Google Ads Performance Review
A straight look at what the account is doing, what stands in its way, and what it could do next.
Every number in this report comes directly from the Google Ads account (619-166-0111), pulled on 24 and 27 July 2026. Nothing here is modelled or invented, and on the one occasion an estimate is used it says so plainly.
The last 90 days
Section 03 flagged a figure that could not possibly be real: £40.3 million of conversion value recorded in April, in a month when the account spent £2,044. That has now been traced, and it came down to a single day. On 2 April the Google and YouTube app that connects the shop to Google Ads reported seven add to cart events carrying £39.7 million of value between them, which is roughly £5.5 million a basket. The same signature turns up in smaller doses on a handful of other days, always on add to cart or begin checkout.
The important part is what it does not touch. Of the twenty five highest value days in the entire history of the account, every single one is an add to cart or a begin checkout. Not one is a purchase, and no day of purchases has ever exceeded about £5,000. So the fault sits in the shop's engagement tracking, the purchase figures have never been caught up in it, and those six engagement events are the ones demoted on 27 July, so it no longer influences the bidding either. One question does remain open, and it is worth being straight about: whether Google is claiming credit for the right orders. That is a question of attribution rather than faulty numbers, and the only way to settle it is to compare against what the shop actually sold.
On the surface, the last three numbers look genuinely strong. The catch, and the reason this review exists, is that the account's measurement cannot currently be trusted, and the campaign is being steered by the least trustworthy parts of it. That is fixable, and quickly.
The account contains a single live campaign, a Google Performance Max campaign running at £120 per day. Performance Max works like this: you hand Google your product data, images and text, and its automation decides where the ads appear (search results, Shopping, YouTube, Gmail, websites), who sees them, and what to bid. There are no keyword campaigns, no dedicated search campaigns, and no other active campaigns anywhere in the account. A second campaign was removed early on after spending about £60 in total. The live campaign has been serving since September 2025 and bids to a goal of maximising conversion value.
Performance Max steers entirely by whatever the account tells it a conversion is worth. Almost everything in this report follows from that one fact.
There is a real foundation here. None of the problems later in this report are structural to the business. They are settings.
In the last 90 days the campaign put Drin-KIT in front of 3.1 million ad impressions and won 47,418 clicks at an average of 17p each. The searches behind that traffic are the right ones: beer fonts and taps, keg couplers, cellar cooling, cask equipment, hand pulls, line cleaning. The market Drin-KIT serves is actively searching, and this account reaches it for pennies.
The account recorded 390 purchases in 90 days at an average tracked order of about £169, which is £65,709 of tracked purchase revenue against £8,226 of spend. Taken at face value, that is roughly £8 back for every £1 spent. Face value deserves one honest caveat: given the state of the tracking described below, the purchase figures should be checked against real store sales before anyone banks them. But even allowing a wide margin for error, the direction is encouraging.
No disapproved ads. No policy issues. Sitelinks, call assets and a full pool of ad images in place, and an unbroken eleven month history. People are also searching for the brand by name: searches for "drinkit" and "drink kit" earned roughly 300 clicks in 90 days on their own. That is a base worth building on.
Thirteen conversion actions are configured, and six of them are website engagement events: page views, item views, on-site searches, add to cart, begin checkout and add payment info. All six are set as primary, which means they count as "conversions" and feed the bidding algorithm with the same authority as a real purchase. The result: of the 108,102 conversions the account reported over 90 days, only 390 were actual sales. That is 0.36%. Add To Cart alone was credited with £645,820 of conversion value, nearly ten times what real purchases were credited with. The campaign's goal is to maximise conversion value, so Google is dutifully chasing that noise, optimising towards people who look rather than people who buy.
April 2026 alone recorded £40.3 million of conversion value, in a month where the account spent £2,044. Neighbouring months recorded a few hundred thousand at most. That is a measurement fault, not revenue. Until the fault is understood and the tracking rebuilt, no headline value figure from this account can be quoted with a straight face, to a bank, to an accountant, or in a board conversation. For a business that reads its numbers properly, this is the single most important thing to fix.
