[{"data":1,"prerenderedAt":634},["ShallowReactive",2],{"guide-broad-vs-detailed-targeting":3},{"id":4,"title":5,"answer":6,"authorId":7,"body":8,"category":496,"ctaVariant":497,"dataset":496,"description":498,"examples":499,"extension":512,"faqs":513,"heroImage":538,"intro":539,"meta":540,"navigation":541,"path":542,"publishedAt":543,"seo":544,"sources":545,"stats":586,"stem":632,"updatedAt":543,"__hash__":633},"blog\u002Fblog\u002Fbroad-vs-detailed-targeting.md","Broad vs Detailed Targeting on Meta (2026)","On Meta in 2026, broad targeting plus strong, varied creative usually beats heavy interest segmentation. Detailed targeting is now a suggestion, not a fence: Meta's Help Center treats your inputs as 'audience suggestions by default' and serves your ad beyond them when its AI predicts better results, with only location and minimum age remaining hard constraints. Meta's Andromeda retrieval engine reads creative signals to match ads to people, so the lever has moved from interests to the creative itself. Narrow still wins in three cases: location-bound local businesses, regulated Special Ad Categories, and small list-based custom or lookalike audiences.","xanny-lee",{"type":9,"value":10,"toc":483},"minimark",[11,16,20,23,26,30,33,47,50,53,56,59,62,69,73,76,79,87,91,94,97,100,103,106,174,178,181,184,245,253,261,265,268,271,274,277,280,341,344,347,351,354,357,360,420,423,427,430,433,445,448,454,458,461,464,467],[12,13,15],"h2",{"id":14},"what-broad-and-detailed-actually-mean-in-2026","What \"broad\" and \"detailed\" actually mean in 2026",[17,18,19],"p",{},"Start with the words, because the platform changed them under you. Detailed targeting is the box where you add interests, behaviors, and demographics. Broad targeting means you leave that box mostly empty and let Meta's audience system find buyers across Facebook, Instagram, Reels, and Stories. For years the choice felt like a dial between control and reach.",[17,21,22],{},"In 2026 that dial is mostly cosmetic on the detailed side. Meta's own Help Center now describes detailed-targeting inputs as audience suggestions by default: the system shows your ad to people who match your suggestion, and to others when it predicts better performance. Independent analysis from Jon Loomer reaches the same conclusion, that only location and minimum age remain hard constraints. Everything else you type into the targeting box is a hint the machine starts from and walks past.",[17,24,25],{},"So the honest framing is not broad versus detailed as two equal options. It is a default broad system that accepts optional suggestions, where heavy suggestions buy you less precision than they used to. The question stops being how tightly can I narrow this and becomes what signal am I giving the system to find the right person.",[12,27,29],{"id":28},"why-detailed-targeting-became-a-suggestion","Why detailed targeting became a suggestion",[17,31,32],{},"Two changes drove this. The first is structural: Advantage+ Audience is the default audience experience for most objectives now, and within it your detailed inputs are treated as suggestions rather than filters. Even when you switch to original audiences on certain performance goals, the suggestion behavior can persist. Meta also consolidated many granular interest categories starting in mid-2025, folding niche segments into broader ones, which thinned the long tail of fine targeting that power users relied on.",[17,34,35,36,41,42,46],{},"It is worth separating two things Meta named almost identically, because mixing them up muddies every targeting decision. Advantage+ Audience is the audience setting on the ad set: you hand it controls it must obey (location, minimum age, language, and the custom audiences you exclude) plus optional suggestions it is free to expand past. Advantage+ Sales, formerly Advantage+ Shopping, is a different animal: a whole campaign type that automates audience, placement, and budget end to end. One is a box on the ad set; the other is the entire campaign. You can run a manual campaign that uses Advantage+ Audience, or an Advantage+ Sales campaign that relies on it underneath. Our walkthroughs of ",[37,38,40],"a",{"href":39},"\u002Fblog\u002Ffacebook-advantage-plus-audience","Advantage+ Audience"," and ",[37,43,45],{"href":44},"\u002Fblog\u002Fadvantage-plus-sales-campaigns","Advantage+ Sales campaigns"," cover each in full; for this decision, what matters is that both treat your interests as suggestions, not fences.",[17,48,49],{},"There is a second-order effect worth understanding in how you stack interests. Adding them with OR logic (someone into running, yoga, or cycling) widens the pool, but layering conditions with AND logic (into running, and a parent, and a frequent traveler) multiplies the constraints and shrinks it fast. Power users reached for that AND-layering to feel surgical, yet every layer hands the optimizer a smaller room to search. In 2026 the narrowing buys almost nothing the creative would not have delivered, and it costs you the volume the model needs to learn.",[17,51,52],{},"The second change is the engine underneath. In December 2024 Meta announced Andromeda, an AI retrieval system that scans tens of millions of active ads and matches