth paying for, I will tell you that instead.
What I look at
The whole stack, because that is where these faults hide.
| Storefront | Theme and template logic, product data and metafields, the conditions that decide whether an option renders at all, layout stability, page weight and speed |
| Search | Crawl and index control read against each other rather than separately, structured data, canonical and duplicate-content faults, what the site tells an engine it is |
| Feeds | Every row validated against the specification with a strict parser and against the price the page charges. Link targets, price tiers and taxonomy, plus the price the channel shows a shopper |
| Analytics | Whether the instrument works before anything is concluded from its readings. Duplicate events, missing parameters, attribution that cannot separate paid from organic |
| Money | Whether the revenue the business believes it made matches what its own records say, reconciled against the platform's statements rather than against a dashboard |
| Fulfillment | What the customer receives after paying, checked against what they bought |
What usually turns up
The faults nobody has reported.
You already know about the slow page and the dated template, because they are visible and somebody has raised them. A teardown goes after the other kind: the fault that renders a correct looking page, throws no error, and leaves nothing on the screen to notice.
These four came from outside the store's own team: a vendor's integration, a sales channel app, a source document. Faults I have found in my own work are listed separately, on the about page.
| What was wrong | What it had been doing |
|---|---|
| A vendor's product feed linking to variants | For the life of the channel, about four rows in five advertised a price above the entry price |
| The same vendor's links carrying no tracking | For the life of the channel, its shopping traffic could not be told apart in analytics |
| A sales channel app sending every file format to Google Shopping | Most of the catalog was listed at its most expensive format |
| A data sheet whose text layer drops checkbox state | Four features read as present that were unchecked on the sheet. Caught before publication |
What it is not
Said plainly, so nobody is surprised.
- There is no upsell section, and no proposal stapled to the back. The document is complete whether or not you ever hire me.
- Automated tools are used where they are the right instrument, and named where they were. A crawler cannot tell you that a condition never matched, which is most of what this is for.
- Taste is not a finding. I will tell you if a layout is costing you money and prove it with a measurement, and otherwise I will leave your design alone.
- Nothing on your site changes. I ask for read access, and where I cannot get it I write down what I could not see rather than guessing at it.
How it runs
Four steps, and you can stop after any of them.
- You send the domain and a sentence on what you think is wrong. The sentence helps even when it turns out to be wrong, because it tells me what the business has been watching.
- I look before I quote. The price depends on the size of the catalog and how much of the stack is in play, so I do not name one until I have seen the site.
- I work through it and send the findings in writing. You get the document whether it holds five findings or one.
- You decide what happens next. The document is yours either way, and you can hand it to anyone.
How the money works
No price list, because there is no standard job.
A nine-product catalog on a clean theme and a six-hundred-product catalog with three vendors writing to it are not the same week of work, and quoting them the same way would mean overcharging one of you. What is standard is the shape of the arrangement, so here it is in full.
The teardown is quoted per site
I look before I quote, and what comes back is one figure with no hourly meter behind it. If the work continues, that figure comes off your first month in full.
The rebuild runs on a retainer
Roughly six months, depending on the size of the catalog and how much of the stack is in play. A project fee ends at the launch, and the launch is the easy part. A retainer paces the work to what genuinely needs doing and means I am still here when the first thing breaks.
It runs on a term for a second reason too. This work compounds and almost none of it shows inside thirty days: Google's own guidance on hiring for search puts the window at four months to a year before improvements are implemented and start to pay.
Then it drops to a reduced retainer plus a share of the growth
Once the rebuild is done the fee comes down, and a share of growth above the baseline the business was already doing takes its place. The revenue you already had is never taxed. If I do not move the number, I am paid the reduced fee and nothing else, which means from that point on the upside I am working toward is the one you are.
The second phase is why this works
Most of what I have found on a running store broke after it was working. A theme update dropped the line that loads the store's own JavaScript, and three shipped features went dead with nothing on the page to show it. The other two were my own doing. A price restore I built carried one column too many and reverted twelve markup fixes without a word. A feed I consolidated lost about three quarters of its Pinterest traffic over the following weeks, as the old listings that carried the account's best pins aged out. I found all three because I was still there.
Those are operating problems, and they surface weeks or months after a build is paid for. The second phase exists so that someone is watching when they do.
What you are not buying is hours
There is no timesheet and no charge for a call or for writing something down. Billing by the hour would pay me to take longer.