Microsoft Advertising best practices for B2B in 2026: what the documentation settles

Five practices come through the cross-check. Microsoft stands apart in this series: its documentation is dated, and specific enough to settle most arguments on its own. That helps, because I've never worked on Microsoft Advertising (LinkedIn, my former employer, belongs to Microsoft but runs its own ad platform) and have no hands-on time with it; nothing below comes from campaigns of mine.

Nothing made the list on one source alone: every practice here rests on at least two independent ones. The sources were Microsoft Learn, the Microsoft Advertising blog, trade reporting in Search Engine Land and Search Engine Journal, and agency posts. Unusually for this series, Microsoft Learn pages carry last-updated dates, all of them from June or July 2026, which most Google Ads Help pages and LinkedIn's best-practice pages don't. Those dates mark a page's last update rather than its publication, and every one of them below carries that label.

Before anything else, a word on the uplift percentages that circulate for LinkedIn Profile Targeting, the specific improvements in conversion rate or click-through rate: they lead back to aggregator pages whose cited sources can't be traced. None of them appears here; a figure with no origin can't tell you what to expect in your own account.

The Microsoft Ads service page covers what I offer on this channel. The job here is narrower: what the published sources say about each practice, including where Microsoft and its practitioners disagree.

Less than most people assume, and Microsoft says so directly. Its LinkedIn profile targeting documentation (last updated 27 July 2026) is unambiguous: "LinkedIn profile targeting will not narrow your ads' audience. For example, by targeting a specific company, you aren't excluding everyone who does not work for that company", and it tells advertisers to "think of LinkedIn profile targeting as 'bid only', as opposed to 'target and bid'." Optmyzr's guide to the feature describes the same usage, applying industry, company and job function "as bid adjustments to search campaigns", and agency posts from 1Digital Agency (28 July 2026) and Hustle Marketers (24 September 2026) do the same.

For anyone planning account-based work, the constraints matter as much as the mechanism, and the company limit comes first, since it decides whether an ABM list fits at all.

Source: Microsoft Learn, LinkedIn profile targeting, last updated 27 July 2026.

A 1,000-company ceiling with no bulk import is a real operational limit on account-based campaigns, so keep it in mind when you come across advice to upload an ABM list to Microsoft as though it were LinkedIn's own platform.

Start in observation mode, then apply bid adjustments narrowly and raise them gradually. Search Engine Land's guide to LinkedIn targeting on Microsoft recommends using professional demographics "in observation mode to learn which industries, job functions, or company types naturally engage with your Audience ads before applying delivery constraints", and applying bid adjustments only "to campaigns/ad groups where search terms already demonstrate business value". A separate Search Engine Land report on job seniority filters confirms the same option exists for the new seniority dimension, and Digital Applied's seniority guide recommends running it as an observation layer with no delivery restriction before any hard targeting.

The sources part company on how hard to push the bid once observation is done, a plain disagreement about magnitude with no data behind any of the three positions.

Source: as linked above; none of the three publishes a sample or method.

Two more points come from Digital Applied alone, so read them with that label attached: seniority on Microsoft is self-declared rather than employer-verified, and the feature is available in 23 markets, a list that leaves out the United Kingdom, Germany, France and the Netherlands. If either matters to your program, check it in the interface before you plan around it. Microsoft Learn's own targeting page, meanwhile, still lists only company, industry and job function, so on seniority the platform's documentation trails the trade press.

Thirty conversions in thirty days, and all three sources that name a figure name the same one. Microsoft's Performance Max best practices (last updated 17 June 2026) recommends adding a target return "after you've accrued a minimum of 30 conversions within the last 30 days". Search Engine Land states the same threshold for any automated bidding, and Search Engine Journal's piece on making Microsoft profitable gives the fallback: "if you can't reach 30 conversions in 30 days, switch to enhanced cost-per-click (eCPC) or add micro-conversions to maintain optimization."

Agreement like that is rare. On Google, the same question produced a three-to-one spread between the platform's documented floor and what practitioners say is actually needed; on Microsoft, platform and practitioners land on the same number, which makes 30 in 30 the most reliable figure in this article.

None of it works without conversion tracking, which every source treats as the precondition rather than a step. Microsoft's UET setup documentation (last updated 12 June 2026) states that before you can track conversions or target a remarketing list "you need to create a UET tag in Microsoft Advertising and then add the UET tag tracking code to every page of your website", and Search Engine Journal puts the consequence plainly: "Microsoft can't optimize performance with only clicks and impressions." Microsoft also documents that Performance Max campaigns take "two to four weeks or two to three conversion cycles to stabilize", which is the window to judge them over.

That the placement isn't optional: you can manage it, but you can't avoid it. Microsoft's documentation on ads in Copilot (last updated 19 June 2026) is explicit on both halves: "All eligible campaign and ad types are automatically opted in to serving in Copilot. Advertisers cannot opt out of serving ads, and there's no guarantee that their ads will be displayed in Copilot."