them to users based on signals in the creative itself: the visuals, the hook, the format. Meta reported that Andromeda lifted recall by 6% and ads quality by 8% on selected segments, and enabled a 10,000x increase in model capacity at the retrieval stage where millions of ads are narrowed to candidates. Those are Meta's own figures, so treat them as directional. The direction, though, is clear and corroborated by neutral trade coverage in Social Media Examiner: the system reads your creative to decide who sees it.",[17,54,55],{},"When a machine that good is choosing the audience from creative signals, a manual interest filter mostly gets in its way. That is the mechanical reason detailed targeting faded from fence to suggestion.",[17,57,58],{},"It helps to picture the old workflow versus the new one. In the old workflow you did the matching yourself: you guessed which interests correlated with buyers, stacked them, and Meta delivered inside the lines you drew. Your guesses were the ceiling on performance. In the new workflow Meta does the matching against a vastly larger feature space than any interest list could express, reading the creative against real-time behavior at the moment of the auction. Your job moved from drawing the lines to feeding the model something worth matching. A tight interest stack does not make that model smarter; it just hands it a smaller room to search and a worse chance of finding the person who would have converted.",[17,60,61],{},"The practical consequence is that the two classic broad-targeting fears have largely inverted. The fear used to be that broad would burn budget on irrelevant impressions. In 2026 the bigger risk is that a narrow audience starves the system of the volume it needs to learn, so it never exits the learning phase cleanly and your cost per result stays high. Broad gives the optimizer room; narrow, in most accounts, just gives it a handicap.",[17,63,64],{},[65,66],"img",{"alt":67,"src":68},"Diagram contrasting a rigid interest filter funnel against a broad audience where an AI retrieval engine matches creative signals to people","\u002Fimages\u002Fblog\u002Fbroad-vs-detailed-targeting-fence-vs-suggestion.webp",[12,70,72],{"id":71},"the-signal-loss-that-pushed-everyone-broad","The signal loss that pushed everyone broad",[17,74,75],{},"There is a deeper reason the platform leans on broad now, and it is not only better AI. Starting in April 2021, Apple's App Tracking Transparency feature shipped with iOS 14.5 and forced every app, Meta included, to ask users for permission before tracking them across other apps and websites. Most people said no. A large share of the third-party signal Meta used to follow a shopper from an ad to a purchase went dark almost overnight, and the slower erosion of third-party browser cookies pulled in the same direction. The raw material that made hyper-precise manual targeting feel possible got scarce.",[17,77,78],{},"Machine learning answers scarce signal the same way every time: with more volume and cleaner first-party data. A narrow audience hands the model a thin trickle of conversions to learn from, exactly when the model is hungriest for examples. A broad audience gives it room to find the patterns that survived the signal loss. So the honest version of \"broad won\" is less that broad got smarter and more that narrow got starved. The privacy change moved the center of gravity from who you name in the targeting box to how much signal and reach you can feed the system.",[17,80,81,82,86],{},"The fix Meta points advertisers to is the Conversions API. Where the browser Pixel loses events to ad blockers, connection errors, and iOS limits, the Conversions API sends the same events server to server, from your backend straight to Meta, so the conversion signal keeps flowing when the browser drops it. Meta's own Help Center says Conversions API data is less affected by those losses and, paired with the Pixel, helps the delivery system lower your cost per action. The order of operations follows from that: get your first-party signal clean and server-side first, then go broad, because broad targeting without a healthy conversion feed is just a big audience the system cannot learn from. Our ",[37,83,85],{"href":84},"\u002Fblog\u002Ffacebook-conversions-api-setup","Conversions API setup guide"," walks the wiring; the point here is that the audience setting and the signal feed are two halves of one decision.",[12,88,90],{"id":89},"broad-plus-strong-creative-beats-heavy-segmentation","Broad plus strong creative beats heavy segmentation",[17,92,93],{},"Here is the trade most accounts should make. Stop spending your planning hours stacking interest layers and spend them building distinct creative, because creative is what the retrieval engine matches on. Meta points to a concrete payoff from leaning on its creative tooling: it reports that campaigns using image generation within Advantage+ creative saw an 11% higher click-through rate and a 7.6% higher conversion rate than campaigns not using those features. Again, Meta's own numbers, so directional, but they line up with how Andromeda works.",[17,95,96],{},"The scale of the shift is visible in adoption. In a separate report on Meta's AI ad