The standard
The standard you can hold us to
Any agency can call itself thorough. Here is what we actually do on an ordinary Tuesday, with the numbers you can hold us to, and what happens when we miss one.
- Published commitments, each with a threshold
- 15
- Things we turn down, each with what it costs us
- 3
- Between full account reviews
- 3 days
Before launch, every week and every month
Any agency can call itself thorough. Below is what we actually do on an ordinary Tuesday, with the numbers you can hold us to. We publish it because you can check a list, and you cannot check a résumé.
Before anything goes live
Tracking is proved before spend starts, not after
We submit a real form and place a real call, then follow both through to Google Ads, Meta and your CRM, checking that the value which arrived is the value we sent. You get the screenshots. If a conversion doesn't land in all three, nothing launches — including the campaign you're in a hurry for.
The negative keyword list exists on day one
Built from twelve months of your own search terms plus the known waste in your category, before the first impression. Each one carries a written note on what relevant traffic it might cost you. We do not blanket-block prices, reviews, competitor names, your own industry's name, or local terms — in a spreadsheet those look like waste, and in a search box they are usually the buyer.
Every search ad group launches with 15 headlines, 4 descriptions, and almost nothing pinned
We pin only what has to hold still: a licence number, a regulated price, a required disclaimer. Everything else stays free to compete. If you inherit an account from us with a headline pinned to position one for convenience, that is us breaking our own rule and you can say so.
Performance Max launches with brand traffic excluded
We block your own company name before the campaign starts, not after the first month looks suspiciously good. And we split the campaign up by the jobs you actually want. Dump everything into one bucket and the report looks great while the money goes to whatever was cheapest.
Every week
Search terms are read weekly, and the negative list changes the same day
We read every search term that got more than one click. Below that line we sample, because at $2,000 a month most of those terms got exactly one click each, and we would rather tell you we sampled than pretend we read all of them.
Budget pacing is checked Monday and Thursday
Any campaign more than 20% off its month-to-date pace is corrected or flagged to you that day. Nobody finds out on the 30th that a third of the month went into one weekend.
Meta frequency and exclusions, weekly, on every ad set
Cold audiences past 3.0 frequency in a rolling seven days go into replacement, not onto a watchlist. Converters, existing customers and job applicants come out of prospecting the week they appear, not the quarter after someone notices the cost drifting.
One senior person reads your whole account every week
The agents run the checks and build the queue; a named person then reads the campaigns, the ads, the search terms and the landing pages themselves. A list of what software found is not the same object as an account, and we don't treat it as one.
Every month
The written review lands by the fifth business day, and it includes what didn't work
Numbers from your ad accounts and your CRM, not from a dashboard we control. Every review names at least one thing we got wrong or one thing we still cannot explain. A month where we can't find one is a month we didn't look hard enough.
New creative ships monthly whether or not the current set is still working
Four new statics and one new video per active campaign, minimum. The replacement has to exist before fatigue arrives — building it afterwards means two bad weeks that you paid for and we could have avoided.
Every form and tracked number is tested live on the first business day
A real submission, a real call, confirmed all the way into the CRM. Broken forms are the most expensive silent failure in this work, and they are found by testing, not by noticing revenue slipped. The same day, every page we shipped is re-run for speed: under 2.5 seconds to largest contentful paint on a mid-range Android over 4G, or it goes back to the top of the queue.
The same 25 buyer questions, run against AI answers, every month, dated
Google AI Overviews, ChatGPT and Perplexity. The same list each time, so a change means movement rather than a different question. You see where you appeared, where a competitor did, and where the answer named nobody — that last column is the opening.
How a decision gets made
One change per campaign per measurement window, with the expected result written down first
Ship five edits in a week and nobody can say which one moved the number, including us. Our prediction sits in the log next to the outcome, so you can see how often we were right.
The threshold and both responses are declared before the measurement, never after
“We run this to 400 clicks; under 6 conversions we turn it off, over 14 we scale it, in between we hold.” Decided in advance, in writing. A rule invented once the data is in is not a decision rule — it is a description of the data.
Every change carries a timestamp, a reason, and the name of the person who approved it
You get that log always, without asking. Nothing on your account was done by “the algorithm”.
What we refuse to do, and what refusing costs us
No percentage of your ad spend, no platform rebate, no referral fee from a tool we recommend
Ten percent of a $40,000 monthly budget is $4,000 — four times our entire fee, from one client, for identical work. We turn that down on purpose. A fee that grows with your spend makes “spend less here” a sentence that costs us money to say, and it is a sentence you need us able to say.
We don't start on an account we can't measure
If we cannot check that your tracking works, or your CRM will not tell us which jobs got sold, we fix that before anything else. If it cannot be fixed, we do not take the account. This is the condition that loses us signups, every one of which would have paid perfectly well while we spent into the dark.
We don't sign more clients than the bench can read every week
The weekly read above is a real hour from a real person, and there is a number of accounts past which it quietly becomes a skim. When we are at that number you get a start date instead of an invoice.
When we miss one
We will miss some of these. When we do, it goes in your monthly review before you find it — which standard, which date, what it cost you, and what changed so it does not repeat. If a pre-launch standard is not met, the campaign does not launch until it is fixed.