Reference
Meta Ads, translated.
Plain-language definitions written for founders, not platform docs. Every term that actually moves ROAS, with no jargon left unexplained.
A
- Advantage+ Shopping Campaign (ASC)
- Meta's fully automated campaign type that hands audience selection, placement, and creative combination to the algorithm. ASC works best when your pixel has at least 50 conversions per week — below that, the algorithm is effectively guessing. The trade-off is real: less manual control, but lower CPAs when the signal is clean. Treat it as a complement to manual campaigns, not a replacement.What is ASC?
- Attribution window
- The look-back period Meta uses to credit a conversion to an ad. The default is 7-day click + 1-day view. A wider window inflates reported ROAS; a narrower one deflates it. If you changed your attribution window mid-flight, your before/after ROAS comparison is meaningless — you are comparing different measurement rulers. Always record the window alongside every ROAS figure you share.Conversion windows explained
- Audience overlap
- When the same user qualifies for multiple ad sets, Meta enters you into auction against yourself. Typical e-commerce accounts see 30–45% overlap between prospecting and retargeting audiences — meaning you are paying twice to reach the same shopper. The fix is structural: exclusions and campaign-level budget controls, not bid adjustments.
B
- Bid strategy & cost cap
- Your bid strategy tells Meta how aggressively to compete in auctions and at what price ceiling. Lowest-cost (no cap) maximises volume; cost cap holds your average CPA at a target but can throttle delivery when the auction is competitive. Choosing the wrong strategy for your margin is one of the fastest ways to overspend — know your break-even CPA before you set a cap.
- Broad targeting
- Running ads with no interest, demographic, or custom audience restriction — letting Meta's algorithm find buyers from scratch. Counterintuitively, broad often outperforms detailed targeting on mature pixels with strong conversion history, because Meta has more room to find high-intent signals. It is not laziness; it is trusting the model. But it requires a clean, high-volume pixel signal to work.
C
- CAPI (Conversions API)
- A server-to-server signal pipeline that sends conversion events directly from your backend to Meta, bypassing browser-based restrictions. Accounts with full CAPI implementation typically report approximately 19% more conversions than pixel-only setups. That gap is not fake — it represents real purchases the pixel missed because of iOS privacy changes, ad blockers, and cookie restrictions. Without CAPI, Meta's algorithm optimises on an incomplete dataset.What is CAPI?
- CBO (Campaign Budget Optimization)
- CBO moves budget authority from the ad-set level to the campaign level, letting Meta shift spend in real time toward whichever ad set is winning the auction. It reduces manual budget shuffling but removes your ability to guarantee spend on specific audience segments. If you have one anchor audience that must always receive budget — a high-LTV retargeting pool, for example — ABO (ad-set budget) gives you that control; CBO does not.
- Click-through rate (CTR)
- Clicks divided by impressions, expressed as a percentage. Link CTR (clicks to your site) is more diagnostic than all-click CTR (which includes reactions and video plays). A low link CTR signals the creative or copy is not compelling enough to pull people off-feed. A high link CTR with low conversion rate points to a landing-page mismatch. Neither number alone tells you where the problem is.
- CPA (Cost per acquisition)
- Total spend divided by the number of purchases (or other conversion events) attributed to your ads. CPA is the budget-health number — it tells you what you paid for each customer. But CPA without context is dangerous: a CPA that looks healthy against your topline price can be catastrophic if your gross margin is thin. Always pair CPA with gross margin to get to break-even CPA, the only threshold that matters.
- CPM (Cost per thousand impressions)
- What you pay per 1,000 impressions. CPM is Meta's currency — everything you spend flows through this rate. High CPM means the auction is competitive for your audience. Your effective CPM rises when your ad relevance score is low (Meta charges more to show poor ads) and falls when it is high. CPM is largely outside your control; CTR and conversion rate are not — so focus there.
- Creative fatigue
- The decay in ad performance that occurs when the same audience has seen the same creative too many times. The typical window before a top-performing creative starts to decline is approximately 14 days. Frequency is the leading indicator — once your frequency climbs above 3–4 on a cold audience, you are in fatigue territory. The fix is refreshing the asset, not the bid.
D
- Dayparting
- Scheduling your ads to run only during specific hours or days. Meta's delivery system already optimises for times when conversion probability is high within a given budget — manual dayparting often overrides that optimisation for the worse. The legitimate use case is businesses with hard operational constraints (e.g., a call centre that closes at 10 pm). For most e-commerce brands, it is an unnecessary override.
E
- Event deduplication
- When both your browser pixel and CAPI fire for the same purchase event, Meta can count it twice — inflating reported conversions and teaching the algorithm on ghost data. Deduplication is the mechanism that prevents double-counting, typically using a shared event_id. Get this wrong and your reported ROAS can be meaningfully higher than your actual ROAS — which means you scale campaigns that are losing money.Event deduplication explained
F
- Frequency & frequency capping
- Frequency is the average number of times a unique user has seen your ad. It is a function of your budget, audience size, and campaign duration — not a dial you turn directly. High frequency on cold audiences signals wasted spend; high frequency on warm retargeting audiences can be intentional if your conversion cycle is long. Frequency capping limits how often a single person sees an ad in a given period. Use it on prospecting to prevent fatigue before a user converts.
