Paid media strategies, benchmarks and playbooks for B2B pipeline and eCommerce revenue.
47 articles, page 1 of 6
Churn rate is the share of customers or revenue you lose in a period. The number is quoted constantly without the period attached, which is how the same 5% gets used as a healthy target and as nine times itself. Here is the formula, the dated benchmarks, and the arithmetic that turns churn into a cost per acquisition ceiling.
Most performance marketing agency rankings recycle directory blurbs. This one opened every case study, re-resolved every domain, and recorded what the numbers actually say.
Cost per mille is what you pay for one thousand ad impressions. Measured across the accounts we manage it ran from $5.46 to $320.80 inside the same quarter, for the same advertisers, which is why no single average CPM is worth quoting. Here is the formula, the real numbers, and the three jobs the metric is good for.
Most multi-location advertisers set up Google Ads to mirror the org chart: one campaign, sometimes one account, per branch. In an account we manage, merging 14 location accounts into one cut cost per lead by 30%, and a later attempt to split five markets back out produced the five worst performers in the account. The money was never lost at the metro level. It was lost in pockets below it, which no amount of splitting finds.
A sales qualified lead is a lead a salesperson has accepted as worth working, against criteria written down in advance. Here is the definition, what the published versions of it actually say, the one conversion rate worth calculating, the fifteen-year-old survey behind the statistic everyone quotes, and the ninety-day limit that decides which qualification stage you can send back to LinkedIn.
Most Google Ads agency rankings recycle directory blurbs. This one opened every case study, re-resolved every domain, and recorded what the numbers actually say.
Almost every guide to Google Ads Quality Score promises a cost per click discount for raising the number. Google's help documentation says, in one sentence, that Quality Score is not an input in the ad auction. Both things cannot be true. This is what the score actually measures, what the auction uses instead, and how to read the three component columns as the diagnostic they were built to be.
Most Bing Ads versus Google Ads comparisons stop at cost per click, which is the one number that is almost always lower on Microsoft and almost never the number that decides anything. In one business to business software account the cheaper clicks were real and the cheaper conversions were not, and three platform defaults nearly invalidated the test before it produced a figure.
Customer retention rate is the percentage of the customers you had at the start of a period who are still customers at the end of it, counting nobody you acquired in between. Here is the formula, the current benchmarks from primary data, the arithmetic that turns a retention rate into an allowable acquisition cost, and the reason the widely repeated claim that five percent more retention produces twenty five percent more profit is true at one starting point and wrong everywhere else.
A practitioner's comparison of 15 B2B paid media agencies, with the channels each runs, the pipeline they can prove, and the type of technology company each is built for.
Almost every landing page A/B test is decided on the first conversion in the funnel, because that is the only event with enough volume to reach significance quickly. In one account we manage that choice would have picked the losing page. Both arms landed within a dollar of each other on cost per trial across four weeks, while every self-serve purchase in the window came from one of them. The signup number was not wrong, it was measuring the wrong thing.
A marketing qualified lead is whatever two teams at one company agree it is, which is why the benchmarks for it are close to meaningless. What matters for a paid media budget is narrower and checkable: which lead stage you can send back to Google, Meta and LinkedIn, and whether that stage produces enough events, fast enough, for the bidding to learn from it.
Target ROAS is a ratio over the conversion value your tag reports, which is almost always gross order value. Returns, payment fees, fulfilment and spend on products nobody can buy all sit outside that ratio, so raising the target does not fix a profit problem that lives below it. In one consumer electronics account the handed-down break-even was 5.7, the profit and loss said about 5.34, and the real money was leaking somewhere the bidding algorithm could not see.
Net revenue retention is the percentage of recurring revenue a cohort of existing customers still produces a year later, after upgrades, downgrades and cancellations. Here is the formula, the current medians from the largest private-market survey, why the widely quoted 120 percent figure comes from a different kind of company measured over a different window, and how retention sets the ceiling on what you can pay to acquire a customer.
Most advice about Google Ads account structure is still arguing about a job structure no longer has. Google publishes the four steps it uses to pick which ad group serves a search, and the account you build sits underneath those steps rather than above them. This is what structure still decides, what it cannot decide any more, the limits Google actually publishes, and the order in which to rebuild an account that was organized for the old rules.
