Google Ads case study: 6.45x ROAS for a San Antonio store

TL;DR

• A San Antonio barbecue rub company came to us with a Google Ads account that spent every day and sold very little.

• The account finished at 6.45 conversion value per unit of cost, 787.59 conversions, $4.77 cost per conversion, $24.3K conversion value.

• Five changes did it: fixing conversion tracking so it reported revenue, cleaning the search terms, splitting brand from non-brand, repairing the product feed, and switching to value based bidding only once there was enough data to switch on.

• Nothing here needed a bigger budget. The money moved, it did not grow.

• For context, WordStream's 2025 benchmarks put ecommerce search ads at a 2.81% conversion rate and the all industry average cost per lead at $70.11, up from $66.69 the year before.

• If you want to copy one thing, copy the order. Tracking first, search terms second, bidding last. Doing it backwards is how accounts get stuck.

The client is a barbecue rub and sauce company in San Antonio. Small team, one warehouse, a website that had been losing money on Google for months before anyone sat down and worked out why.

The multiple is the easy part to quote. The work underneath it is the part worth copying.

What are the actual numbers?

Four figures, straight from the account. Conversion value over cost of 6.45. Total conversions of 787.59. Cost per conversion of $4.77. Conversion value of $24.3K. Those are the real results and they are the only numbers from this account in this post.

Two useful things fall out of that set if you do the arithmetic yourself. Multiply 787.59 conversions by $4.77 and you get a shade under $3,760 in spend. Divide $24.3K of conversion value by 787.59 conversions and the average order comes to around $30.85. So this is a low ticket store buying cheap conversions in volume, not a business closing four figure deals twice a month. That matters, because the tactics that suit a $31 average order are not the tactics that suit a $3,100 one.

Conversion value over cost is the figure Google reports where most people say ROAS. A 6.45 means every dollar of spend was matched by $6.45 of tracked revenue. It's a tracked number, not an accounting number: it counts what the tag recorded, before cost of goods, shipping, card fees or returns. A store with a 40% gross margin at 6.45x is genuinely profitable on ads. The same 6.45x on a 12% margin product would be a much tighter conversation.

What was actually wrong with the account?

The complaint was the ordinary one. Spend went out every day, Google's interface reported plenty of conversions, and the bank account disagreed. The owner had already been told twice by different people that the campaigns "just needed more budget."

They didn't. The account had three separate problems stacked on top of each other, and each one made the next harder to see.

Conversion tracking was counting the wrong things. Newsletter signups, contact form submissions and add to carts were all firing as primary conversions alongside purchases. Google's bidding was doing exactly what it was told, which was to buy as many of those events as possible. It got very good at buying newsletter signups.

Second, no revenue was being passed. The purchase tag fired but sent no value, so every $9 sample tin and every $80 gift box looked identical to the algorithm. You can't bid toward value that you never report.

Third, the search terms were a mess. Broad match with a thin negative list had the ads showing on recipe searches, on "how to make brisket rub" and on the names of two competitors' products the store did not stock. People clicked. Of course they clicked. They were never going to buy.

What did Mavix change in the account?

Five moves, in a deliberate order. Nothing exotic, and no new budget.

 

The ordering is the part people skip. Fixing bidding while the tracking is still lying to you just makes the account confidently wrong. Fixing tracking first means the next four changes have something honest to optimise against.

Why did it work?

Because the account stopped buying the wrong thing. That's close to the whole answer. The budget did not change much, the offer did not change, and the website got only minor work. What changed was the definition of success that the bidding was pointed at.

There is a second reason, and it is less flattering to the industry: a lot of small accounts are not underperforming because of strategy. They are underperforming because of plumbing. A purchase tag with no value in it is a plumbing problem. A missing negative keyword list is a plumbing problem. These get diagnosed in an afternoon and then ignored for a year because they are not interesting to talk about in a monthly report.