The same page names one control that does exist, and it's the familiar one: negative keywords apply, and work "the same as it does for web searches on the Microsoft Advertising Network", so your negative list covers Copilot as well as search. Selection, on the other hand, is a mechanism you don't control: it's conversational rather than query-based, and Microsoft says it considers "the context of the entire conversation, not only the last prompt", a real departure from keyword matching and the reason a B2B advertiser can't reason about Copilot placement with search-term logic alone.

Eligibility carries one documented requirement that's easy to miss: search ads need logo extensions or business logo automated extensions to be eligible at all. Only Microsoft's own page states it, and no independent source has tested it. Search Engine Journal's coverage adds just one point, that no separate campaigns are needed, since Copilot serving is enabled inside existing search campaigns. Everything else published on the subject is Microsoft's guidance about its own inventory, and you should read it that way.

Of all the disagreements between Microsoft and practitioners in these sources, this is the clearest, and for once the practitioners are the majority. Microsoft's Google import post (February 2025) sells the import as near-frictionless, promising it will "seamlessly transfer your creative assets, including images", preserve "audience signals, search themes, and more", and have the campaign ready "in just a few minutes".

Three independent practitioner sources treat the import as a starting point, not a strategy. ZATO's 2026 Microsoft guide recommends optimizing "Microsoft campaigns separately within Microsoft as opposed to importing the changes" made elsewhere; 1Digital Agency argues for rewriting copy and building Microsoft-specific audience layers with independent bid strategies; and Search Engine Land's Microsoft guide advises building dedicated campaigns instead of relying on the import, precisely so LinkedIn targeting and Microsoft-exclusive audiences can be used at all. Search Engine Land's reason carries the most weight: the features that make Microsoft worth running for B2B are the ones Google doesn't have, so they can't be imported.

Microsoft's own documentation on what gets imported (last updated 26 June 2026) backs the practitioners more than its blog does. It states that "any bids or budgets that are too low will be raised to meet our minimums", that the process "converts whichever bid strategies you have set in Google Ads to maximize clicks", and that unsupported location targets are automatically expanded to a nearby state, region or province. It also warns that its own list of imported items "is not a comprehensive list". An import that quietly switches your bid strategy, raises your budgets and widens your geography is a first draft; the actual practice is the audit that comes after it.

Conversion tracking first, then the import audit, then LinkedIn signals, then everything else, and none of the sources argues with that order. UET is a documented precondition for both conversion tracking and remarketing, the 30-conversion threshold can't be reached without it, and every automated bid strategy on the platform depends on both. The import audit comes second because by then bid strategies, budgets and geography have already been changed on your behalf.

LinkedIn Profile Targeting comes third, not first, which is the opposite of how it's usually sold. It only modifies bids, it caps at 1,000 companies with no bulk import, and its published uplift figures have no traceable source. It's a real differentiator, but a weak one to build a plan around.

Copilot stays out of the ordering, since serving there isn't a choice. The practices above also leave one question open: how much reach Microsoft actually offers a B2B advertiser, which comes down to market share, not practice. Our Microsoft Advertising B2B benchmarks article gathers the published data on that. For the platform most of this traffic is imported from, the equivalent synthesis is our Google Ads best practices article, and the Google Ads service page gives the commercial argument for running both.

Does LinkedIn Profile Targeting restrict who sees a Microsoft ad?

No. Microsoft documents it as bid-only rather than target-and-bid, and states that targeting a company doesn't exclude everyone who doesn't work there (Microsoft Learn 2026). For company, industry and job function, Microsoft documents no mode that restricts delivery; trade coverage describes one for seniority only, so plan around bid adjustments.

How many companies can a Microsoft campaign target?

No more than 1,000 per ad group or campaign, and you can't import lists from files or other platforms (Microsoft Learn 2026). Check an account-based list against that ceiling before you build the campaign around it.

How many conversions does Microsoft automated bidding need?

Thirty in thirty days, one of the rare figures on which platform documentation and independent practitioners agree exactly. Below that, the published advice is to fall back to enhanced cost-per-click or add micro-conversions (Search Engine Journal 2026).

Can you opt out of ads serving in Copilot?

No. Microsoft states that all eligible campaign and ad types are automatically opted in and advertisers cannot opt out, with no guarantee of being shown (Microsoft Learn 2026). The control you do have is negative keywords.

Is importing from Google Ads good enough for a B2B Microsoft account?

The sources disagree, and most of them say no. Microsoft markets the import as near-instant, while three practitioner sources recommend rebuilding so Microsoft-only features can be used, and Microsoft's own documentation confirms the import may raise budgets, convert bid strategies to maximize clicks, and widen location targeting. Audit those three fields after every import.