The "what to ask" glossary — 20 terms from algorithm, ads, SEO and analytics

The 20 terms

Ordered organic → paid → format → model → SEO. Revised: 2026-09.

1 · Reach Score Organic 🔗

What it is: A 0-to-100 score that estimates how much the platform will distribute a post before you publish it, computed from measurable factors: hook strength, expected dwell, save likelihood, CTA and readability on a small screen.
Why it matters in 2026: It's the only score that can change your draft, not just flatter the report. If the score only gets calculated after publishing, it's a scoreboard, not an optimizer.
The question you should ask: "Show me the exact formula and weights, and which posts it got wrong over the last 20 predictions." Anyone who won't publish their rubric is selling a slogan.

2 · Qualified session Analytics 🔗

What it is: A visit that did something: used the calculator, audited a piece of text, requested the PDF. Not "time on page" or "60% scroll" — a named action, counted as an event.
Why it matters in 2026: With AI summarizing your content before the click, traffic no longer means interest: action does. Optimizing for qualified sessions is the only honest way to measure content that converts.
The question you should ask: "Which actions count as 'qualified' in your reports?" If the answer includes "more than 2 minutes on site", they're measuring patience, not intent.

3 · Dwell time Organic 🔗

What it is: How long your content stays visible (or open) before the person moves on. On LinkedIn and X it's the central retention signal; on Google, the time between leaving the results and coming back.
Why it matters in 2026: It determines how many more people see the post afterwards: high retention → more distribution into the next audience layers. It's the reason one dense, good paragraph beats ten obvious sentences.
The question you should ask: "What specific recommendation for MY format raises dwell, and can you measure it before/after?" "Short sentences and emojis" is not a recommendation, it's a meme.

4 · Save rate Organic 🔗

What it is: Saves ÷ impressions. The "new like": a save says "I'm going to need this later", which is the closest thing to purchase intent a feed ever records.
Why it matters in 2026: Platforms keep weighting saves and shares above likes for sustained distribution. A heavily saved post keeps earning reach for weeks; a heavily liked one dies in 48 h.
The question you should ask: "If I raise my save rate 50%, what changes in my reach?" Nobody honest can give you an exact number — anyone who does, without caveats, is making it up.

5 · Anti-link penalty Organic 🔗

What it is: The reduced distribution a feed applies to posts with external links in the body (and, by some measurements, in the first comment too). It's the silent B2B tax: your whole strategy pointing to your site, and the channel anchoring it down.
Why it matters in 2026: It's the biggest hidden cost in content marketing: you won't find it in any dashboard. The pattern that works: no links inside the post, the link goes in the first comment (or the bio, or via DM), plus a save CTA the algorithm actually rewards.
The question you should ask: "Show me the before/after of the same post with and without a link: numbers, dates, segmentation." An A/B test without dates and sample size is an anecdote with formatting.

6 · Truncation Format 🔗

What it is: The cut the platform applies to your text with "see more": ≈140 visible characters on LinkedIn and ≈280 on X, depending on device and app. Whatever doesn't fit there doesn't exist for 80% of the audience.
Why it matters in 2026: The hook wins or dies in the visible zone. A great post with a mediocre truncation is a great post nobody saw: people don't tap "see more" out of politeness.
The question you should ask: "Can you show me the actual preview on a mid-range Android and on the desktop app?" If they only check their laptop feed, they're not looking where your audience lives.

7 · Early velocity Organic 🔗

What it is: Interactions in the first 60–90 minutes after publishing. The filter the algorithm uses to decide whether your post moves past your immediate network to colder audiences.
Why it matters in 2026: It explains why publishing "at the right time" weighs more on LinkedIn (professional audience with narrow peaks) than on Instagram (long consumption tail). And why replying to comments in the first hour beats scheduling three weeks out.
The question you should ask: "Are you scheduling to maximize early velocity, or to make the content calendar look busy?"

8 · Hook rate Video 🔗

What it is: The percentage of viewers who get past the first 3 seconds out of everyone who started the video. Common rule of thumb (not physics): below ~30% the hook fails; above ~50% it's strong.
Why it matters in 2026: The retention curve — not "% watched" — is what algorithms look at to keep distributing. A video holding 60% at second 5 beats a "complete" 15-s one that drops to 20%.
The question you should ask: "At which exact second do they leave? Show me the retention curve, not the average." Averages without curves = cosmetic reporting.

9 · CTR Ads 🔗

What it is: Clicks ÷ impressions. The honest reading of CTR isn't absolute, it's against the median of its own segment: the same number can be excellent in LinkedIn B2B and mediocre in e-commerce retargeting.
Why it matters in 2026: It's the cheap metric that separates "wrong message" from "wrong audience": low CTR with high cost per lead → creative; high CTR with bad leads → targeting or the form.
The question you should ask: "What segment median are you measuring this CTR against, and what bid range is your auction in?" A CTR without a reference is a decorative number.

10 · Break-even ROAS Ads 🔗

What it is: The minimum ROAS (attributed revenue ÷ ad spend) to not lose money: 1 ÷ gross margin. At a 60% margin your break-even is 1.67 — below that you're losing money even while "making sales".
Why it matters in 2026: "Raising ROAS" without knowing your break-even is training without a scale: it tells you to try harder, not where. The break-even takes 30 seconds to compute with YOUR margin and it conditions every budget.
The question you should ask: "What's my break-even ROAS, and what ROAS are you delivering right now against that number?" If the answer is "the higher the better", switch agencies without negotiating.