elements, Social Media Today, citing Meta, said more than a million advertisers were using Meta's generative-AI ad tools, producing over 15 million ads in a single month. That same Social Media Today write-up attributes to Meta a 22% ROAS lift for advertisers who turned on Advantage+ creative and a 7% conversion lift from its generative-AI image tools. Vendor figures, all of them, so directional, but the volume tells you the competitive baseline: if your rivals are shipping varied creative at that pace, a tightly segmented campaign with two tired images is not competing on the right axis.",[17,98,99],{},"What \"diverse creative\" means in practice is different angles, not different colors. Three to five concepts that attack the product from genuinely different motivations beat ten variants of one idea. A brand like Shakura can run a before-and-after pigmentation angle, a dullness angle, and a glow angle under one broad audience, and let Andromeda route each to the person it fits. A brand like Fitness Achievers can run strength, fat-loss, and community hooks the same way. The audience is broad; the creative does the segmenting.",[17,101,102],{},"Think of each distinct concept as a probe into a different part of your potential audience. The pigmentation angle finds the person worried about dark spots; the glow angle finds the person who just wants to look brighter for an event; the dullness angle catches the one in between. You did not have to name those segments or guess their interests. You expressed them as creative, and the retrieval engine did the sorting in milliseconds at auction time. That is the whole argument for moving effort from the targeting box to the creative brief: the creative is now the most expressive targeting input you have, and it is the only one Meta will not override.",[17,104,105],{},"This also reframes how you read a losing ad. Under heavy detailed targeting, a flat result was ambiguous, since you could never tell whether the interest, the offer, or the image was at fault. Under broad targeting the variables collapse: the audience is constant across your concepts, so when one angle wins and another dies, the creative is the difference. That clarity is worth more than any interest report, because it tells you exactly what to make more of.",[107,108,109,128],"table",{},[110,111,112],"thead",{},[113,114,115,119,122,125],"tr",{},[116,117,118],"th",{},"Approach",[116,120,121],{},"What you control",[116,123,124],{},"Where the work goes",[116,126,127],{},"When it wins",[129,130,131,146,160],"tbody",{},[113,132,133,137,140,143],{},[134,135,136],"td",{},"Broad plus diverse creative",[134,138,139],{},"Location, min age, objective, creative set",[134,141,142],{},"Building distinct angles and hooks",[134,144,145],{},"Healthy signal, several concepts, scaling",[113,147,148,151,154,157],{},[134,149,150],{},"Heavy detailed targeting",[134,152,153],{},"Interests, behaviors, demographics (as suggestions)",[134,155,156],{},"Stacking and excluding segments",[134,158,159],{},"Rarely the best lever in 2026",[113,161,162,165,168,171],{},[134,163,164],{},"Narrow constrained targeting",[134,166,167],{},"Hard fences (location, category rules, lists)",[134,169,170],{},"Defining the legitimate constraint",[134,172,173],{},"Local, regulated, or list-based cases",[12,175,177],{"id":176},"why-narrow-audiences-stall-in-the-learning-phase","Why narrow audiences stall in the learning phase",[17,179,180],{},"The clearest mechanical cost of going too narrow shows up in the learning phase. When you launch or significantly edit an ad set, Meta's delivery system explores who responds before it settles into efficient delivery. It exits learning once the ad set gathers roughly 50 optimization events (purchases, leads, or whatever you optimized for) within about seven days, measured at the ad set level on a rolling window. Clear that volume and delivery stabilizes and cost per result usually drops. Miss it and the ad set sits in a state Meta labels \"Learning Limited,\" where it is unlikely to reach 50 events in a week and delivery never fully optimizes.",[17,182,183],{},"A narrow audience walks straight into that wall, and a small budget makes it worse, so run the arithmetic on your own numbers before you launch. Say a purchase costs you about $20 and you set a $40 daily budget. That is two purchases a day, roughly 14 a week, well short of 50, so the ad set is structurally Learning Limited no matter how good the creative is. You have three honest fixes. Raise the budget so the math can clear 50 (here you would need north of $140 a day). Optimize for a more frequent event higher in the funnel, like Add to Cart or Initiate Checkout, which fires far more often than Purchase and can feed the model enough volume to learn. Or consolidate several thin ad sets into one broader audience so the events pool instead of scattering. Notice that two of the three fixes mean going broader, not narrower.",[107,185,186,199],{},[110,187,188],{},[113,189,190,193,196],{},[116,191,192],{},"Lever",[116,194,195],{},"What it controls",[116,197,198],{},"Healthy state",[129,200,201,212,223,234],{},[113,202,203,206,209],{},[134,204,205],{},"Optimization events per week",[134,207,208],{},"Whether the ad