There is a specific kind of complacency that shows up in a worryingly large number of paid media accounts:

A Microsoft Ads setup created via the "Import from Google" button a few years ago, never touched since, quietly absorbing five or ten percent of the search budget while everyone focuses on the 'real work'.

When set up this way, the channel typically produces results modest, unremarkable results - and nobody asks too many questions about it.

The problem is, this approach is leaving significant performance on the table.

Not in the theoretical sense.

In the "£18,000 more revenue with £1,200 less spend in a single month" sense. 

(That's the actual data Matt Beswick shared publicly earlier this year, sparking a wave of skepticism that turned out to be entirely unwarranted).

To break down exactly how marketers should be setting up their Microsoft campaigns to achieve similar results, Navah Hopkins, (Microsoft Ads Liaison) and Matt Beswick (Microsoft Ads specialist) joined us for a session to cover the full picture, including:

• Where the platform sits in 2026

• What the latest AI features actually mean in practice, and what you should care about

• And exactly how to build Microsoft campaigns from the ground up in a way that competes with — and in many accounts, beats — Google on CPA.

Watch the full session on-demand here, or keep reading for the written rundown:

Timestamps:

0:00 - Intro to our panel
07:15 - Session agenda
11:35 - The LinkedIn attitude
13:00 - Why Microsoft Ads deserves your budget
19:40 - New Microsoft Ad features in 2026
24:50 - Copilot (& why it's actually good now!)
29:55 - AI Max for Microsoft Ads
33:00 - PMax for Microsoft Ads
39:28 - Auto-generated creative & Ad Assist
41:50 - Microsoft Clarity (free tool)
45:35 - Navah's new feature summary
48:25 - Creating Bing campaigns that beat Google
51:52 - Placement exclusions in Microsoft Ads
55:00 - [Free resource] Placement lists for Microsoft Ads
57:17 - Live audience Q&A

Why Microsoft Ads deserves its own strategy in 2026

The most persistent myth about Microsoft Advertising is the demographic.

You might think the platform typically serves an older, less relevant audience.

Perhaps Bing is for people who haven't bothered to change their default search engine - or for the demographics that, on large, lack the technical knowhow to do so.

Or, that it's fine for certain B2B use cases - but irrelevant for anything that involves B2C audiences.

Navah brought one cold hard stat to tear down those walls straight away: 

73% of reachable users on the Microsoft network are 45 or younger. 

But that headline stat isn't the punchline itself.

The point is that the platform spans every surface where Microsoft is embedded in daily life: Bing, Copilot, Edge, Outlook, Windows. The audience assumption is wrong, and it's costing advertisers who rely on it.

"The amount of reachable humans who are younger is drastically growing.

And the fact that Bing now reaches over a billion users monthly... I don't think enough people realize how many humans value the accessibility and inclusivity principles that Microsoft represents."
— Navah Hopkins

Matt came at the same argument from the data side. When he first shared the revenue-versus-spend comparison publicly, the response was disbelief.

His explanation for why it looks the way it does was straightforward: Microsoft simply performs better across every metric that matters for his accounts.

"The data is telling me that's what I need to do. We're getting lower CPAs, higher average order values, stronger ROAS, more calls. Every metric we see is stronger. I would not be doing my job properly if I didn't spend more time in Microsoft." — Matt Beswick

The structural reason most advertisers under-invest, he suggested, is less about the evidence and more about inertia: agency time concentrates around Google because that's where the budgets are, and Microsoft's returns only become visible once someone decides to treat it seriously as a platform in its own right.

Copilot: AI diagnostics, creative, and keeping humans in the loop

Microsoft has been building Copilot as a genuinely useful tool for advertisers rather than another opaque automation layer, and Navah spent a meaningful portion of the session explaining what that actually looks like in practice.

The most practically useful recent addition is root cause analysis for performance shifts. When a campaign drops or spikes, Copilot can now identify which change appears to have caused it, trace it back through account history, and flag it clearly.

This is diagnostic intelligence that removes the grunt work of combing through thousands of rows of data - and critically, Navah drew a distinction worth holding onto: this feature has nothing to do with the recommendations tab.