Customer lifetime value is the profit a customer produces over the whole relationship, discounted to what it is worth today. The formula most published guides give leaves out gross margin and the discount rate, which on median business-to-business software retention returns a number about three times too high. Here is the arithmetic, the three methods, and how to turn the answer into a cost per lead you can bid against.
Most advice about choosing an attribution model in Google Ads is describing a menu that Google took away in 2023. First click, linear, time decay and position-based are gone, and conversion actions that used them were migrated to data-driven. That leaves two models, which means the decision that used to sit in the model dropdown has moved to the conversion window and the primary conversion actions. This is where the real control is now.
Most in-house marketing team vs agency comparisons set a base salary next to a monthly retainer and call it arithmetic. Here is the same decision with the employer costs measured at source, the platform coverage counted, and a spend threshold you can test against your own numbers.
Average order value is total revenue divided by number of orders. That much is agreed. What is not agreed is which revenue, and the answer changes the number: Shopify and Google Analytics 4 compute it differently, returns are not deducted from either, and a product mix shift moves AOV without a single customer changing their behaviour. Here is the formula, the measured benchmarks, and the version of the metric your bids can actually use.
Meta reviews the destination, not only the creative. When a deleted product page shows a not-found message to a human but returns the status code for a working page, the ad promises a product and delivers a dead end, which is the exact shape of a misleading link under Meta's own spam policy. Because ads are re-reviewed at any time and a catalogue re-crawls itself, each pass finds the same fault again and enforcement climbs from the ad to the account. The fix took one change on the website. Finding it took killing a more plausible theory first.
CAC payback period is the number of months of gross profit it takes to repay what you spent acquiring a customer. Here is the formula the benchmark reports actually use, the current medians by company size, why the famous 12-month rule was published as a failure line rather than a target, and how payback sets the ceiling on your ad budget.
Most negative keyword advice assumes negatives behave like keywords in reverse. They do not. Google's own documentation says negative keywords will not match close variants or other expansions, which means a downloaded list of root words blocks far less traffic than its length suggests, while a single-word negative broad blocks far more than you intended. This is how the match types, the four levels, and the current limits actually work.
Every Google Ads bid strategy optimizes for exactly one thing, and most account problems come from asking one to deliver something else. This is what each strategy does, what Google documents about the conversion volume each needs, why the learning period is counted in conversion cycles rather than days, and what the August 2026 change to target-based bidding did to campaigns that were quietly beating their targets.
Optimization score is not a measure of how well your account is performing. Google's own documentation says it is calculated partly from your recent recommendations history, which means it partly measures whether you have been reviewing the suggestions. When a client asked us to rule on four recommendations from her Google representative, the account data agreed with one of them. The other three described a problem that was not happening, a campaign that was already at zero spend, and a budget constraint the delivery data did not show. Meanwhile a campaign nobody had flagged was buying trials at roughly twice the cost of search.
Google Ads cost whatever the auction charges, and there has never been a price list. What Google does publish is more useful than any average: you are charged the minimum needed to beat the advertiser below you, a campaign can spend up to twice its average daily budget in a single day, and the month is capped at 30.4 times that budget. Meanwhile the industry's most-repeated average cost per click comes from data collected in 2017. Here is what each number is worth, and how to produce your own in 30 days.
Culture, passion and chemistry cannot be verified before you sign. Account ownership, fee structure, change history and the exact thresholds behind a Google Partner badge can. This guide rebuilds the selection process around evidence you can check yourself, including the warning signs Google publishes in its own help pages and the three numbers the Partner badge actually certifies.
Facebook ads cost whatever the auction charges, and Meta has never published a rate card. What Meta does publish is more useful than any average: a 175% daily spending limit, a weekly cap of seven times your daily budget, an auction price floor it will not disclose, and two separate budget rules that contradict each other by a factor of two. Here is what each of those means for the number you will actually pay, and how to find your own cost floor in 30 days.