set can exit learning",[134,210,211],{},"Around 50 or more within 7 days",[113,213,214,217,220],{},[134,215,216],{},"Audience size",[134,218,219],{},"The pool the events can come from",[134,221,222],{},"Broad enough to supply the volume",[113,224,225,228,231],{},[134,226,227],{},"Optimization event",[134,229,230],{},"How often the event fires",[134,232,233],{},"Frequent enough to reach about 50 per week",[113,235,236,239,242],{},[134,237,238],{},"Significant edits",[134,240,241],{},"Whether learning resets",[134,243,244],{},"Batched, then left alone to stabilize",[17,246,247,248,252],{},"One more trap: editing a live ad set restarts learning. Meta counts a change to the audience, the optimization event, or the creative as a significant edit, and a large enough budget or bid swing can count too, as can pausing the ad set for more than seven days. So a habit of nudging a narrow audience daily keeps resetting the very phase it is struggling to exit. Batch your changes, then leave it alone. Our ",[37,249,251],{"href":250},"\u002Fblog\u002Ffacebook-ad-learning-phase","learning phase guide"," goes deeper on what resets it and how to scale without tripping it.",[17,254,255,256,260],{},"Narrow also gets expensive for a second reason that compounds the first: saturation. A small audience runs out of fresh people fast, so the same users see the ad again and again. Frequency climbs, CPM tends to rise as the auction works harder to reach a shrinking pool of unseen people, and creative fatigue sets in sooner. You end up paying more to show a tired ad to people who have already seen it, which is the opposite of what narrowing was supposed to buy you. A broad audience pushes that saturation point further out by giving delivery more unseen people to reach. For how often is too often, see our guide on ",[37,257,259],{"href":258},"\u002Fblog\u002Fgood-facebook-ad-frequency","healthy ad frequency",".",[12,262,264],{"id":263},"when-narrow-targeting-still-makes-sense","When narrow targeting still makes sense",[17,266,267],{},"Broad is the default answer, not the only answer. A handful of cases keep the fence standing, and they share one trait: the constraint is a fact about who can buy, not a guess about who might want to.",[17,269,270],{},"Location-bound businesses are the clearest. A clinic, a salon, a single-city service cannot serve someone three states away, so the geographic radius is a real constraint, not a preference. A brand like UR Klinik runs instant-form lead ads where location stays a hard fence and the creative does the rest of the targeting inside that radius. Broad-within-a-radius is still broad; you are just honoring the one constraint that matters.",[17,272,273],{},"Regulated verticals are the second. If your ad promotes credit or a financial product, employment, housing, or social, electoral, or political issues, Meta requires you to flag it as a Special Ad Category, and the platform then strips most of your targeting on purpose to prevent discrimination. Detailed-targeting interests, audience exclusions, and lookalikes all become unavailable, age opens to 18-65+ across all genders, and in the US your location targeting is forced to a minimum 15-mile radius with no ZIP-code precision. Narrowing here is not a strategy you chose, it is a compliance regime you operate inside, so the move is to honor the rules and let the creative carry the work the targeting no longer can.",[17,275,276],{},"The third case is small list-based audiences: a high-value CRM segment or a tight lookalike you want to seed precisely, especially on a cold account with no conversion history. Feed that list to Meta as a custom audience, let Advantage+ expand from it, and you get the precision of a known seed plus the reach of the broad system.",[17,278,279],{},"The fourth case is the one the broad-always crowd tends to skip: a genuinely small market. If your total addressable audience is tiny, a specialized B2B tool sold to a few thousand companies, a luxury product with a narrow buyer pool, a hyper-specific hobby niche, then broad has little to expand into, and an interest or a tight custom audience can be the only way to point delivery at the handful of people who could ever buy. The tell is the real size of the market, not your nerves: if only a small population could plausibly purchase, seeding the system with that signal beats asking it to find needles across the whole country. Even then, keep the creative diverse, because a small audience saturates fast and fresh angles are what slow the fatigue.",[107,281,282,295],{},[110,283,284],{},[113,285,286,289,292],{},[116,287,288],{},"Narrow case",[116,290,291],{},"Why it is a genuine fence",[116,293,294],{},"How to honor it without over-narrowing",[129,296,297,308,319,330],{},[113,298,299,302,305],{},[134,300,301],{},"Location-bound business",[134,303,304],{},"You cannot serve a buyer outside the area",[134,306,307],{},"Set the radius, then go broad inside it",[113,309,310,313,316],{},[134,311,312],{},"Special Ad Category",[134,314,315],{},"Targeting is restricted by law",[134,317,318],{},"Flag the category, comply, lean on creative",[113,320,321,324,327],{},[134,322,323],{},"Small list to seed",[134,325,326],{},"A cold account needs a known