"Root cause analysis is specifically when you're talking to Copilot within Microsoft Advertising and you ask: what's going on with this campaign? It will tell you these are some things that seem off, or these changes were made and they might be related. It has nothing to do with recommendations."
— Navah Hopkins

On the broader question of whether Copilot's judgment can be trusted, Navah shared an example that makes the case plainly. She asked Copilot whether to use broad match or exact match for an account with low conversion volume.

The response: use exact match, because broad match needs conversions to function. Hardly the answer a platform trying to inflate spend would give.

"When you know the AI system is on the human side, it's much easier to trust it — and give it tasks you might otherwise hesitate to share that data with."
— Navah Hopkins

The underlying principle here is one Navah returned to throughout the session: Microsoft's approach to AI is built around empowering human choice rather than replacing it.

The framework from two Microsoft blog posts published in April — describing three eras of the web (the human web: "help me find it"; the LLM web: "help me choose"; the agentic web: "do it for me") - maps where the industry is heading.

But Microsoft's position is that the AI tooling should support human judgment at every stage, not preempt it.

Matt's take on Copilot was more specific, and worth noting: Microsoft's AI ad surfaces are live and delivering measurable results right now in a way Google's simply aren't yet.

"On Microsoft, we're actually seeing our ads served in Copilot quite frequently. On other platforms, we're just not getting placed in AI overviews or AI mode at the moment. It's giving us another placement, another place to be found, that we're simply not getting on Google right now."
— Matt Beswick

Navah's data on those Copilot placements underlines why this matters: ads served inside Copilot convert at a 194% better rate than equivalent SERP ads, with 25% higher relevancy scores.

The catch is that relevance thresholds are higher — creative and keyword strategy need enough flexibility to fit the conversational context of a Copilot exchange, which is why PMax and AI Max are better positioned to unlock those placements than rigid exact match setups.

AI Max, Performance Max, and how they sit together

For advertisers running Google's Performance Max, Microsoft's version covers much of the same conceptual ground - but with structural differences that Navah flagged as significantly underappreciated.

On new customer acquisition goals, Microsoft has taken a more conservative approach than Google. If the platform isn't fully confident someone is a new customer, they're classified as existing rather than unknown.

"If we are not 100% confident that this is a net new customer, we actually classify unknown as existing customer. You can be much more confident that if we say someone is new, the data is accurate."
— Navah Hopkins

In practice, this means you may see lower new customer volumes reported than on Google - but the numbers you do receive are more reliable.

Customer match lists help from day one: Microsoft accepts a minimum of 300 emails, which is achievable for most advertisers and should be built out early.

Matt's observation on Microsoft PMax versus Google PMax focused on offline conversion handling - an area where Microsoft's smaller data set might be expected to be a disadvantage, but turns out not to be:

"Microsoft seems to understand our offline conversion journey better than Google at the moment. Phone calls, converted leads — Google seems to favour the immediate online conversions because they happen straight away. Microsoft seems to understand more about the ones we get fewer of." — Matt Beswick

AI Max remains in closed pilot at the time of writing, but the direction is pretty clear: it expands keyword matching across Copilot and Bing, surfaces product details like free shipping directly inside Copilot conversations, and introduces an audience generation tool that builds targeting segments from plain-language descriptions.

Navah's framing of where it fits is arguably the most useful context: AI Max is a step toward AI adoption for advertisers currently running tight keyword strategies.

If you're already on PMax, the incremental lift is likely to be modest. If you're predominantly on exact match, that's where it earns its place.

Auto-generated creative: a testing asset, not a strategy

Navah's Search Engine Journal article on auto-generated creative framed the debate clearly, and she covered the essentials efficiently.

A LinkedIn poll she ran found that 84% of respondents weren't using auto-generated creative in their campaigns. Whether that reflects sound strategic reasoning or default skepticism is the more interesting question.

