Google Ads not converting? Most of the leak is after the click

Before anyone touches the site, do the arithmetic that decides whether the campaign can work at all. Take gross margin per order — not revenue, but what is left after product cost, payment fees, shipping and returns. Multiply it by the rate at which paid sessions become orders. That is what one click is worth to you. As an illustration: a $60 average order at 35% margin, converting paid sessions at 1.5%, makes a click worth roughly 30 cents. If your average cost per click is $1.20, no landing page rescues that. The maths has three levers — margin, order value, conversion rate — and a storefront rebuild moves one of them, sometimes two. Nothing on your site moves your cost per click. Run these numbers first, because the honest answer is sometimes that this keyword, or this channel, cannot pay at your price point.

• Margin per order after product cost, fees, shipping and returns
• Conversion rate of paid sessions specifically, not of all traffic
• Repeat purchase value, but only if you can actually evidence it
• Average CPC in the auction you are really competing in

Two leaks sit close together here. The first is intent: a broad keyword pulls in people researching, comparing or hunting for something cheaper, and they behave nothing like someone typing a model number. The second is destination. Sending every ad to the homepage is the most common and most expensive habit in ecommerce accounts — the visitor already searched for a specific thing, arrives at a general shop window, and has to repeat the search they just did. The fix is unglamorous: each ad group points at the page that answers its query, and that page repeats the ad's promise in its own first heading. If the ad says free returns for 30 days, the page says it above the fold, in the same words, beside the price. Message match is not persuasion; it is the visitor confirming within a second that they are in the right place.

• One ad group, one destination: the category or product the query names
• Repeat the ad's headline promise in the page's own first heading
• Price, availability and delivery terms visible without scrolling
• Route research queries and buying queries to different pages

Google names landing page experience as one of the three components it reports in Quality Score, alongside expected click-through rate and ad relevance, and its own guidance for that component talks about relevant original content, transparency, and a page that loads quickly and works on mobile. What Google does not publish is a formula converting milliseconds into cents of cost per click, and anyone who quotes you one is inventing it. So treat speed as removing an obvious negative rather than as a lever with a known exchange rate. Separately from advertising, Google does publish concrete Core Web Vitals thresholds — LCP under 2.5 seconds, INP under 200 milliseconds, CLS under 0.1 — and those are a sane target for a paid landing page for the plain reason that people abandon pages that make them wait, whatever the auction thinks of it.

Check the device split in your campaign reports before anyone argues about desktop layout. For most consumer stores the majority of ad clicks are mobile — often a mid-range Android on an ordinary connection, arriving cold from an in-app browser with no cache and no session history. That is the environment the landing page has to survive: a hero image that renders fast, price and add-to-cart reachable with a thumb, filters that do not reload the page, and no cookie banner or promo interstitial covering the buy action. A theme that looks fine in a desktop lab test can still take five seconds to become interactive on the hardware your paid traffic actually uses. Stores where most sessions are mobile often find their worst paid performance was never a targeting problem — it is the first four seconds of a phone session that nobody has watched on a real device.

This is the leak that makes the rest of this page pointless if you skip it. Conversions get double-counted when the confirmation page is reloadable, or when a native Google Ads tag and an imported analytics conversion are both marked primary. They get lost when the order confirmation is never reached, when the tag fires before payment settles, or when a plugin conflict strips parameters from the URL. Then attribution moves the goalposts again: where consent is required, Google's Consent Mode governs whether ad identifiers may be stored at all, and some of what you see in the interface is modelled rather than observed. That does not make the data worthless — it means you should know which numbers are measured and which are estimated before you switch off a campaign that merely looked unprofitable. Reconcile against WooCommerce orders on the same date window before trusting either side.

• One primary conversion action per goal — native tag or imported, not both
• Test a real order end to end, including a refresh of the confirmation page
• Reconcile WooCommerce order counts with the ad platform on identical dates
• Know which conversions are observed and which are modelled under consent mode

In Shopping campaigns you do not write the ad — the feed does. Title, image, price, availability, GTIN and product type decide which queries you appear for and how you look next to the competitor beside you. Titles generated straight from a WooCommerce product name are usually too short and missing the words shoppers actually type: brand, model, size, colour, material, roughly in that order of usefulness. Prices and stock that drift out of sync with the store produce disapprovals and paid clicks on items you cannot ship. Images carrying watermarks or promotional text get rejected outright. And the click still lands on your product page, so that page has to confirm the same price and availability the feed advertised. Feed and product data quality is usually where a Shopping account gains more than any bid adjustment will give it.

We do not manage ad accounts. We do not touch bids, keywords, budgets or negatives, and we work alongside whoever does — usually the agency or freelancer already in the account. What stays in WooCommerce is everything the business runs on: products, variations, stock, pricing, coupons, orders, refunds, tax and shipping rules, payment gateways, plugins and the checkout itself, in the same WordPress admin your team uses today. The product feed keeps its existing source of truth. What changes is the destination: the category, product and campaign landing templates the ads point at, rebuilt as a Next.js storefront in front of the existing store, plus a tracking layer rebuilt so it fires once, correctly, on a real order. Engagements start at $1,999; pricing explains how scope is set before anything is agreed.

If your tracking is unreliable, do not spend a dollar on the storefront. Fix measurement first — you cannot judge a landing page change against a number you do not trust, and a month of clean data costs far less than a rebuild aimed at the wrong page. Beyond that, several situations make this work premature. If the unit economics in the first section do not close, a faster page cannot bridge a gap of that size. If you spend a few hundred dollars a month on ads, the absolute upside is smaller than the invoice, and targeted image work, caching or removing one heavy plugin gets you most of the available speed. If the account itself is unstructured — a single campaign, broad match, no negative keywords — the account is the cheaper fix and should go first. When the destination genuinely is what is failing, we will say so; when it is not, we will say that instead. A short diagnostic conversation usually settles which it is.