starting signal",[134,328,329],{},"Feed a custom audience or tight lookalike, let it expand",[113,331,332,335,338],{},[134,333,334],{},"Genuinely small market",[134,336,337],{},"The total buyer pool is tiny",[134,339,340],{},"Seed the niche signal, keep creative varied to fight fatigue",[17,342,343],{},"Exclusions are worth a separate note, because they are narrowing for hygiene rather than for precision. You can no longer rely on detailed-targeting exclusions the way you once could, but custom-audience exclusions still bind, since they sit alongside the hard constraints rather than in the suggestion layer. Excluding recent purchasers from a prospecting campaign, or carving existing customers out of an acquisition push, is a legitimate narrowing the system honors, and it is the kind that still works.",[17,345,346],{},"Outside those cases, narrowing usually costs you. You shrink the pool the system can optimize over, raise the price of reaching it, and gain little relevance the creative would not have delivered anyway. The test to apply is simple: is this constraint a fact about who can become a customer (a radius, a legal category, a known list, a tiny market) or a guess about who might want the product? Facts deserve a fence. Guesses belong in the creative, where the engine can prove you right or wrong instead of taking your word for it.",[12,348,350],{"id":349},"the-cost-picture-in-neutral-numbers","The cost picture, in neutral numbers",[17,352,353],{},"Lead with figures that do not come from a vendor's marketing page. WordStream by LocaliQ's cross-industry Facebook benchmarks put the average cost-per-click at $0.77 on the traffic objective and $1.88 on lead generation in 2024. Average lead-gen conversion rate landed at 8.78%, cost-per-lead at $21.98 (down from $23.10 the prior year), and lead-gen CTR at 2.53% against 1.57% for traffic campaigns. Search Engine Land, reporting the same dataset, summed up the trend as clicks and conversions up while costs came down.",[17,355,356],{},"Read those as wide midpoints, not targets. Cost-per-lead swings enormously by vertical, offer, and geography, so a number two or three times the average can be perfectly healthy in a competitive niche. The point for this decision is that none of these benchmarks is driven by how many interest layers you stacked. They move with objective, creative strength, and the quality of your conversion signal. That is the tell: the audience setting is not where your cost per result is won or lost.",[17,358,359],{},"One caution on the cost claims you will see attributed to broad targeting and Advantage+. Vendor blogs frequently cite a figure that broad or AI targeting cuts cost-per-acquisition by some fixed percentage, often around a third. That number gets recycled widely but rarely traces to a dated, neutral source you can open and check, so do not plan around it. The defensible figures are the neutral WordStream benchmarks above for cost, and Meta's own directional ROAS and conversion lifts for the AI tooling. Anchor your expectations to those, run your own holdout test, and let your account's real cost per result settle the broad-versus-narrow question for your offer.",[107,361,362,375],{},[110,363,364],{},[113,365,366,369,372],{},[116,367,368],{},"Metric (all industries, 2024)",[116,370,371],{},"Traffic objective",[116,373,374],{},"Lead-gen objective",[129,376,377,388,399,410],{},[113,378,379,382,385],{},[134,380,381],{},"Cost per click",[134,383,384],{},"$0.77",[134,386,387],{},"$1.88",[113,389,390,393,396],{},[134,391,392],{},"Click-through rate",[134,394,395],{},"1.57%",[134,397,398],{},"2.53%",[113,400,401,404,407],{},[134,402,403],{},"Conversion rate",[134,405,406],{},"n\u002Fa",[134,408,409],{},"8.78%",[113,411,412,415,417],{},[134,413,414],{},"Cost per lead",[134,416,406],{},[134,418,419],{},"$21.98",[17,421,422],{},"Source: WordStream by LocaliQ, Facebook Ads Benchmarks 2024.",[12,424,426],{"id":425},"how-to-set-up-a-broad-campaign-without-losing-the-plot","How to set up a broad campaign without losing the plot",[17,428,429],{},"Broad is not \"leave everything blank and hope.\" It is a deliberate setup that gives the system clean signals. Keep your location and minimum age accurate, because those are the constraints Meta actually enforces. Pick the objective that matches the real outcome, since the objective shapes who the system hunts for far more than any interest does. Then put your effort into the creative set.",[17,431,432],{},"Build three to five concepts around different motivations, not different crops of the same photo. Vary the format too: a UGC testimonial, a static results frame, a short transformation reel each give Andromeda a different signal to match. A brand like Skinlycious can carry a broad audience on a single strong before-and-after testimonial because the creative itself is doing the audience-finding that detailed targeting used to do. If you are seeding a cold account, hand Meta a custom audience or a tight lookalike as a starting suggestion and let it expand, rather than caging delivery inside interests.",[17,434,435,436,41,440,444],{},"If you are seeding that cold account, the specifics matter. Build the custom audience from the best signal you have, a