The third reason is the average order value of $30.85. At that price a $4.77 cost per conversion leaves room. If the same cost per conversion had come with a $12 average order, the account would have been busy and still losing money. Cheap conversions are only good news relative to what a conversion is worth to you.

What did the first eight weeks look like in practice?

Here is the sequence, roughly as it ran. If you are doing this yourself, this is a workable order.

1. Week one: audit the conversion actions. List every one, note which are primary, note which pass a value. Don't change bids yet.

2. Week one: pull 90 days of search terms into a spreadsheet, sort by cost, and read the top 200 by spend. Not the top by clicks. By cost.

3. Week two: rebuild the conversion setup. Purchases primary with dynamic value, everything else secondary. Expect the reported conversion count to drop hard, because it was inflated before.

4. Week two: add the negative keyword lists that came out of the search term read, and apply them at account level so they can't be forgotten.

5. Week three: split brand out into its own campaign so it stops subsidising the numbers on everything else.

6. Weeks three and four: fix Merchant Center. Titles that lead with what the product is, correct GTINs, clear the disapprovals.

7. Weeks four to six: leave the bidding alone and let conversion history rebuild on the corrected tracking. This is the hardest step because it looks like doing nothing.

8. Weeks six to eight: switch to value based bidding, set a target you can defend from the data you now have, then stop touching it for two weeks.

Step seven is where most people break the sequence. The account looks worse on the reported numbers right after step three, because it stopped counting signups as sales, and the temptation to intervene is strong. Intervening resets the learning and buys another fortnight of noise.

Is 6.45x good, or just normal?

It's good for a store at this price point, and it's well above the public averages, but a single multiple is not comparable across businesses. Margin and average order value decide whether any given number is worth having.

 

WordStream's 2025 study, which Search Engine Land also covered, put the all industry average conversion rate at 7.52% and the average cost per lead at $70.11, a rise of about 5% on the year before. Ecommerce search sits well below the blended figure at 2.81%, which surprises people who compare their store against an average dominated by lead generation.

Against that backdrop a $4.77 cost per conversion reads as very cheap, and it is, but the comparison is not like for like. A $70.11 lead for a law firm and a $4.77 order for a rub company are different units of the same word. The honest way to judge your own account is against your own contribution margin, not against a blended industry number.

What do people get wrong here, and what does it cost them?

They add budget to an account that is broken. It's the most expensive mistake in small business paid search, and it's the one most often recommended.

If your tracking counts newsletter signups as sales, doubling the budget doubles the newsletter signups. If your search terms are full of recipe queries, doubling the budget buys more recipe readers. The account gets worse in absolute terms while the reported conversion count goes up, which is why the mistake survives so long. Estimates of how much small business budget goes to irrelevant clicks vary widely by source and by how well the account is run, so treat any single percentage you see quoted as an estimate rather than a measurement. The direction is not in dispute.

The second mistake is treating Performance Max as a black box and leaving it there. That excuse got weaker in 2026. Google rolled out a proper search terms report for Performance Max along with campaign and account level negative keyword support and channel level reporting, which several PPC trade outlets covered through the year and Google documents in its own Ads Help pages. If you have a PMax campaign you have not opened the search terms report on, that's a job for this week.

The third is changing three things at once and then arguing about which one worked. Change one thing, wait out the learning period, write down what happened. It's slower and it's the only way you end up knowing anything.

Does any of this transfer to a service business?

Most of it does. The tracking discipline, the search term work and the brand split apply identically whether you sell rubs or roof repairs. The product feed step obviously doesn't, and value based bidding is harder because a form fill has no natural price attached to it.

For a service business the equivalent of passing revenue is passing a lead value. Give a booked consultation a different value from a general enquiry, and a quote request a different value again. It doesn't need to be exact. It needs to be directionally right and consistent, so the bidding can tell a good lead from a cheap one. An account that reports every form fill as one identical conversion is asking Google to optimise for volume and then acting surprised when it delivers volume.

