Guides
3 pieces
What changed most recently, newest first.
11 min read
Ten AI SEO agents get ranked by how many pipeline stages they automate. None of the rankings measures whether the output ranks, and the one practitioner reporting a number writes the pages himself.
11 min read
Two SaaS SEO guides publish conversion percentages for SaaS search, and neither says what converted. A third goes further and reports demo requests and trial signups as one measure. On organic traffic the trial to paid rate is 18.2 percent without a card and 48.8 percent with one, so a rate quoted without its event is not a number you can plan against.
11 min read
An ideal customer profile is usually described as a filter you derive from your best existing customers. Lenny Rachitsky's interviews with more than twenty-five founders describe something else: a hypothesis you find by selling, and most got the first version wrong.
Search is not one place anymore.
17 pieces
11 min read
Two SaaS SEO guides publish conversion percentages for SaaS search, and neither says what converted. A third goes further and reports demo requests and trial signups as one measure. On organic traffic the trial to paid rate is 18.2 percent without a card and 48.8 percent with one, so a rate quoted without its event is not a number you can plan against.
11 min read
An ideal customer profile is usually described as a filter you derive from your best existing customers. Lenny Rachitsky's interviews with more than twenty-five founders describe something else: a hypothesis you find by selling, and most got the first version wrong.
11 min read
A strategy is a decision about what you will not do. The published guidance names no order, no budget and no exclusion. What is left is a list of activities and no way to choose. Forrester goes furthest, naming channels only to reject them as strategies.
10 min read
Published sets of SaaS metrics run from five entries to twenty-two, and the widest of them prints the metrics side by side with no arithmetic between them. Several of them contain each other, and one name carries two quantities that produce two verdicts on the same quarter.
11 min read
Search this term and you get galleries holding more than ten thousand example pages. Across the eight readable ones, no conversion rate appears beside a screenshot. Unbounce's 41,000-page benchmark puts SaaS at a 3.8 percent median, well under the 6.6 percent it reports across all industries.
12 min read
Product-led growth means the product does the selling. The model gets explained everywhere and priced nowhere. A salesperson carries a published quota, and dividing it by your price gives you the number of customers one rep would have to close.
12 min read
Account-based marketing means treating a named list of companies as the market. The usual advice explains how to run it far more than whether it fits you. This one gives you a fit test built from published numbers, and the arithmetic to run it on your own accounts.
11 min read
B2B content marketing gets defined everywhere and priced nowhere. Not one published account names a price or a date. The published numbers that look like an answer measure a demand engine, a sales cycle or a search ranking, and reading them as a content timeline is how a plan gets a date it cannot keep.
12 min read
No published definition of lead nurturing prints a cadence number, and only Oracle says what the timing keys on. The vocabulary that decides what you build sits in the platform documentation: what enrols someone, what delays them, and what happens when they convert mid sequence.
13 min read
The case for demand generation usually opens with a statistic about the buying journey, and three different ones are in circulation. Eight values circulate in all, from 45 percent to 90 percent, and they count different things. The firm behind the most quoted one published a correction the same year.
10 min read
Search saas marketing funnel and half the results describe the road to a purchase while the other half describe what happens after signup. Both call it the SaaS funnel. Here is where each one ends, why the disagreement exists, and which endpoint to pick.
12 min read
A lead score can be derived from what closed, or assigned by judgement. One of those is arithmetic and the other is a guess, and they look identical on the page. Here is each method, and the detail about retroactive rescoring almost nobody mentions.
11 min read
Published models give four stages, three or six. Google suggests a question asking for five. Here is what the count changes, what it does not, and where the handoff sits.
12 min read
Four published reports price a B2B lead. They name four different scopes, publish four different statistics, and not one of them says how many campaigns it looked at. Here is how to read them, and how to build a number of your own.
11 min read
Gartner says buying teams work on their tasks without a consistent order or journey. Stage models say the opposite, and five different stage counts are in circulation. Here is what Gartner publishes, and what gets attributed to Gartner that Gartner never said.
10 min read
The benefit figures quoted for alignment lose a step of attribution at every hop, and the anchor study behind them is from 2010. The part you can write down and check is smaller, and more useful.