customer list or recent purchasers, and seed a lookalike off your highest-value customers rather than every buyer, so the match points at quality. Start the lookalike at 1%, the closest match to your seed, and widen toward 2% or 3% only once cost per result climbs as the tight audience saturates. Meta lets you size a lookalike from 1% up to 10% of a country's population, but jumping straight to 10% on a cold account hands the system a loose signal before it has learned anything. The laddering, 1% first then wider, is practitioner convention rather than a Meta-published rule, but it follows directly from how saturation and the learning phase work. Our guides on ",[37,437,439],{"href":438},"\u002Fblog\u002Ffacebook-lookalike-audiences","lookalike audiences",[37,441,443],{"href":442},"\u002Fblog\u002Ffacebook-custom-audiences","custom audiences"," cover the build.",[17,446,447],{},"Then read results at the creative level, not the audience level. Which concept the system favored tells you more than which interest converted, because in 2026 the concept is the targeting.",[17,449,450],{},[65,451],{"alt":452,"src":453},"Diagram of a broad campaign setup showing location and age as hard fences, an objective, and a fan of diverse creative concepts feeding the audience","\u002Fimages\u002Fblog\u002Fbroad-vs-detailed-targeting-creative-setup.webp",[12,455,457],{"id":456},"a-simple-decision-framework","A simple decision framework",[17,459,460],{},"When you are staring at the audience screen, run this order. Is the business location-bound, in a Special Ad Category, built on a specific list you must seed, or selling into a genuinely tiny market? If yes, set the legitimate constraint and stop narrowing past it. If no, go broad: location, minimum age, objective, and a diverse creative set, then let Advantage+ work.",[17,462,463],{},"If you feel the urge to add interests, treat them as a suggestion to test, not a wall to hide behind, and expect delivery to expand beyond them. If results disappoint, the fix is almost never more segmentation. It is sharper hooks, a fresh angle, or a different format, because the lever that used to live in the targeting box now lives in the creative.",[17,465,466],{},"If you would rather settle broad versus narrow for your own account than take anyone's word for it, run a clean test. Build two ad sets that differ in exactly one thing: one broad (location, minimum age, objective, nothing else), one carrying your usual interest stack. Give them equal budgets, the identical creative set, and the same optimization event, then let both clear the learning phase before you read anything, because a result mid-learning is noise. Judge on cost per result after that, not on early click-through or a single strong day. One variable, equal spend, same creative, patience past learning: that is the whole test, and it answers the question for your offer better than any benchmark can.",[17,468,469,470,41,474,478,479,260],{},"That reframing is also where the practical workflow lives. The strong, varied creative broad targeting now demands is exactly what slows teams down, so the loop that scales is research, then make, then measure: study what is working in the market by browsing the Meta Ad Library and a searchable archive of real ads, generate and edit several distinct concepts, launch them straight to Meta, and feed the winners back into the next round. Tools like AdPlay.ai exist to compress that loop, but the discipline matters more than the tool: feed the system distinct creative and let broad targeting do its job. For the mechanics of building those concepts, see our guides on ",[37,471,473],{"href":472},"\u002Fblog\u002Ffacebook-ad-creative-testing","creative testing",[37,475,477],{"href":476},"\u002Fblog\u002Ffacebook-ad-hooks","writing scroll-stopping hooks",", and for the audience layer itself, our deeper walkthrough of ",[37,480,482],{"href":481},"\u002Fblog\u002Ffacebook-ad-targeting","how Meta targeting works",{"title":484,"searchDepth":485,"depth":485,"links":486},"",2,[487,488,489,490,491,492,493,494,495],{"id":14,"depth":485,"text":15},{"id":28,"depth":485,"text":29},{"id":71,"depth":485,"text":72},{"id":89,"depth":485,"text":90},{"id":176,"depth":485,"text":177},{"id":263,"depth":485,"text":264},{"id":349,"depth":485,"text":350},{"id":425,"depth":485,"text":426},{"id":456,"depth":485,"text":457},null,"neutral","Broad vs detailed targeting on Meta in 2026: why detailed targeting is now a suggestion, when broad plus strong creative wins, and when narrow still makes sense.",[500,504,508],{"brand":501,"hook":502,"format":503},"Skinlycious","Testimonial ad for skin that cleared up after weeks of one routine","UGC",{"brand":505,"hook":506,"format":507},"Shakura","Before and After ad for pigmentation faded over a treatment course","Carousel",{"brand":509,"hook":510,"format":511},"UR Klinik","Announcement ad for a clinic consultation slot near you","Static","md",[514,517,520,523,526,529,532,535],{"question":515,"answer":516},"Is broad or detailed targeting better on Meta in 2026?","For most accounts with a working signal and several distinct creatives, broad targeting wins. Meta's Andromeda retrieval engine now reads creative signals to match ads to