Cost per conversion will also look nothing like $4.77. For most Texas service businesses a realistic number sits far closer to that $70.11 all industry average, and for legal or medical work it can run several times higher. Judge it against what a customer is worth to you over a year, not against a barbecue rub company.

What should you check in your own account first?

Two screens, in this order. Conversions, then search terms. Both are free, both take under an hour, and between them they explain most underperforming small accounts.

On the conversions screen, read the Goal column and the Value column together. Every conversion action marked Primary is something you are paying Google to go and find more of. If a newsletter signup is sitting there as Primary, you have found your problem before lunch. Then check whether purchases pass a real value or a fixed one.

On the search terms report, set the date range to the last 90 days, sort by cost descending, and read down the list until you stop recognising your own business. Everything below that line is a negative keyword. Sorting by cost rather than by clicks is the small change that makes this exercise useful, because the queries wasting real money are rarely the ones with the most clicks.

If paid search is only part of the picture and you also want the organic side, the trade offs between the two channels in this state are worth understanding before you commit a budget to either. We covered the organic side for Texas businesses here: https://mavixmarketing.com/blog/best-seo-agency-in-texas-for-startups-and-growing-businesses-2026

FAQ

Q) 1 What does 6.45 conversion value per cost actually mean?

Every unit of currency spent on ads was matched by 6.45 units of tracked conversion value. It's what most people mean by ROAS. It counts tracked revenue before cost of goods, shipping and returns, so it is not the same as profit.

Q) 2 Is $4.77 per conversion realistic for a small business?

For a low priced ecommerce product with clean tracking, yes. For a service business, no. The all industry average cost per lead in WordStream's 2025 data was $70.11, and service categories sit well above ecommerce on cost.

Q) 3 How long did this take?

The structural work ran over about eight weeks, and most of that was waiting rather than doing. Rebuilding conversion tracking and negative keyword lists takes days. Letting the bidding relearn on corrected data takes weeks, and rushing it undoes the rest.

Q) 4 Do I need to spend more to get results like this?

No, and adding budget to an account with broken tracking usually makes things worse. Fix what the account counts as a sale before you change what it costs you.

Q) 5 Should I be running Performance Max?

It's worth testing now that the reporting has caught up. Google added a full search terms report, negative keyword support at campaign and account level, and channel level reporting through 2026, so the old complaint about visibility carries less weight than it did.

Q) 6 What if my conversion volume is too low for value based bidding?

Then stay on a simpler strategy until it isn't. Google's guidance points to roughly 50 conversions in the trailing 30 days before value based bidding has enough to work with, and switching early gives you an expensive random number generator.

Q) 7 Can I do this myself?

The audit part, yes. The two screens above will tell you most of what is wrong. The rebuild is where people lose time, mainly because the account looks worse for a few weeks before it looks better and it takes some nerve to sit through that.

Q) 8 How do paid ads and SEO compare for a Texas small business?

They answer different questions. Ads buy visibility today at a price per click, organic earns it over months and then costs nothing per visit. Small businesses account for 99.8% of all Texas businesses according to the SBA's 2025 state profile, so the competition on both is real. We wrote about the organic side here:  https://mavixmarketing.com/blog/mavix-digital-marketing-agency-the-leading-seo-agency-in-texas-for-affordable-results-driven-growth 

Conclusion

This account did not need a clever strategy. It needed someone to notice that it was counting the wrong things, then to fix them in an order that let each fix mean something. Six point four five is the headline, and the boring plumbing work is what produced it.

Most accounts we open have at least one of these three faults, and the owner is usually the last person to be told. So it's worth asking the question directly: when did you last read your own search terms report, sorted by cost?

Get a free 15 minute Google Ads teardown

Send us the account and we'll read the conversion actions and the search terms and tell you plainly what is wrong with them, in writing, with no pitch attached. Book it at  https://cal.id/mavixmarketing/strategy-call