Most teams report one lead conversion rate and steer on it. A paid media budget crosses four separate conversion rates on its way to a customer, they multiply rather than average, and the published benchmarks for each one are measured on different denominators. Here is the formula, the 2026 benchmark tables with their sample sizes attached, the widely repeated figure that turns out to have no original source, and the arithmetic that tells you which stage is costing you the most.
LinkedIn Ads cost is decided by an auction, not a rate card, and LinkedIn's own pricing page contains no prices at all. The published averages contradict each other by a factor of three on clicks and five on impressions. Here is what LinkedIn does document, how to pull the bid floor and suggested bid for your exact audience, why a cost per click cannot be compared across campaign objectives, and how to work back from break-even cost per lead to a budget that survives a quarter.
Most guides to Google Ads conversion tracking stop once the tag fires. The tag is the easy part. What decides whether Smart Bidding buys customers or form interactions is five settings underneath it, plus one documented exception that quietly overrides two of them.
Cost per lead is total spend divided by leads generated. The useful question is not what the average CPL in your industry is, but what your own economics allow: gross profit per customer multiplied by your close rate is your break-even cost per lead, and a good CPL sits 20% to 30% below it. Here is the formula, the 2026 benchmark tables, the reason two respected datasets disagree by up to 6x for the same industry, and when a rising CPL is the best news in the account.
Most rankings of the best PPC agencies rewrite the same directory blurbs. This one opened every agency's case studies and checked whether the numbers hold up.
Break-even ROAS is the return on ad spend at which advertising stops losing money. The formula everybody publishes uses gross margin and understates the real floor by more than 80% on a typical store. Here is the version that includes fees, fulfilment and returns.
Performance Max stopped being a black box some time in 2025, and most advice about it has not caught up. Here is what the campaign type optimises for, every control that exists today, what the reporting now exposes, and the two 2026 changes that quietly moved the numbers in live accounts.
The LTV to CAC ratio divides customer lifetime value by customer acquisition cost. Here is how to calculate both halves properly, what the 3:1 rule really says in its original source, why CAC payback period is the sturdier number, and why the ratio decays as you scale spend.
Customer acquisition cost is total sales and marketing spend divided by new customers acquired. Here is how to calculate CAC from ad platform data that counts conversions instead of customers, the break-even number your margin implies, and where the 3:1 LTV to CAC rule actually came from.
MER is total revenue divided by total marketing spend. Here is the formula, what a good marketing efficiency ratio looks like by revenue stage, the break-even number your margin implies, and the rule for deciding when to act on MER instead of platform ROAS.
Return on ad spend is one division, and almost everyone gets the inputs wrong. Here is the ROAS formula, what the number means, the four ROAS figures a single account can produce, and the percentage-versus-ratio trap in Google's target ROAS field.
A practitioner's comparison of 12 PPC agencies for B2B SaaS, with the channels each runs, the trials and pipeline they can prove, and the type of SaaS each is built for.
A practitioner's comparison of 12 Meta Ads agencies for DTC ecommerce, with the way each runs Facebook and Instagram, the results they can prove, and the brand each is built for.
A practitioner's comparison of 11 Google Shopping agencies for ecommerce, with the feed and Performance Max approach each runs, the ROAS and revenue they can prove, and the type of online store each is built for.
A practitioner's comparison of 11 LinkedIn Ads agencies for B2B, with the account structure each runs, the pipeline and cost per lead they can prove, and the type of company each is built for.
A practitioner's guide to what a good ROAS actually is in 2026: benchmarks by channel and industry, the break-even math that decides your real target, and why the headline number matters less than the profit underneath it.
Most Bing Ads agency lists repeat partner claims nobody checked. This one ranks 11 agencies against Microsoft's own partner directory, names the tier each one actually holds, and flags the CPC statistic the whole category quotes without a source.
A practitioner's Google Ads audit checklist: the 12 checks that find wasted spend in a live account, the order that stops you fixing the wrong thing first, and what the 2026 automation changes mean for the review.
Most PPC pricing guides are written for agencies deciding what to charge. This one is written for the person paying the invoice: real 2026 fee ranges, the four models compared, and a profit-based test for whether your fee is fair.
A practitioner's comparison of 11 ecommerce PPC agencies, with the channels each runs, the results they can prove, and the type of store each is built for.