The actual data Navah shared suggests the outcome depends almost entirely on execution quality. In a 30-day test at a logistics company, a top-performing AI-generated ad outperformed a human-created ad across both click-through rate and conversion rate. A lower-performing AI ad underperformed the human equivalent. The mean isn't the story, the variance is.

"This is not a blanket call to say yes to AI and auto-generated creative all the time. It's a call to allow it to be part of your testing process."
— Navah Hopkins

The more pressing point: Microsoft has now enabled Ad Assist by default for all customers outside of sensitive verticals.

That means auto-generated assets are now filling RSA text fields in under-resourced campaigns whether advertisers have opted in or not. It's worth auditing your account to understand where auto-generated assets are currently active — and making an active, intentional decision about whether to test them, constrain them, or turn them off.

For regulated or sensitive industries, Navah confirmed that Microsoft pre-filters those verticals out of auto-generated creative at launch, and that brand kit controls in Ad Studio can lock in fonts, colours, and messaging as those tools develop further.

Microsoft Clarity: the free tool most advertisers are ignoring

Matt opened this section with a question: "How is Microsoft Clarity still free?" - which is both genuine and a prompt for every marketer to check they've actually installed it.

The baseline Clarity functionality (heat maps, session recordings, rage click and fast-back tracking) competes with tools that cost hundreds of pounds per month.

In Matt's accounts, it's been practically useful for identifying friction points on landing pages that explain why high-intent visitors aren't converting, and for surfacing those insights quickly enough to act on them.

"There are loads of little nuggets of gold that can help you improve performance across all your advertising and organic channels. And then with this new AI angle, we can start to see places where we're not getting coverage on AI-generated answers and build a content plan around the things we're missing."
— Matt Beswick

The more significant development is what Clarity is becoming. New AI citation reports let advertisers see not just how humans are behaving on-site, but how AI systems are interacting with and referencing the site — and, crucially, where competitors are being cited in AI-generated responses instead of them.

For anyone running PMax or preparing for AI Max, this is direct intelligence on the content gaps that affect both organic discoverability and ad relevance in AI surfaces.

Navah reinforced Clarity's commitment to remaining free permanently and highlighted its A/B testing capability as an important safeguard: before making sweeping landing page changes based on AI engagement signals, test them first.

Optimizing for how AI systems interact with a page at the expense of how humans convert on it is a real risk worth guarding against.

Building Bing campaigns from scratch: The 5-step framework

The second half of the session shifted into Matt's practical guide to building Microsoft Ads accounts without relying on a Google import — drawn from his PPC Live article, "The Case for Building Microsoft Ads From the Ground Up."

The article is worth reading in full alongside this session; what follows is the distilled version.

The starting position is the most important principle: do not import from Google.

"It's a completely different platform. It's got different users. They react in different ways to different messaging. I don't import anything from Google apart from the merchant feed." — Matt Beswick

Matt's five steps, in order:

• Step 1: Don't export everything from Google Ads. The temptation is to import the whole account structure to get live quickly. The problem is that Microsoft's algorithm, audience makeup, and keyword dynamics are different enough that a direct copy underperforms from day one — and you'll spend time optimizing around a flawed foundation rather than building the right one.

• Step 2: Embrace consolidation. If you have 15 to 20 campaigns in Google, start with four or five in Microsoft. Concentrating the data gives the algorithm what it needs to learn effectively, accelerates the ramp-up period, and keeps the account manageable while you build familiarity with how the platform behaves.

• Step 3: Use Dynamic Search Ads. Matt's accounts run 60,000-plus SKUs — creating individual search campaigns at that scale isn't viable, so DSAs do much of the heavy lifting. While Google has deprecated DSAs, Microsoft continues to support and develop them. Navah noted that PMax can replicate much of the intent signal value DSAs provide for those looking further ahead.

• Step 4: When it's working, leave it alone. This applies everywhere, but it matters more on Microsoft where lower conversion volumes mean the signal takes longer to accumulate. Unnecessary changes reset the learning period. Patience is a genuine competitive advantage here.

• Step 5: Scale spend. Once the account is performing, invest in it proportionally. The common mistake is keeping Microsoft capped as a secondary channel even after the data clearly justifies more budget.