11 · Creative fatigue Ads 🔗

What it is: The predictable decay of an ad: frequency climbs (the same people see it again and again) and CTR falls. Accounts don't die from bad targeting: they die from old creative.
Why it matters in 2026: With narrow B2B audiences (2,000–20,000 people is normal), fatigue arrives in weeks, not months. RankStudio Ads measures it per creative combining CTR (30%), CPM (20%), CVR (30%) and CPA (20%) — frequency is shown as context but doesn't score. Whoever ships 3 fresh variants a month and rotates them with criteria has a structural edge over whoever "optimizes bids".
The question you should ask: "At what frequency and after how many weeks do you rotate a creative, and are the new variants ready BEFORE they're needed?"

12 · PDF carousel 1080×1350 Format 🔗

What it is: The document LinkedIn (and X, to a lesser extent) renders as a swipeable image gallery. The vertical 4:5 ratio (1080×1350) takes up the most real pixels on the phone screen of the person scrolling.
Why it matters in 2026: Still the best organic-reach-per-cost format for B2B: every swipe adds dwell, and the last slide is a save asset ("bookmark this") when it's designed well.
The question you should ask: "Does this carousel have an actionable last slide (summary + reason to save), or is it one closing sentence split across three slides to look long?"

13 · Transcreation Content 🔗

What it is: Not translation: rewriting a piece for the language, culture and physics of another platform. A LinkedIn post is not an X thread even when it says the same thing, and Madrid Spanish is not Mexico City Spanish.
Why it matters in 2026: Small teams publish into 3 markets with one person. Transcreation is the lever that multiplies good content without multiplying its cost; literal translation is what breaks it in the last 20%.
The question you should ask: "Show me a before/after: what changed besides the words — structure, examples, CTA, format." If only the words changed, that's a translator in a suit.

14 · BYOK (Bring Your Own Key) Model 🔗

What it is: A model where you bring your AI provider's API key to the tool: it doesn't resell you tokens or wall you behind monthly credits. The real cost (tokens) is yours and visible.
Why it matters in 2026: Two reasons: cost — a heavy user pays provider price, not "Pro plan with 500 rewrites" — and privacy — your text can skip the vendor's servers entirely, if the product is built right.
The question you should ask: "If I plug in MY key, what happens to my text: does it stay in my account, do you use it for training, do you store it on your servers?" A vague answer means "we keep your text", under another name.

15 · AI SEO Model 🔗

What it is: Using AI agents for the entire SEO cycle — picking topics, writing, publishing, requesting links — not just for writing. The label warns you: if whoever sells you "AI SEO" only generates text with a model, they're selling you a prompt.
Why it matters in 2026: The model is the same for you and your competitor, so writing stopped being the edge. It moved to process discipline — measure before writing, publish only as fast as you get indexed — and to the one thing the model doesn't have: your real experience.
The question you should ask: "Which part of your system can't my competitor replicate by hiring the same model?" If the answer is "nothing", the value isn't in the system.

16 · Long-tail keyword Organic 🔗

What it is: The more specific variant of a search: whoever searches "post audit" also searches "how to audit a post before publishing". Fewer people search it and fewer competitors fight over it, and the intent is clearer.
Why it matters in 2026: The safest tactic is expanding from what the search engine already shows you: if your site gets impressions for a topic, its variants cost you less than starting from zero. The long tail is the first place a small site wins.
The question you should ask: "Did these keywords come from MY Search Console, or from a list you hand every client?"

17 · Indexing Organic 🔗

What it is: The search engine's decision to include your page in its index. Without indexing, your page doesn't exist for Google, no matter how published it is.
Why it matters in 2026: It sets the real pace of SEO: publishing faster than you get indexed is throwing content into a drawer. The healthy rule — don't commission the next batch until the previous one is in the index — is the difference between growing and hoarding.
The question you should ask: "How do you verify that what's published is indexed before producing more?"

18 · Backlink Organic 🔗

What it is: A link from another site to yours: a public vote from someone with an audience. You don't need hundreds — a well-chosen twenty beats mediocre hundreds — because the expensive part isn't the link, it's the judgment to know who to ask.
Why it matters in 2026: You get them the way you always have: asking, one by one, from related sites. That's why the cheap link package is the classic red flag: whatever sells by the dozen is the first thing the search engine learns to discount.
The question you should ask: "How is each link obtained: one-by-one outreach or a package?"

The 2 counterweights — these are yours to use

These aren't terms they'll explain to you. They're the two question-shaped concepts that change the meeting.

19 · Vanity metric Counterweight 🔗

What it is: Any number that goes up in the report and can't be traced to money: raw impressions, followers, likes. They're not useless — they're the thermometer; the problem is treating the thermometer as the disease.
Why it matters in 2026: With feeds measuring in real time, it's trivially easy to show reach growth without moving a single opportunity. Every agency and tool that "optimizes reach" hides here if there's no bridge to qualified actions.
How to use it in the meeting: In front of any pretty number: "And what does this translate into — qualified sessions, leads, pipeline?" The question, not the number, is the counterweight.

20 · "100% SEO score" Counterweight 🔗

What it is: The flavor-of-the-month tool's perfect score. A 100 doesn't mean "guaranteed ranking": it means "you passed every check on a list". Lists run out; ranking isn't on that list.
Why it matters in 2026: With generative AI citing (or not), the old green-checklist game is dead: what ranks today is getting cited and getting the visitor to do something — qualified sessions again. A perfect score doesn't book a meeting; specific content does.
How to use it in the meeting: "If the score were predictive, you'd pay me when you drop below 98." Whoever sells scores instead of results is doing weather commentary with an umbrella up.

How to use this without becoming what we criticize: the definitions are operational, not revealed truth — the thresholds (3 s, ~140 characters, 60–90 min) are practical rules that shift with every platform update, and there are no client logos or fake counters here precisely because we don't have them. The reviewable part is the third line of every term: the question. If your vendor can't survive all 20, the problem isn't the glossary.