Microsoft Audience Network Targeting: LinkedIn, In-Market & Remarketing

The Microsoft Audience Network gives you six targeting families for native ads: LinkedIn profile targeting (company, industry and job function — no other native channel has it), in-market audiences built from Microsoft's purchase-intent signals, remarketing and dynamic remarketing powered by the UET tag, customer match lists, similar audiences, and the standard demographic, location and device controls. Each audience layer can be applied as "target and bid" (serve only to this audience) or "bid only" (serve broadly, boost bids for the audience). This guide covers how each option actually works, where the data comes from, and how experienced buyers layer them.

Where the targeting data comes from#

Targeting is only as good as the identity graph behind it, and Microsoft's is unusual. The Audience Network draws on first-party data from properties Microsoft owns outright: Bing search history (declared intent), MSN and Microsoft Start reading behavior, Edge browsing signals, Outlook.com engagement — and the LinkedIn professional graph, which is the asset no competing native network can touch.

Your ads serve as native placements across MSN/Microsoft Start, Outlook.com inboxes, the Microsoft Edge new-tab feed and select partner publishers. If you want the full picture of that inventory — what the feed looks like, who reads it, how the ads render — our MSN native ads guide covers the supply side; this article stays on targeting. For orientation on the network itself there is also a compact MSN native ads glossary entry.

Scale, so you know what is behind the targeting: OpenAdLibrary's index holds 281,839 live Microsoft Audience Network creatives as of July 2026 — the largest single network in our 725,882-creative, 49-network corpus. This is not a niche experiment; it is one of the biggest native surfaces on the open web.

LinkedIn profile targeting: the option nobody else has#

The headline feature. Microsoft owns LinkedIn, and the Audience Network is one of the only places outside LinkedIn itself where you can buy against the professional graph. Three dimensions are available:

• Company — target people who work at specific organizations, by name.
• Industry — target by employer industry category.
• Job function — target by functional role family (finance, engineering, marketing and so on).

Notice what is absent: individual job titles and seniority tiers, which LinkedIn's own ad platform sells. On the Audience Network you get the coarser trio — function, industry, company — which is still transformative for B2B on native inventory. The practical plays:

• ABM-flavored prospecting: load a named-account list as company targeting and put your solution in front of those employees while they read the news — at native-feed click prices instead of LinkedIn's premium CPCs. Media buyers consistently report Audience Network clicks costing a small fraction of comparable LinkedIn buys; treat that as directional, not a rate card.
• Function × industry stacking: "finance roles in healthcare organizations" style combinations for vertical SaaS and services.
• Availability caveat: LinkedIn profile targeting is offered in a set of major markets (the US and several other tier-1 countries) and the list evolves — verify current market coverage in Microsoft's official documentation before building a geo plan around it.

One discipline note: LinkedIn dimensions applied as hard targeting can shrink audiences brutally. The standard pattern is to apply them as bid boosts ("bid only") over a broader campaign first, confirm the segments convert, then graduate the proven combinations to "target and bid."

Creative has to carry its share of the targeting, too. A feed reader has no idea they were selected by employer industry — a generic consumer-style headline squanders the selection. Name the professional context ("for finance teams," "if you run field operations") and the LinkedIn layer starts earning its boost; the audience data gets them served, the copy gets the right ones to click.

In-market audiences: intent without the cookie drama#

In-market audiences are Microsoft-curated segments of users showing recent purchase-intent signals — search queries, content consumption, clicks — for a category: "in market for auto insurance," "in market for business software," hundreds more. They are the workhorse cold-traffic option because they encode declared search intent (a Bing query is about as honest as signals get) without requiring any data from you.

Practitioner notes:

• Start with in-market for cold prospecting. It is the closest thing native has to search intent at feed prices.
• Match the segment to the funnel stage, not just the product. Someone "in market for mortgages" responds to rate-comparison angles; the same person six months earlier sits in a broader home-buying segment and needs education-stage creative.
• Segment quality varies by category and geo. Test two or three adjacent segments against each other rather than assuming the obvious one is best.

A worked example of how the funnel-stage point plays out. Say you sell business VoIP. The obvious build targets an "in market for business phone systems" style segment with a demo-request lander — right segment, bottom-funnel ask, and it will convert a thin slice. The stronger build runs two ad groups: the tight in-market segment against the demo lander, and a broader business-services segment against a comparison-guide advertorial that warms the click before the ask. On feed inventory, where the reader was not searching for you ten seconds ago, the second ad group usually produces more pipeline per dollar — intent data tells you who is shopping, but the creative still has to meet them where the feed put them.

Remarketing, dynamic remarketing, customer match and lookalikes#

The retention-and-recovery family, all gated on one prerequisite: the UET (Universal Event Tracking) tag must be live on your site before any of it works. UET is Microsoft's site tag — base snippet on every page, conversion tracking events layered on top — and it feeds both your conversion goals and your audience lists.

• Remarketing lists — rule-based audiences from site behavior: all visitors, cart abandoners, pricing-page viewers, converters to exclude. Standard recency windows apply; build the lists early because they only populate forward from creation.
• Dynamic remarketing — product-level retargeting for retail: pair UET product events with a Microsoft Merchant Center feed and the network re-shows the actual products a user viewed. If you run a product catalog, this is usually the highest-ROAS thing on the whole network.
• Customer match — upload hashed email lists to target (or exclude) known customers and leads. Eligibility requirements apply to the feature; check current conditions in the official docs.
• Similar audiences — Microsoft-generated lookalikes seeded from your remarketing lists, the standard expansion play once a seed list has enough volume.