10 min read
The published advice converges on the same nine steps. None of them says what content is being chosen instead of, or how much it has to produce to win that comparison. The answer moves by a factor of a hundred with your price.
6 pieces
Being findable in the search results people still click.
11 min read
Ten AI SEO agents get ranked by how many pipeline stages they automate. None of the rankings measures whether the output ranks, and the one practitioner reporting a number writes the pages himself.
10 min read
There is no B2B ranking algorithm, and nobody publishing on this claims otherwise. What changes is the size of the terms, the number of people deciding, and how long a page takes to pay. Two guides publish figures for that.
10 min read
Search volume is the wrong way to pick B2B terms, and the published guidance says so without putting a number in its place. Choosing between two keywords stays a matter of taste. Here is the arithmetic that settles it.
12 min read
Shopify decides much of your technical SEO by default and documents the two problems it creates. Plus eleven public robots.txt files read on one day, serving two different Shopify defaults.
13 min read
Every ecommerce SEO guide is the same four headings. Across three major storefront platforms the audit category named SEO scores 92 to 100 and separates nobody. Here is where the work that is left sits.
12 min read
Most audit checklists mix one thing Google publishes a threshold for with fifty things it does not. Here is the difference, and the ecommerce-only section that generic audits leave out entirely.
14 pieces
Being cited by the assistants that answer instead of linking.
10 min read
OpenAI, Anthropic and Perplexity each publish the crawler that decides whether your pages can appear in their search answers, and each rule is set in robots.txt. Here are the three rules, what they control, and where their published evidence stops.
11 min read
AI SEO means either using AI to do your SEO work, or getting your pages named inside AI answers. Those are different purchases with different mechanisms. Here is how to tell which one you have, what each costs, and what the platforms publish about the second.
11 min read
You cannot apply for a Google knowledge panel. Even the firms that sell the service say so on their own pages. What you can do is make your brand legible enough that Google builds one.
10 min read
Digital PR earns coverage on other people’s sites. Earned media is also where AI answers get most of what they quote: 84 percent of the links they cite come from it, and 0.3 percent from paid.
9 min read
Google Alerts reports what shows up in Google Search. That is a real boundary, and it decides which of the three places your brand gets mentioned you will ever hear about.
10 min read
Google says an unlinked mention passes no signal to its link graph. A published visibility score counts it as its unit. The same publisher argued both sides five years apart, and both times it was right.
10 min read
The word means something different everywhere you look. One audit is entirely measurement, another is mostly diagnosis, and none says how many prompts it ran.
10 min read
A percentage needs something to divide by. In AI share of voice that is the total mentions across a set of prompts somebody picked and usually will not show you. Some tools call that set your whole category, a number nobody can count.
9 min read
Five published procedures, and they disagree about how many times to ask. The disagreement resolves once you notice they are measuring two different things: one prompt, or a portfolio.
11 min read
Visibility scores are rates computed over a prompt set somebody chose, with a formula the vendor picked, not a standard. The formula, the engines and the prompts are each set per tool, and each one changes what is being counted. Here is what an AI visibility score can and cannot support.
12 min read
GEO comes from a 2024 paper with a measured claim. A benchmark tested those methods across six domains and found three of fifty-four combinations significantly positive. This is what the evidence supports, what it does not, and what it costs to measure.
11 min read
Adoption estimates for llms.txt run from 951 domains to 28 percent. The measurements are not in conflict: each counts a different universe. What almost nobody puts next to them is the traffic number.
11 min read
Four questions come up most often, and three are answered word for word in Google’s own documentation, including whether you can turn them off. There is no overview-only switch.
13 min read
Google says no special optimization is needed to appear in AI Overviews. The published studies measured what drives citation, and they disagree. Here is why, and what to do about it.
4 pieces
What the auction rewards, and what it ignores.
12 min read
Performance Max buys all of Google inventory in one campaign. Google publishes both sides of the trade: the controls you keep, and the rule that decides whether your own Search campaign outranks it.
12 min read
Google states that Shopping ads run on your product data, not on keywords. That is why keyword work does so little here. This is what decides whether your products show, and what a small budget can and cannot reach.