Both campaigns target the United Kingdom, but with Google's wider setting that also serves people outside the UK who merely show interest in it. Of the £8,226 spent in the last 90 days, £7,180 reached people actually in the UK. Around £1,000, roughly one pound in every eight, bought clicks from Ireland, Spain, Nigeria, Saudi Arabia, Oman, Cyprus, Greece and at least half a dozen more countries. For a UK trade supplier, that is spend with almost no realistic path to an order.
The account contains zero negative keywords, at any level. Google's own search insight shows the largest named search theme by clicks is "drain kit": 318 clicks in 90 days from people shopping for plumbing parts. Home bar hobbyist searches and rival brand names flow through unfiltered as well. Every irrelevant click costs real money, and right now there is no mechanism anywhere in the account to stop any of it.
With everything riding on a single Performance Max campaign there is no dedicated brand campaign, no keyword level control, and limited visibility into where the money actually goes. The change history shows the account is barely touched: two changes in the last 28 days, both to the budget amount, one of them applied automatically by Google itself. The ad content is rated Average by Google's own measure. And on searches where Drin-KIT was eligible to appear, it actually appeared less than a quarter of the time, with most of the missed share lost on ad rank rather than budget.
Every weakness above has a direct, inexpensive counter. In rough order of leverage:
Make Purchase the only primary conversion action and demote the six engagement events to observation only. Same budget, same campaign, but from that day Google optimises for buyers instead of browsers. This is the single highest leverage change available in the account, and it costs nothing.
Switch location targeting from "presence or interest" to "presence only". At the current rate that re-aims roughly £4,000 a year of spend from overseas browsers back onto UK buyers. It is a five minute change.
Add negative keywords for the plumbing traffic ("drain kit" and its cousins) and the obvious irrelevancies, then make a deliberate decision on home bar hobbyist traffic: serve it knowingly as a secondary audience, or exclude it and concentrate on trade. Either choice is defensible. Today the choice is being made by accident.
A small branded search campaign to properly own the searches for Drin-KIT's own name, and a trade search campaign on the proven head terms: keg couplers, beer fonts and towers, cellar cooling, cask dispense. Search campaigns bring keyword level reporting, exact control over where money goes, and a way to claim the missed impressions. Google is currently missing three out of four eligible search appearances, mostly on rank, and properly built search campaigns are how that share gets claimed.
The ad content pool is rated Average, the trade story barely features, callouts and structured snippets do not exist, and some sitelinks are half finished. Sharper trade first messaging, complete assets and a fuller creative pool push ad rank up, and ad rank is exactly where most of the missed impression share is being lost.
Google's own recommendation engine projects more volume at higher budget, and the data confirms a modest budget constraint. But scaling before the measurement is fixed just buys more noise, faster. Fix the tracking, verify for a fortnight, then grow the budget against numbers everyone can trust.
In order. The first three are small, precise changes; the rest build on them.
Set Purchase as the sole primary conversion action, move the six engagement events to secondary, investigate and annotate the April value fault, and reconcile tracked purchases against actual store sales so the numbers earn back their credibility.
Change both location settings to presence only. Verified in minutes, saves for as long as the account runs.
Exclude plumbing and other irrelevant themes from day one, review the search categories monthly, and take the deliberate trade versus home bar decision described above.
Brand campaign first, then a tightly themed trade campaign on the proven head terms, both reporting at keyword level, alongside a cleaned up Performance Max rather than instead of it.
Trade first messaging, complete callouts, structured snippets and sitelinks, and a fuller image and headline pool, aiming Google's ad strength rating at Excellent rather than Average.
Turn off auto applied recommendations so budget decisions stay human. Then a one page monthly readout: real purchases, real revenue, real cost per order, and nothing else. Budget rises only when that page justifies them.
This section exists for whoever kicks the tyres. Three grades of number appear in this report:
Spend, clicks, impressions, settings and change history come straight from Google's own billing grade ledger. Quote these freely. Examples: £8,226 spent in 90 days, 47,418 clicks, zero negative keywords, two changes in 28 days.
Purchase counts and purchase values (390 purchases, £65,709) look plausible and are the closest thing to truth in the account, but they are recorded by the same tracking setup that produced the faults below. Reconcile against store sales before treating as gospel.
The headline "conversions" total (108,102 in 90 days) and every aggregate value figure that includes the engagement events or April 2026 (£40.3 million). Do not quote these anywhere, for any purpose.