Two habits worth stealing: always exclude recent converters from prospecting campaigns (feed inventory is cheap enough that waste hides easily), and build a "high-intent non-converter" list — cart or pricing-page visitors minus buyers — as your first dedicated remarketing campaign. It is reliably the best CPA on the account.

Demographics, location and device#

The unglamorous layer that quietly decides efficiency:

• Age and gender targeting or bid adjustment, useful in both directions — the MSN/Outlook audience skews older than social feeds, which is an asset for finance, health, home and insurance offers and a correction to make for youth-market products.
• Location — country, region/state, city-level and radius options, plus geo targeting bid adjustments. The same one-geo-per-campaign discipline that applies on every native network applies here.
• Device — desktop, tablet and mobile splits. The Audience Network has unusually strong desktop supply (Edge new-tab, Outlook on desktop, MSN on Windows), and desktop-vs-mobile behavior differs enough that mature accounts split them.

Worth stating explicitly: because all of this rides on Microsoft's own properties and logged-in graph, the targeting stack is largely insulated from third-party-cookie deprecation. The signals feeding in-market segments and LinkedIn dimensions are first-party by construction, which is one reason sophisticated buyers have been quietly shifting prospecting budget toward this network while retargeting pools on cookie-dependent channels shrink. It also means audience scale varies by how dominant Microsoft properties are in each geo — audience sizes that look great in the US can thin out fast in markets where Bing and MSN have less footprint, so sanity-check estimated reach per geo before committing budget splits.

Target-and-bid vs bid-only: structuring it all#

Every audience association on the network comes in two modes, and choosing deliberately is most of the skill:

The pattern that works: new signals enter as bid-only boosts on top of broad or in-market campaigns; anything that proves conversion lift over a real sample graduates to its own target-and-bid campaign with dedicated budget and creative. This keeps learning volume high while ring-fencing spend for what is proven.

Negative controls matter as much as positive ones — exclude converted audiences, exclude irrelevant placements at the site level when the placement report justifies it, and use category exclusions where brand fit demands. Exclusion options and their granularity change over time, so verify in current documentation rather than assuming parity with Google's display controls.

Three layered builds that work#

• DTC / lead-gen prospecting stack: broad geo + device campaign → in-market segment (target and bid) → LinkedIn industry as bid-only boost → converter exclusions. Feed-native creative angles, and expect the older-skewing audience to reward clarity over hype. If you are coming from Meta, our Meta diversification playbook puts this channel in portfolio context, and the DTC comparison in native vs Facebook ads sets expectations on CPC and creative style.
• B2B stack: LinkedIn job function × industry (bid-only at first) over an in-market business segment → named-company target-and-bid campaign for ABM accounts → remarketing on whitepaper/pricing visitors. B2B on native feeds works precisely because nobody expects it — your competitors are bidding LinkedIn while you reach the same professionals at feed prices.
• Retail stack: dynamic remarketing on product viewers → customer match exclusions → similar audiences seeded from buyers → in-market categories for cold expansion.

Common targeting mistakes on this network#

Five patterns that burn budgets here, all avoidable:

1. Stacking every layer on day one. LinkedIn function × in-market × age × device × custom list produces an audience of four hundred people and a campaign that never exits learning. Layers earn their place one at a time.
2. Treating LinkedIn targeting as LinkedIn Ads. You get function, industry and company — not titles or seniority. Campaigns planned around "target CFOs" need re-scoping to "finance function at companies of the right profile," with creative doing the seniority filtering.
3. Importing Google display audiences one-to-one. The Microsoft graph is built from different signals (search on Bing, LinkedIn, MSN behavior); segment names that look identical can perform very differently. Re-test rather than assume.
4. Ignoring the age skew. This inventory reaches an older, wealthier reader than social feeds. Offers aimed at that demographic (finance, health, home, insurance) under-invest here; youth-market offers over-invest.
5. Set-and-forget lists. Remarketing windows, customer match files and exclusion lists all decay. A quarterly audience audit is cheap; six months of re-targeting churned emails is not.

Verify what's actually running before you build#

Targeting theory is cheap; the live feed is evidence. Because Microsoft publishes no ad library for Audience Network inventory, the way to study it is an independent index: OpenAdLibrary's ad intelligence platform lets you browse those 281,839 live Audience Network creatives by advertiser, vertical, geo and days running. The vertical mix as of July 2026 — ecommerce (9,978 classified creatives), finance (9,029), travel (8,830), insurance (8,406), software (7,909), education (5,025) — tells you which categories have already made the economics work, and creative longevity tells you which specific angles are paying. This WalkFit creative had been running 30 days at capture, a strong profitability signal on CPC inventory:

Study the advertisers who persist in your vertical — their audience choices are invisible, but their geo split, device split, creative angles and landing pages are all observable, and those choices encode what their targeting data taught them. If a competitor runs desktop-heavy in three tier-1 geos with retiree-focused creative, you are looking at the output of their audience data, reverse-legible from the outside. Where this network sits against the rest of the native landscape: best native ad networks has the ranked comparison.

Targeting checklist#

1. UET tag live sitewide before anything else; conversion goals defined.
2. Remarketing lists created on day one (they populate forward only).
3. Cold traffic starts with in-market segments, one geo per campaign.
4. LinkedIn dimensions enter as bid-only boosts; graduate winners to target-and-bid.
5. Confirm LinkedIn-targeting market availability in official docs before geo planning.
6. Converters excluded from all prospecting campaigns.
7. Desktop and mobile split once volume supports it.
8. Placement/site exclusions applied from data, not assumption.
9. Every layer added one at a time — a stack you built in one afternoon is a stack you cannot debug.
10. Competitor angles and longevity checked against the live index before creative briefs.