9 min read
One term, two products. Google runs a campaign type called Demand Gen and the phrase also names a marketing strategy, and the results mix them. Here is the boundary in Google's own words, and the budget number two sources disagree about by a factor of two.
11 min read
Google publishes six inputs to Ad Rank. Most explainers cover three. The other three are what explain why a perfectly good ad never shows, and why raising your bid does not always fix it.
2 pieces
A channel with a published mechanism and no published returns.
13 min read
OpenAI lists 52 countries where Ads Manager is self-serve. This is the build order, the three settings you cannot undo, the country by country budget floor, and what refuses your ad before a shopper ever sees it.
11 min read
OpenAI documents the auction and the opening bid. Nobody publishes the return. An arXiv audit by Lurie and colleagues of 3,602 ad-bearing conversations shows the channel matches conversations, not shopping queries.
3 pieces
Turning the traffic you already have into revenue, and knowing when a result is real.
13 min read
The definition is easy and the execution is where tests die. What to split on, how to tell the split worked, and the two checks that come before you read a result.
12 min read
Baymard’s 70 percent abandonment figure is an average of 50 studies across 19 years. Here is what it can and cannot tell you, and which checkout numbers you can count yourself.
14 min read
The most-shared A/B test result in ecommerce was examined in a KDD paper and did not survive. Here is what it got wrong, what the published success rates are, and how to run a program that compounds.
8 pieces
The pieces written with a store in front of them.
12 min read
Baymard’s 70 percent abandonment figure is an average of 50 studies across 19 years. Here is what it can and cannot tell you, and which checkout numbers you can count yourself.
12 min read
Performance Max buys all of Google inventory in one campaign. Google publishes both sides of the trade: the controls you keep, and the rule that decides whether your own Search campaign outranks it.
12 min read
Google states that Shopping ads run on your product data, not on keywords. That is why keyword work does so little here. This is what decides whether your products show, and what a small budget can and cannot reach.
12 min read
Shopify decides much of your technical SEO by default and documents the two problems it creates. Plus eleven public robots.txt files read on one day, serving two different Shopify defaults.
13 min read
Every ecommerce SEO guide is the same four headings. Across three major storefront platforms the audit category named SEO scores 92 to 100 and separates nobody. Here is where the work that is left sits.
12 min read
Most audit checklists mix one thing Google publishes a threshold for with fifty things it does not. Here is the difference, and the ecommerce-only section that generic audits leave out entirely.
11 min read
OpenAI documents the auction and the opening bid. Nobody publishes the return. An arXiv audit by Lurie and colleagues of 3,602 ad-bearing conversations shows the channel matches conversations, not shopping queries.
14 min read
The most-shared A/B test result in ecommerce was examined in a KDD paper and did not survive. Here is what it got wrong, what the published success rates are, and how to run a program that compounds.
19 pieces
The pieces written for a purchase several people sign off.
11 min read
Two SaaS SEO guides publish conversion percentages for SaaS search, and neither says what converted. A third goes further and reports demo requests and trial signups as one measure. On organic traffic the trial to paid rate is 18.2 percent without a card and 48.8 percent with one, so a rate quoted without its event is not a number you can plan against.
11 min read
An ideal customer profile is usually described as a filter you derive from your best existing customers. Lenny Rachitsky's interviews with more than twenty-five founders describe something else: a hypothesis you find by selling, and most got the first version wrong.
11 min read
A strategy is a decision about what you will not do. The published guidance names no order, no budget and no exclusion. What is left is a list of activities and no way to choose. Forrester goes furthest, naming channels only to reject them as strategies.
10 min read
Published sets of SaaS metrics run from five entries to twenty-two, and the widest of them prints the metrics side by side with no arithmetic between them. Several of them contain each other, and one name carries two quantities that produce two verdicts on the same quarter.
11 min read
Search this term and you get galleries holding more than ten thousand example pages. Across the eight readable ones, no conversion rate appears beside a screenshot. Unbounce's 41,000-page benchmark puts SaaS at a 3.8 percent median, well under the 6.6 percent it reports across all industries.
12 min read
Product-led growth means the product does the selling. The model gets explained everywhere and priced nowhere. A salesperson carries a published quota, and dividing it by your price gives you the number of customers one rep would have to close.