The full data extraction behind this report, including every query that failed and why, is on file and available on request.
This section was added after the report first went out, and it will keep growing. It is the running record of what has actually been done to the account, so whenever you open this page you are looking at the current picture rather than the original snapshot.
The largest single leak this report identified was in section 03: "drain kit", a plumbing search, was the biggest named search theme taking clicks off the campaign, and nothing in the account was filtering it out. On 25 July, the day after the report landed, that exclusion was in place, along with four others: optics, Perfect draft, Drip pan liners and Polykegs. Three of those are ones only somebody who knows this trade would think to write down, and a batch of new creative went up the same day as well, fourteen landscape images and five videos. It is worth being precise about what that finding was worth once it had been acted on, because it was the single biggest filtering win available anywhere in the account: the Drain kit exclusion on its own stops 318 plumbing clicks every 90 days, very nearly all of the plumbing traffic the campaign had been absorbing. That is exactly the pattern this is meant to produce. The review finds the leak, it gets closed inside a day, and everything below is built on top of it.
The six engagement events described in section 03, page views, item views, on-site searches, add to cart, begin checkout and add payment info, are now demoted to observation only. They are still recorded and you can still look at them whenever you want, but they no longer tell Google what to chase. Two things now drive the bidding: completed purchases, and phone calls from ads. The phone calls were kept deliberately, because in this trade an order that arrives by telephone is every bit as real as one that arrives through the basket. The exclusion list was extended at the same time with a preventive layer, so searches about drains, bathrooms, heating and household appliances are filtered out before they cost anything. Deliberately left alone: anything touching pushfit, since that is a line you actually sell and a blunt exclusion there would have cost you real customers.
The campaign had been serving people who were merely interested in the United Kingdom as well as people who are actually in it, and that single setting was what sent roughly one pound in every eight to countries with no realistic path to an order. It has now been switched to Presence, so the ads reach people who are here. Checked directly in the account rather than taken on trust: the campaign's location setting was changed at 16:28 on 27 July and now reads Presence rather than Presence or interest. That was the last of the three quick fixes.
Two of those numbers are money and one is direction. The £4,000 a year is not a saving that turns up in the bank, it is spend that has stopped going overseas and now reaches buyers in the UK, which is better value from the same budget rather than a smaller bill. The plumbing clicks are a real saving but a modest one, somewhere around £220 a year at this account's average click price, and that figure is an estimate because Google does not report cost against these search categories. The conversion goal change is the one nobody can put a figure on in advance, and it is almost certainly the biggest of the three: £120 a day was being steered by a signal in which 99.6 per cent of the events counted were not sales. It is now steered by sales and by phone calls.
When the bidding goal changes, Google has to learn again. For a week or two the reported number of conversions will fall sharply and the cost per conversion may look worse. That is the correction doing its job, not something going wrong: the account has simply stopped counting page views as sales. The numbers worth watching from here are purchases and phone calls, not the headline conversion count.
Searches from people building a bar at home run to more than 350 clicks every 90 days, which is more traffic than the plumbing ever was. Whether that is a welcome second audience or a distraction from trade customers is a commercial judgement about your business rather than a technical one, so it has been left exactly as it is instead of being quietly decided for you. Say the word either way and it is a few minutes' work.
All three of the quick fixes in section 05 are now done. The next step is to let the fortnight settle, then put the account's 390 recorded purchases side by side with what the shop actually sold over the same period. If those two line up, the bigger items in section 05, the brand and trade search campaigns and the creative rebuild, become worth a proper conversation.
This review is the same discipline I apply to the accounts I already look after, including the ones you know about. Between us, the whole of the week one work is now done, and section 07 shows exactly what changed and why. The natural next step is to let the fortnight settle, then reconcile the purchases the account records against what the shop actually sold, so you can see the difference rather than take my word for it.
The bigger rebuild in weeks two to four is a conversation for after that, if the clean numbers earn it. And if you would rather keep it in house, this report is yours either way: take it to anyone you like, question every number, and use it as a checklist. It will still have been worth the read.
Lewis
Lewis Bacon · High Click Rate Ltd · info@highclickrate.co.uk