people, so a strong, varied creative set does the qualifying that interest layers used to do. Detailed targeting still has a place, but it works best as a starting suggestion you hand to Advantage+ rather than a hard fence. Narrow manual targeting earns its keep mainly for local businesses, regulated categories, and small list-based audiences.",{"question":518,"answer":519},"Is detailed targeting still a hard filter?","No. Meta's Help Center describes detailed-targeting inputs as audience suggestions by default: it shows your ad to people who match your suggestion and to others when it predicts better performance. Independent analysts confirm only location and minimum age remain true hard constraints. So you can still add interests, but you should expect delivery to expand past them rather than stay walled inside.",{"question":521,"answer":522},"What is the difference between Advantage+ Audience and Advantage+ Sales campaigns?","They are easy to confuse because Meta named them alike, but they sit at different levels. Advantage+ Audience is an audience setting on the ad set: you give it controls it must obey (location, minimum age, language, and custom-audience exclusions) plus optional suggestions it can expand past. Advantage+ Sales, formerly Advantage+ Shopping, is a whole campaign type that automates audience, placement, and budget end to end. You can use Advantage+ Audience inside a manual campaign, or run an Advantage+ Sales campaign that relies on it underneath. For the broad-versus-detailed decision, what matters is that both treat your interests as suggestions, not hard fences.",{"question":524,"answer":525},"When should I still use narrow targeting?","Four cases. First, location-bound businesses like a clinic or a single-city service, where the geographic radius is a genuine constraint. Second, Special Ad Categories (credit, employment, housing, social or political issues), where Meta restricts targeting by law and removes interests, exclusions, and lookalikes. Third, small list-based custom or lookalike audiences, such as a high-value CRM segment you want to seed precisely. Fourth, a genuinely small total market, like niche B2B, where broad has little room to expand. Outside those, narrowing usually starves delivery without improving relevance.",{"question":527,"answer":528},"Will broad targeting work for a brand-new ad account with no data?","It can, but give it help. A cold account has little conversion signal, so feed Meta a strong seed (a customer list as a custom audience, or a lookalike off your best customers) and let Advantage+ expand from there. Start a lookalike at 1%, the closest match to your seed, and widen toward 2% or 3% only as cost per result climbs; jumping straight to 10% hands the system a loose signal before it has learned. Keep the creative diverse from day one, and as conversions accumulate you can loosen the seed and lean further into broad.",{"question":530,"answer":531},"Does broad targeting work for niche or B2B products with a small total market?","This is the main exception. Broad works by giving the system a large pool to expand into, so when the total addressable market is genuinely tiny, a specialized B2B tool, a luxury item, a hyper-specific niche, there is little for broad to expand toward. In that case an interest or a tight custom audience can be the right call, because you are pointing delivery at the few people who could ever buy rather than asking it to search the whole country. The tell is the real size of your buyer pool, not nerves about wasted spend. Even then, keep the creative varied, because a small audience saturates fast.",{"question":533,"answer":534},"How many creatives do I need for broad targeting to work?","Enough genuinely different angles that the system has something to match against different people, not five near-identical variants. Meta's retrieval engine matches ads to people using signals in the creative, so it rewards distinct hooks, formats, and visuals. A practical starting set is three to five concepts that attack the product from different motivations, then expand the winners. See our creative testing guide for a structured way to run that.",{"question":536,"answer":537},"Does broad targeting lower my cost per result?","It depends on creative and signal, not the audience setting alone. Neutral benchmarks put the average Facebook lead-gen cost-per-lead at $21.98 and conversion rate at 8.78% across industries in 2024, but those swing widely by vertical and offer. Broad targeting tends to help once your creative is strong and your conversion signal is healthy; it tends to hurt when the creative is thin. Treat the audience as the easy decision and the creative as the lever.","\u002Fimages\u002Fblog\u002Fbroad-vs-detailed-targeting-hero.webp","You stacked five interest layers, narrowed the age band, and excluded everyone who looked wrong, then watched Meta spend half the budget on people outside every box you built. That is not the platform ignoring you. In 2026 your detailed-targeting picks are hints the system starts from and walks past, and the real work of finding buyers has shifted onto the creative. Once you stop fighting that, the decision between broad and narrow gets simpler, and