12 min read
Account-based marketing means treating a named list of companies as the market. The usual advice explains how to run it far more than whether it fits you. This one gives you a fit test built from published numbers, and the arithmetic to run it on your own accounts.
11 min read
B2B content marketing gets defined everywhere and priced nowhere. Not one published account names a price or a date. The published numbers that look like an answer measure a demand engine, a sales cycle or a search ranking, and reading them as a content timeline is how a plan gets a date it cannot keep.
12 min read
No published definition of lead nurturing prints a cadence number, and only Oracle says what the timing keys on. The vocabulary that decides what you build sits in the platform documentation: what enrols someone, what delays them, and what happens when they convert mid sequence.
13 min read
The case for demand generation usually opens with a statistic about the buying journey, and three different ones are in circulation. Eight values circulate in all, from 45 percent to 90 percent, and they count different things. The firm behind the most quoted one published a correction the same year.
10 min read
Search saas marketing funnel and half the results describe the road to a purchase while the other half describe what happens after signup. Both call it the SaaS funnel. Here is where each one ends, why the disagreement exists, and which endpoint to pick.
12 min read
A lead score can be derived from what closed, or assigned by judgement. One of those is arithmetic and the other is a guess, and they look identical on the page. Here is each method, and the detail about retroactive rescoring almost nobody mentions.
11 min read
Published models give four stages, three or six. Google suggests a question asking for five. Here is what the count changes, what it does not, and where the handoff sits.
12 min read
Four published reports price a B2B lead. They name four different scopes, publish four different statistics, and not one of them says how many campaigns it looked at. Here is how to read them, and how to build a number of your own.
11 min read
Gartner says buying teams work on their tasks without a consistent order or journey. Stage models say the opposite, and five different stage counts are in circulation. Here is what Gartner publishes, and what gets attributed to Gartner that Gartner never said.
10 min read
There is no B2B ranking algorithm, and nobody publishing on this claims otherwise. What changes is the size of the terms, the number of people deciding, and how long a page takes to pay. Two guides publish figures for that.
10 min read
Search volume is the wrong way to pick B2B terms, and the published guidance says so without putting a number in its place. Choosing between two keywords stays a matter of taste. Here is the arithmetic that settles it.
10 min read
The benefit figures quoted for alignment lose a step of attribution at every hop, and the anchor study behind them is from 2010. The part you can write down and check is smaller, and more useful.
10 min read
The published advice converges on the same nine steps. None of them says what content is being chosen instead of, or how much it has to produce to win that comparison. The answer moves by a factor of a hundred with your price.
30 pieces
The rest of what we have published, across the other formats.
9 min read
The phrase covers two products with two different buyers: software for a company running its own search work, and agency software with white-label reports and client seats. Here is the job each tool is sold for, what each one publishes, and the unit they leave to you.
7 min read
Pipeline velocity compresses four separate numbers into one, and two of them can move in opposite directions and cancel. A quarter where deals doubled in size and halved in win rate reports the same figure as the quarter before it.
7 min read
Most definitions of cost per lead describe a number you calculate. One of the results describes something else entirely: a pricing model you are billed on, where a contract defines what counts as a lead. The two share a name and almost nothing else.
7 min read
Look this term up and the same 3:1 target comes back page after page, and the two opened here name no origin for it. The origin exists, calls it a guideline, and its author later published a correction saying he should have told readers when the ratio is worth calculating at all.
7 min read
Published SaaS churn averages run from 3.22 percent to 14 percent a year, and the reason is not the maths. Enterprise software loses under 10 percent of customers a year and freemium products lose over 50, so the spread inside the category is wider than the spread between the people reporting it.
10 min read
Demand generation and lead generation are published as the same thing, as different things, as cause and effect, as two stages, and as distinct but complementary. Cognism holds two of those positions on two live pages. Here is how to pick a line, and what to fund while the argument runs.
6 min read
A sales qualified lead is one a rep has looked at and kept. Two steps can sit here where teams describe one. Here is the acceptance step, what reps check before they keep a lead, and how the definition changes the rate you can report.
