How we research, write, and correct.
We publish operational writing about how to spend money on advertising, which is a subject where being confidently wrong has a price. This page states how the writing is produced, where the numbers come from, and what we do when we get something wrong — so you can decide how much weight to give any of it.
Who writes it
Everything published here is written by people who operate paid media accounts, not by a content team working from a keyword brief. That is the whole basis on which we ask you to take any of it seriously.
Bylines work two ways. Where a named person wrote and owns a piece, their name is on it and links to a bio page listing what they actually work on. Where a piece reflects the collective method rather than one person’s argument — a checklist we all audit against, a benchmark table compiled from many accounts — it carries the company byline. We do not attach a named author to something that person did not write in order to make the page look more authoritative.
Where the numbers come from
Figures in our writing come from one of three places, and we say which:
- Accounts we operate. Aggregated across clients and never attributed to a named brand without that client’s explicit agreement. Case studies are the exception — those are published with permission and name the client.
- The platforms’ own published documentation and disclosures. Cited when the exact wording matters, because Google and Meta both revise these quietly.
- Our own estimates and rules of thumb. Labeled as such in the text. When we say a threshold is "roughly" or "around" a number, that is not hedging for style — it means the number is a working heuristic from experience rather than a measured constant.
What we do not do
No affiliate links, anywhere. No paid placements, sponsored mentions, or vendor-funded reviews. Nobody pays to be recommended here, and no tool we name has any commercial relationship with us unless the page says so explicitly.
We do not publish comparison content designed to lose. If we write about when a freelancer beats an agency, or when you should keep paid media in house, the answer is the one we would give a friend rather than the one that routes to a contact form. Several pieces on this site end by telling a segment of readers not to hire us.
We do not restate a platform’s help center and call it analysis. If a piece does not contain something you could only know from operating an account, it should not exist.
What a piece has to earn before we publish it
Most advertising content is a rearrangement of the same received wisdom, and it is written that way because rearranging is safe. We hold pieces to a harder standard, and it is worth stating so you can check whether we met it:
- It changes how you see the problem. Every piece has to contain at least one idea that inverts the conventional read of its subject — earned from running accounts, not invented to sound contrarian. A piece with nothing to overturn is a supporting note on one that has, and we file it that way.
- It shows the arithmetic. Where a claim depends on a calculation, the calculation is on the page with numbers you can re-run against your own account. "This improves ROAS" without the working is an assertion, and assertions are what you already have too many of.
- It gives you a decision rule, not a description. "Do this unless that, and that means this threshold." Hedged advice is easy to write and impossible to act on.
- It answers the objection you are already forming. If a smart reader would push back three paragraphs in, the pushback is addressed in the piece rather than left hanging.
- It tells you where the advice stops working. Every substantial piece states its own limits — the conditions under which we would tell you to ignore it, and the things the data genuinely cannot settle. This is the section most publishers leave out, and leaving it out is how you end up trusting advice that does not apply to you.
An honest note about the older archive
The standard above is the standard going forward, and not every older piece on this site meets it yet. A number of earlier articles are correct but thinner than they should be — they make a good argument in eight hundred words where the subject deserved two thousand, and they stop short of the edge cases.
We are rewriting those rather than quietly deleting or reshuffling them, working through the ones that matter most to a buying decision first, and each revised piece shows its revision date. We would rather tell you the archive is uneven than let you discover it.
Accuracy and freshness
Paid media documentation goes stale fast, and stale advice on this subject is not merely unhelpful — it costs money. Pieces that describe platform mechanics are reviewed when the platform changes them, not on a calendar, and a piece whose substance has changed shows its revision date rather than quietly keeping its original one.
Benchmarks and cost figures carry the year in their title deliberately. A CPM benchmark without a date attached is a number pretending to be a constant.
Corrections
If something here is wrong, we would rather hear it than not. Email hi@adsrunner.com and say which page and what is wrong.
Substantive errors — a wrong figure, a mechanic we described backwards, a recommendation that would lose money — are corrected in place and the revision date updated. We do not silently delete a piece that turned out to be wrong, because the reader who acted on it deserves to find the correction rather than a 404. Typos and clarity edits are made without ceremony.
Where AI is and is not used
We build AI infrastructure for our own account operations, so it would be strange to pretend we do not use it. Being specific matters more than being pious about it.
AI is used in our product to analyze account data — that is what the audit and the platform do, and it is the thing we are actually good at. In our writing, models assist with editing passes, structural suggestions, and checking a draft against our own data. What does not happen is a model generating a published article: the argument, the judgment about what is worth saying, and the operational experience behind it come from a person, because a model has never run a budget and cannot tell you what it feels like when one goes wrong.
Nothing here is published without a person who operates accounts reading it and agreeing with it.
Who you can hold responsible.
Sam Nouri
Founder of ADSRUNNER. Twelve years running paid media, now building the AI infrastructure the agency runs on.
Sophie Mills
Performance marketing strategist at ADSRUNNER, working across Google and Meta for ecommerce and lead-gen accounts.
Questions or corrections: hi@adsrunner.com. Our writing lives in insights.