the few cases where narrow still earns its keep get easier to spot.",{},true,"\u002Fblog\u002Fbroad-vs-detailed-targeting","2026-10-18",{"title":5,"description":498},[546,550,553,557,560,563,566,570,574,577,580,583],{"label":547,"url":548,"year":549},"Meta Engineering, Andromeda: next-gen personalized ads retrieval engine","https:\u002F\u002Fengineering.fb.com\u002F2024\u002F12\u002F02\u002Fproduction-engineering\u002Fmeta-andromeda-advantage-automation-next-gen-personalized-ads-retrieval-engine\u002F","2024",{"label":551,"url":552,"year":549},"Social Media Today, Meta Shares Insight Into the Performance of Its AI-Based Ad Elements","https:\u002F\u002Fwww.socialmediatoday.com\u002Fnews\u002Fmeta-shares-insight-into-the-performance-of-its-ai-based-ad-elements\u002F734328\u002F",{"label":554,"url":555,"year":556},"Social Media Examiner, Facebook Ad Algorithm Changes for 2026","https:\u002F\u002Fwww.socialmediaexaminer.com\u002Ffacebook-ad-algorithm-changes-for-2026-what-marketers-need-to-know\u002F","2025",{"label":558,"url":559,"year":549},"WordStream by LocaliQ, Facebook Ads Benchmarks 2024","https:\u002F\u002Fwww.wordstream.com\u002Fblog\u002Ffacebook-ads-benchmarks-2024",{"label":561,"url":562,"year":549},"Search Engine Land, Facebook Ads 2024 Data: Clicks and Conversions Up, Costs Down","https:\u002F\u002Fsearchengineland.com\u002Ffacebook-ads-2024-data-clicks-and-conversions-up-costs-down-445162",{"label":564,"url":565,"year":556},"Meta for Business, The Creative Advantage: Unlocking Diversification With Meta Andromeda","https:\u002F\u002Fwww.facebook.com\u002Fbusiness\u002Fnews\u002Fthe-creative-advantage-unlocking-the-power-of-diversification-with-meta-andromeda",{"label":567,"url":568,"year":569},"Apple Developer, App Tracking Transparency Requirements","https:\u002F\u002Fdeveloper.apple.com\u002Fnews\u002F?id=ecvrtzt2","2021",{"label":571,"url":572,"year":573},"Meta Business Help Center, About the Learning Phase","https:\u002F\u002Fwww.facebook.com\u002Fbusiness\u002Fhelp\u002F112167992830700","2026",{"label":575,"url":576,"year":573},"Meta Business Help Center, About Learning Limited","https:\u002F\u002Fwww.facebook.com\u002Fbusiness\u002Fhelp\u002F269269737396981",{"label":578,"url":579,"year":573},"Meta Business Help Center, About the Conversions API","https:\u002F\u002Fwww.facebook.com\u002Fbusiness\u002Fhelp\u002F2041148702652965",{"label":581,"url":582,"year":573},"Meta for Business, Advantage+ Audience","https:\u002F\u002Fwww.facebook.com\u002Fbusiness\u002Fads\u002Fmeta-advantage-plus\u002Faudience\u002F",{"label":584,"url":585,"year":573},"Meta Business Help Center, How to Choose a Special Ad Category","https:\u002F\u002Fwww.facebook.com\u002Fbusiness\u002Fhelp\u002F298000447747885",[587,591,594,597,600,604,607,611,614,618,620,622,625,629],{"label":588,"value":589,"source":590},"ROAS lift Meta reports for advertisers using Advantage+ creative versus not (Meta's own figure, directional)","+22%","Meta Engineering, 2024",{"label":592,"value":593,"source":590},"Recall improvement to the ad retrieval system from Andromeda, across Facebook and Instagram (Meta's own figure)","+6%",{"label":595,"value":596,"source":590},"Ads-quality improvement on selected segments from the Andromeda retrieval system (Meta's own figure)","+8%",{"label":598,"value":599,"source":590},"Increase in model capacity Andromeda enables at the ad-retrieval stage (Meta's own figure)","10,000x",{"label":601,"value":602,"source":603},"Conversion lift Meta reports for businesses using its generative-AI image tools (Meta's own figure, directional)","+7%","Social Media Today (citing Meta), 2024",{"label":605,"value":606,"source":603},"Advertisers using Meta's generative-AI ad tools, producing 15M+ ads in a single month","1M+ advertisers \u002F 15M ads per month",{"label":608,"value":609,"source":610},"CTR lift Meta reports for campaigns using image generation within Advantage+ creative versus campaigns not using those features (Meta's own figure, directional)","+11% CTR","Meta for Business, 2025",{"label":612,"value":613,"source":610},"Conversion-rate lift Meta reports for campaigns using image generation within Advantage+ creative versus campaigns not using those features (Meta's own figure, directional)","+7.6% conversion rate",{"label":615,"value":616,"source":617},"Average Facebook ads cost-per-click, traffic objective, all industries","$0.77 (traffic) \u002F $1.88 (lead gen)","WordStream by LocaliQ, 2024",{"label":619,"value":409,"source":617},"Average Facebook lead-gen conversion rate, all industries",{"label":621,"value":419,"source":617},"Average Facebook cost-per-lead, all industries (down from $23.10 prior year)",{"label":623,"value":624,"source":617},"Average Facebook lead-gen CTR, all industries (traffic-objective CTR 1.57%)","2.53% (lead gen) \u002F 1.57% (traffic)",{"label":626,"value":627,"source":628},"Optimization events an ad set needs within about 7 days to exit Meta's learning phase","~50 events \u002F week","Meta Business Help Center, 2026",{"label":630,"value":631,"source":628},"Minimum US location radius Meta forces on Special Ad Category campaigns","15 miles","blog\u002Fbroad-vs-detailed-targeting","3wptibC90KAAlVI4K9UJ_6gM8bnwCRGbcmxMm3qcRig",1786093698237]