On LinkedIn Profile Targeting (which came up in the Q&A as Microsoft's most distinctive structural B2B advantage) - the advice for getting started is to add audiences in observation mode first, understand how different professional segments interact with your creative, then adapt messaging around the pain points that actually resonate.

Navah's note on the mechanics is worth keeping in mind: combining targeting criteria using AND logic narrows the audience, while OR logic expands it and can spike bids unexpectedly. Keep your targeting lanes clear before adding bid adjustments.

On placement exclusions, Navah highlighted the enhanced URL metrics now available in PMax, giving advertisers the spend and conversion data needed to make an evidence-based decision on whether a placement should be excluded — or, conversely, carved out into a dedicated always-on audience campaign.

Navah's top tip: manage exclusion lists at the MCC level to apply unlimited URL exclusions across multiple accounts, rather than adding them manually at the ad group level.

Final thoughts

If there's one key takeaway from the session, it's that Microsoft Ads in 2026 is not the platform it was a few years ago, and the advertisers still treating it as a cheap Google mirror are making an increasingly costly mistake.

Navah's data on Copilot placement performance, Matt's real-account revenue comparison, and the breadth of AI tooling that's now live or close to it all point in the same direction.

If you want to go further, the full recording includes a live Q&A (starting at 57:17) - covering LinkedIn targeting for B2B SaaS, DSA sunset questions, auto-generated creative in regulated markets, and a detailed discussion of whether Copilot's root cause analysis can genuinely be trusted. It's worth the extra time.

And of course, a huge thank you to both Navah and Matt for bringing real expertise and real account data to the session. Make sure to follow Navah on LinkedIn and follow Matt on LinkedIn - both are consistently among the most useful people to have in your feed if Microsoft Ads is on your radar.

For further Microsoft Ads guidance, check out the session docs & resources below. 

Plus, make sure to use our free junk placement exclusion lists to automatically exclude junk domains from your campaigns. (Bonus exclusion lists are available for those in the IT & Security / Banking & Lending sectors!)

Exclusive junk placement exclusion lists:

• IT & Security — 1k junk placement exclusion list
• Banking, Lending & Credit — 1k junk placement exclusion list
• Lunio's 40k Performance Max placement exclusion list

Additional resources from the session:

• Follow Navah Hopkins on LinkedIn
• Follow Matt Beswick on LinkedIn
• Follow James Deeney on LinkedIn
• Get a free 14-day Lunio traffic audit
  
  
• 73% of reachable users on the Microsoft network are 45 or younger — Navah Hopkins
• Beyond Google — free eBook (PPC.Doctor)
• Win across all three eras of the web — Tim Frank, Microsoft Advertising
• An ads ecosystem built for the AI era — Pallavi Naresh, Microsoft Advertising
• Copilot, AI updates & Microsoft Clarity — Navah Hopkins on LinkedIn
• Microsoft AI Max places search ads inside Copilot responses — DesignRush
• Should you use auto-generated creative? — Navah Hopkins, Search Engine Journal
• How is Microsoft Clarity still free? — Matt Beswick on LinkedIn
• Microsoft's AI vision: transparency, control, and joy — Navah Hopkins
• The case for building Microsoft Ads from the ground up — Matt Beswick, PPC Live
• The Paid Media Lab Playbook
• 2026 Global Invalid Traffic Report
• All in on AI: Agentic Commerce — Microsoft Advertising
• How to steer your brand in AI-powered search — Microsoft Advertising
• How to use AI ad placements for stronger PPC performance — Search Engine Journal
• How to identify and solve click fraud in paid media — Search Engine Journal
• New import center and other product news for May 2026 — Microsoft Advertising
• 5 priorities for lead gen in AI-driven advertising — Search Engine Land
• Query intent vs. conversion intent: why the difference matters — Search Engine Land
• How to use LinkedIn targeting in Microsoft Advertising — Search Engine Land
• Google and Microsoft: how their Performance Max approaches align and diverge — Search Engine Land
• The future of remarketing? Microsoft bets on impressions, not clicks — Search Engine Land