9 min read
Every tool sold as a SaaS SEO tool is a general SEO tool. What changes for a SaaS company is the unit: your conversion is a trial or a demo, and no SEO suite counts those. Here is which tool does which job, what each publishes, and what you still have to build yourself.
6 min read
A product qualified lead is somebody who has already got value out of your product. This page covers the named triggers real companies use, why the stage is missing from HubSpot’s defaults, and when a PQL replaces a lead score instead of sitting beside it.
8 min read
Marketing decides an MQL and sales decides an SQL, and no published standard sets either bar. Here is what each term means in practice, why the conversion rates you can look up disagree by a factor of two, and how to set your own threshold.
6 min read
A marketing qualified lead is one your marketing team judges ready for sales. No outside body sets that bar. Here is what usually goes into the score, what the platforms leave to you, and how to write your own definition in one sentence.
10 min read
A CRO tool can be a testing platform, a session recorder, a survey tool or an analytics product. The two largest testing platforms publish no price and one recorder is free forever. Here is which class you need, what each publishes, and the traffic arithmetic to do before you buy.
8 min read
Six products are sold as answer engine optimization tools, as generative engine optimization tools, and as AI visibility trackers. Same products, three category names. Here is what they do, what they cost, and how to avoid buying two of them.
8 min read
GEO is the name of a method from a 2024 academic paper with a published benchmark. AEO is an industry coinage, named by Google only in a list of things you do not need to do. The work sold under both names is the same work, so buy on evidence and not on the label a vendor prefers.
9 min read
Most GEO tools tell you where your brand stands in AI answers, and one of them watches what AI agents do on your site instead. What you pay for is asking again, on a schedule, across more engines. The untested half is the advice that comes with it, and here is how to tell the two apart before you buy.
10 min read
These tools do four different jobs, and they get ranked as though they did one. Scoring a draft, planning a site, writing a draft, tracking AI answers. Here is which tool is sold for which, and what none of them will tell you.
10 min read
Dozens of tools claim to track whether AI answers name your brand. Three come up again and again: Profound, Peec AI and Semrush. Here is what each publishes about how it works, who each one is wrong for, and what none of them can do yet.
6 min read
A box Google generates about an entity it already recognises. You cannot create one, you can sometimes claim one, and Google states that not all of them are claimable.
6 min read
The phrase names two different jobs: preparing your pages for AI search engines, and using AI tools to do SEO work. Most people mean the first. Vendors often mean the second.
6 min read
AI visibility is whether an answer engine names your brand. Google attaches a privacy question to the same phrase in its own results, and inside the marketing sense the term still covers three different events.
9 min read
Both raise revenue through the same multiplication. How much traffic you already have decides whether you can prove a conversion gain, and that threshold is arithmetic, not opinion.
5 min read
AOV is revenue divided by orders. That is an arithmetic mean, and a mean can sit far from the order a store takes most often.
9 min read
Google publishes how its shopping ads work, down to the auction. OpenAI publishes almost nothing about how products get picked in ChatGPT. That gap is the most useful thing to know before you budget for either.
5 min read
AEO is the industry name for being cited in AI-generated answers. Three of the companies that build those answers publish access instructions, not optimization advice, and none of the three uses the term.
9 min read
Before the channels differ, the counting does. Google Ads counts conversions over 30 days by default and Meta counts a click for at most 7. A head to head return comparison starts from two different rules.
6 min read
Return on ad spend in plain language: the formula, why two tools report different numbers for the same spend, and the arithmetic that tells you what a good ROAS is for your store.
9 min read
The usual answer is that SEO is free and slow, PPC is fast and paid. In three of four ecommerce categories measured, the paid share of clicks roughly doubled in twelve months. Here is what that changes.
6 min read
The three metrics, the three published thresholds, and the three counting rules that decide whether a store passes. Plus what the numbers look like by ecommerce platform.
4 min read
Conversion rate optimization in plain language: the definition, the formula, and the one way the number reliably misleads store owners.
8 min read
Answer engine optimization and SEO are measured differently and fail differently. Here is the split, with the number that shows why ranking first is not the same as being cited.
These pages answer the questions that have published answers. The ones that do not are the reason we look at your store directly.