H
- Hook rate
- The percentage of video viewers who watch past the first three seconds. Hook rate measures whether your opening frame is strong enough to stop the scroll. A weak hook usually means the first frame is too generic, too slow, or too similar to what the user was already scrolling past — judge yours against your own account's best performers rather than a universal threshold. Fixing the hook is almost always higher-leverage than editing the middle or end of a video ad.
L
- Learning phase
- A period at the start of a new ad set (or after a significant edit) during which Meta's algorithm explores the audience to find who converts. The learning phase typically exits after approximately 50 conversions per ad set over seven days. While in learning, performance is volatile and CPAs are often higher than they will be at steady state. Avoid making structural edits during learning — it resets the clock.
- Lookalike audience
- A Meta-generated audience that shares statistical characteristics with a seed list you provide (e.g., your purchasers, email list, or pixel events). A 1% lookalike is the closest match and smallest; a 10% is broader and cheaper per CPM but less qualified. The quality of a lookalike is entirely determined by the quality of the seed — a purchaser-based seed outperforms a website-visitor seed every time.Custom vs lookalike audiences
M
- MER (Marketing Efficiency Ratio)
- Total revenue divided by total marketing spend — blended across all channels, not just Meta-attributed. MER is the only metric that cannot be gamed by attribution windows or pixel gaps. If your Meta-attributed ROAS is 6x but your MER is 1.8x, something is wrong in your attribution, not your business. Use MER as the sanity check against which every channel-level ROAS number is held accountable.
P
- Pixel
- A snippet of JavaScript placed on your website that fires events back to Meta when visitors take actions — view a product, add to cart, initiate checkout, purchase. The pixel is the foundation of attribution, optimisation, and audience creation. A misconfigured pixel (wrong events, missing parameters, double-firing) corrupts every downstream number. Audit your pixel before trusting any campaign metric.
- Prospecting vs retargeting
- Prospecting reaches people who have never interacted with your brand — cold audiences built from lookalikes, interests, or broad targeting. Retargeting reaches people who have already visited your site, added to cart, or engaged with your content. They require different creative (brand introduction vs. purchase urgency), different bidding (higher CPA tolerance vs. tighter cost caps), and different success metrics. Running them in the same campaign muddies both.
R
- Retargeting window
- The look-back period used to build retargeting audiences — typically 7, 14, 30, 60, or 180 days of site visits, add-to-carts, or other events. Shorter windows (7–14 days) reach the warmest, most purchase-ready users but are smaller and more expensive. Longer windows (60–180 days) reach larger audiences but with lower intent. Match the window to your purchase consideration cycle, not your budget pressure.
- ROAS (and break-even ROAS)
- Return on Ad Spend — revenue attributed to your ads divided by ad spend. The 2026 median for e-commerce Meta campaigns is approximately 2.8x. But headline ROAS is meaningless without break-even ROAS, which is simply 1 ÷ your gross margin. If your gross margin is 40%, your break-even ROAS is 2.5x — anything below that and every sale loses money regardless of what the dashboard shows. Always report ROAS alongside your break-even threshold.
S
- Spend reallocation
- Moving budget from underperforming campaigns, ad sets, or creatives toward proven winners — without increasing total spend. Reallocation is often more impactful than scaling because it removes the drag of losing spend while amplifying what already works. 20–40% of the average e-commerce account's Meta budget leaks to audience overlap, creative fatigue, and broken tracking. Reallocating that recaptured spend is free ROAS growth.
T
- Thumb-stop ratio
- The percentage of people who paused on your ad instead of scrolling past — typically measured as 3-second video views divided by impressions. A strong thumb-stop ratio means your opening frame is winning the scroll interruption battle; compare creatives against each other within your account rather than against a universal number. It does not measure whether the viewer then converts — hook and hold are separate problems.
W
- Wasted ad spend
- Budget that generates impressions, clicks, or even attributed conversions without delivering profitable outcomes — typically caused by audience overlap, creative fatigue, broken tracking, structural campaign errors, or bidding against yourself. Illustrative planning ranges suggest 20–40% of the average e-commerce account's Meta budget falls into this category. The number is account-specific; the only way to know yours is to audit it.
- Whitelisting / dark posting
- Running paid ads through a creator's or influencer's Facebook or Instagram handle rather than your brand page. The ad appears to come from the creator (social proof intact) but is targeted and paid for by you. Dark posts are unpublished page posts — they run as ads but do not appear on your brand's public timeline. Both techniques are useful for UGC-style creative that performs better appearing organic than branded.
Benchmark figures cited (20–40% wasted spend; 2.8x median ROAS; ~14-day fatigue window; ~19% CAPI uplift; ~50 conversions / 7 days for learning phase exit) are illustrative planning ranges — Bach.ai surfaces your account's real numbers once connected.
From glossary to action
Stop translating. Bach reads your account and tells you what to fix, in money.
Every term in this glossary maps to a check Bach runs automatically — creative fatigue, audience overlap, CAPI gaps, learning-phase thrash, wasted spend. The result is a ranked list of fixes with estimated revenue impact, not another vocabulary lesson.