Learn
Methodology we actually use running paid media accounts, written up as full pieces rather than a rule of thumb, plus a reference set for the individual terms and effects those pieces lean on.
Methodology
How to review a bid or budget change before you make it
A platform's own bid simulator, read wrong, will justify a change that delivers a fraction of the volume you expected. Here's the checklist that catches it first.
How we run a deep campaign review
A repeatable method for finding the real story behind a campaign's numbers, built around two derived metrics that do most of the diagnostic work: pool and ceiling.
Why your numbers never quite match what actually happened
Two separate mechanisms make ad reporting drift from reality: platforms quietly redefine how they count conversions, and platform numbers were never meant to equal your actual customer count.
Why "yesterday's numbers" are always wrong, and keep climbing for weeks
Search ad platforms count conversions on the day of the click, not the day of the sale, so every recent day looks worse than it will end up being.
Why ad platforms can report more "conversions" than you actually had customers
Meta and Google both claim credit for the same sale more often than you'd expect, and the fix starts with your own order numbers, not theirs.
Comparing CPA across two campaigns can be comparing apples to oranges
Two campaigns can report wildly different cost-per-acquisition numbers for a reason that has nothing to do with which one is actually working: they aren't counting the same thing.
Why one ROAS number doesn't work for a multi-product or multi-offer business
When your site sells more than one thing, a single blended ROAS figure can hide contamination between campaigns and mask which product lines are actually profitable.
Why your Meta CPA data might be lying to you
Meta has to estimate a real share of your conversions instead of directly observing them, which means the CPA number on your dashboard is a guess with a margin of error you were never shown.
Why a 30-day lookback window produces false "wasted spend" flags
The default reporting window in Google Ads and Meta Ads is a fixed 30 days, and using it to decide what counts as wasted spend can both hide real waste and flag things that were about to work.
The negative-keyword match-type mistake that silently blocks profitable traffic
Negative keywords have match types too, and picking the wrong one can quietly wall off real, convertible searches without leaving any trace in your reports.
Google's "Ad Strength" label doesn't predict which ads actually perform
Ad Strength grades how complete and varied your ad's assets are, not whether the ad actually converts, and chasing the label can quietly make your ads worse.
We looked at the popular advice on writing Responsive Search Ad copy. Here's what actually holds up.
Pinning, headline length, sentence case, keyword repetition: we separate the RSA copy advice that has a plausible mechanism behind it from the folk wisdom that doesn't.
Your "daily budget" isn't what you actually spent yesterday
Google and Meta can both spend above your stated daily budget on a given day, by design, as long as the longer-term average holds. Here's the mechanism.
Budget-capped or rank-capped? The diagnostic that changes everything
A stalled campaign looks the same whether the misses were attributed to your budget or to your Ad Rank. Google reports how much time it lost to each, in two columns.
Target CPA vs Target ROAS: which one actually fits how your business makes money
Most explanations define both bidding strategies and stop. This one tells you which to actually run, based on how your conversions create value.
The Bid Change That Looked Like a Mistake for Two Weeks, Then Worked
A hypothetical founder raised a bid target to chase more volume, panicked when the numbers dipped, and reverted right as the campaign was about to turn a corner.
Why the same bid increase can do nothing in one campaign and double traffic in another
A 20% bid increase can be a rounding error in one campaign and a traffic-doubling event in another, and the difference has nothing to do with luck.
Rising impression share isn't always a win
A climbing impression share looks like unambiguous progress on a dashboard, but the metric is a ratio, and ratios rise when the bottom number falls.
How to Tell a Real Demand Ceiling From a Delivery Problem
When spend stops converting or a campaign plateaus, two very different causes produce the identical symptom, and treating one like the other wastes budget and time.
The Verdict-First Framework for Answering "How Are the Ads Doing?"
Most ad reports bury the answer under a pile of charts. Here's why the verdict has to come first, and what that actually requires of the analyst writing it.
Nine ways a confident-sounding ad account recommendation turns out to be wrong
A checklist for catching the mistakes that hide inside plausible-sounding ad account analysis, before you act on them or approve someone else's plan.
The Audience Segment That Looked Like a Winner for One Month
A single strong month from a small audience segment can look like a repeatable pattern when it's really a handful of lucky conversions, and scaling budget into it too fast is how good months turn into wasted spend.
The Week Three Things Changed at Once, and Nobody Could Say Which One Worked
A hypothetical walkthrough of what gets lost when a budget increase, a creative swap, and a negative-keyword cleanup all land in the same week.
Why your ad platform rep's suggestions aren't neutral advice
Google and Meta make more money when you spend more, and their recommendations are structured accordingly. Here's how to evaluate any suggestion on its own merits.
When the platform's AI advisor blames a competitor, check the math
Google and Meta now auto-generate plain-language explanations for metric changes. Here's how to verify one instead of trusting the sentence.
Reading auction insights: "the market got more competitive" vs "you're losing"
A worsening Auction Insights report gets blamed on the market by default. Often the real cause is sitting in your own account settings, waiting to be checked.
Signs Your Ad Tracking Has Gaps, Spottable From Outside Your Own Funnel
Five comparisons anyone can run without platform access or technical skill to sense whether ad tracking numbers can be trusted.
Why yesterday's "best hour to advertise" becomes next quarter's worst
Dayparting rules built on a few weeks of data get treated as permanent facts, but the pattern behind them keeps moving even when nothing you did changed.
The grace conversions rule: how to cut a keyword before the data is significant
Small accounts never reach statistical significance. Here is the rule we use instead: give a keyword the conversions it has not earned, for free, and see whether it is still worse than its peers.
Reference
One page per effect: what it is, where you meet it, why it happens and what it costs.
Measurement and attribution
Claim ratio
Platform-reported conversions divided by what your own records say actually happened, tracked per channel over time.
Why platform conversions do not match your real sales
Platform-reported conversions are a claim about reality, and they are wrong in both directions, sometimes by an order of magnitude.
When a conversion drop is attribution, not performance
Reported conversions can fall hard while the business is flat, because a sibling channel scaled up and took the credit.
Conversion capture rate
The share of real leads the platform can see, and the exact factor by which a low share inflates every cost it reports.
How much of your business ad platforms actually see
Settled totals are all-channel truth. Platform numbers are causal claims about a subset, and the relationship between them breaks the moment anything changes.
Conversion completeness
How much of a period's conversions have arrived yet, which is a property of the individual account and never a rule of thumb.
Platform mechanics
How the three quality components combine into one score
Every combination of the three published Quality Score components maps to one fixed integer, and the whole mapping can be read off measured data.
How often keywords carry no Quality Score at all
A blank Quality Score is an absence rather than a low score, and it covers a large share of a normal account.
How much a keyword's quality rating moves in a month
A two point move inside a month is the ordinary case rather than a signal, which is what makes a weekly Quality Score ritual expensive.
CPC went up: is it your own bids or the competition?
Before blaming a rival for a higher cost per click, read your own impression share and search top impression share. They move in opposite directions for the two causes.
Search intent and negatives
How much spend the search terms report covers
The search terms report only shows the queries Google chose to surface, and on one large account it covered under half of total spend.
Filtering a very large search terms report
Most rows in a big search terms pull never spent a penny, and filtering them out is what makes the report usable at all.
Match type on a negative keyword
The match type you pick for a negative keyword decides how much converting traffic it quietly cuts off.
Spend concentration across search terms
Ranking every search term by spend shows that a small band at the top carries almost all the money, which is where a search term audit should start.