Microsoft Ads ROI: A UK Marketer's Guide for 2026

Microsoft Ads is the channel most UK marketers switch on, import from Google, and then never look at again. That neglect is precisely why the ROI is often better than Google's, why so many accounts quietly leak budget for years, and why a proper Microsoft Ads audit is usually one of the cheapest improvements available in a paid search budget.

Why Microsoft Ads is genuinely worth UK marketers' attention in 2026

The honest case for Microsoft Ads is not that it will replace Google. It will not. StatCounter's UK figures for August 2026 put Google at 91.75% of searches across all devices and Bing at 5.67%. On desktop the gap narrows: Bing takes 10.38% of UK desktop searches against Google's 85.9%. Any plan that treats Microsoft as a like-for-like alternative will disappoint. The case is narrower and more useful: the same query, from a similar buyer, frequently costs less because fewer advertisers bothered to compete for it.

That advantage comes from a specific audience. Bing's share is nearly twice as high on desktop as it is across all devices, and Windows search and the Edge browser both default to Bing, which puts it in front of plenty of people on managed work machines. In practice a good part of that audience is office workers, on company machines, during working hours. For B2B, professional services, software, recruitment and considered high-value purchases, that is a strong audience. For impulse consumer purchases skewing young and mobile, it is a weak one.

Two things have changed since most UK accounts were built. The first is Copilot. Microsoft now serves ads inside Copilot's answers, built automatically from your existing campaign assets; according to Microsoft's ads in Copilot documentation, every eligible campaign is opted in, advertisers cannot opt out, your negative keywords still apply, and there are currently no Copilot-specific metrics. The second is consent. Since 5 May 2025 Microsoft has required consent signals for visitors from the UK, the EEA and Switzerland, and an account that never set them up is under-reporting its own results. (If you are wondering what to call the platform: Bing Ads was renamed Microsoft Advertising in 2019, and most people still say Microsoft Ads.)

The catch is that the ROI advantage is fragile. It exists because the auction is thinner, and it evaporates the moment you run the account on autopilot. Most of the underperformance we see is not a platform problem. It is an attention problem, and much of it is fixable in an afternoon.

Six high-leverage levers for higher Microsoft Ads ROI

These six changes account for the majority of the improvement in the accounts we take over.

Import then prune, don't import and forget

The Google import is the right starting point and the wrong finishing point. Bid strategies, budgets, assets and negative lists all transfer assumptions built for a different auction. After importing, cut the keywords with no Microsoft search volume and reset budgets to reflect the smaller traffic pool. If you keep a scheduled import running, switch off the settings that overwrite your Microsoft work: Update bids, Update bid strategies and Update existing campaign budgets. Otherwise every change you make is reset to Google's values on the next sync.

Get consent mode working for UK visitors

The Universal Event Tracking (UET) tag has to receive a consent signal, through Microsoft's consent mode (the ad_storage parameter) or the IAB consent framework, for visitors from the UK, the EEA and Switzerland. Microsoft's consent FAQ is blunt about the alternative: it stops tracking UET-based conversions and populating remarketing lists for those visitors. If your cookie banner was set up for Google only, check that Microsoft's tag is listening to it too.

Use device bid adjustments deliberately

Microsoft lets you adjust bids by device anywhere from -100% to +900% for computers, tablets and smartphones, and automated bid strategies take those adjustments into account as signals. Given how desktop-weighted the audience is, a blanket bid across devices leaves money on the table in both directions. Check the device split in your own conversion data before assuming: it varies enormously by sector.

Match-type discipline beats keyword volume

Broad match on a thin auction is how budgets disappear. With less overall volume, a single badly matched query can consume a meaningful share of the daily spend. Start exact and phrase, prove intent, then widen deliberately. Review search terms weekly rather than monthly, because the sample is smaller and problems take longer to become statistically obvious.

Set conversion goals before launch, not after

The UET tag needs to be live and firing correctly before the first click, or the first month of data is worthless for optimisation. Only goals with the Include in “Conversions” box ticked feed automated bidding, so decide which goals should steer the bids. Where the sale closes off-platform, use offline conversion import: each conversion needs its Microsoft Click ID (MSCLKID), a conversion uploaded more than 90 days after the click will not be imported, and Microsoft recommends uploading daily. Automated bidding on badly configured conversion data will confidently optimise towards the wrong outcome.

Know where your ads actually show

Ad distribution is set at ad group level: either the entire Microsoft Advertising Network, which adds wider partner traffic such as Yahoo search partners, or Microsoft sites and select partner traffic only. Microsoft's ad distribution documentation notes that every option includes both search and audience ads, so picking a distribution option does not take audience ads out of a search campaign. The controls are reporting and exclusions: run the Website URL (publisher) report, exclude the partner sites that spend without converting, and judge search and partner traffic on their own numbers rather than blended together.

What LinkedIn profile targeting can and cannot do for UK campaigns

LinkedIn profile targeting (company, industry and job function, with job seniority added for Search and audience ads in August 2026) is the feature most often quoted as Microsoft's big advantage over Google, and for many markets it is. It works as a bid adjustment rather than a filter: targeting a company raises or lowers your bid for people who work there, it does not stop your ads showing to everyone else.

There is a catch for UK advertisers that most guides miss. Microsoft's market availability page says the capability is available globally, but the underlying data excludes consumers in EEA countries for regulatory reasons, and the list of markets where consumer data is included (23 of them, from Argentina to Vietnam) does not include the United Kingdom. In other words, a bid adjustment for a job function will not find UK searchers to act on. It remains useful if you also target the included markets, such as the United States, Canada or Australia, and it is worth re-checking that page before each audit, because Microsoft updates it. What it is not, for a UK-only account, is the lever it is often sold as.

For UK B2B the practical substitutes are the ones that work on any platform: keyword intent that only a professional would type, working-hours ad schedules, desktop bid adjustments, remarketing lists built from your own site, and in-market audiences layered on with bid adjustments.

How to actually calculate Microsoft Ads ROI for UK businesses

Platform ROAS is not ROI. The number in the interface divides conversion value by ad spend and stops there, which flatters every channel it touches. A usable ROI figure has to account for three things the platform does not know.

Gross margin, not revenue. A 4:1 ROAS on a product carrying 20% margin is a loss. Divide by margin before you celebrate. For lead-generation businesses, substitute the average value of a closed deal multiplied by your lead-to-close rate.

Management cost. Whether that is agency fees or the hours your own team spends, it belongs in the denominator. A channel spending £1,500 a month that takes six hours a week to manage is more expensive than the invoice suggests.

Attribution honesty. Microsoft's default attribution will claim conversions that other channels contributed to, exactly as Google's does. If you run both, some conversions are being counted twice across the two interfaces. The only reliable check is comparing platform-reported conversions against your own CRM or order data for the same period. For UK visitors, also check that the gap is not simply missing consent signals.

The working formula for most UK SMBs looks like this: take conversions from your own records rather than the platform, multiply by average order value, multiply by gross margin, subtract ad spend and management cost, then divide by the sum of ad spend and management cost. It is less flattering than the dashboard and it is the only number worth reporting to a finance director.

A worked example with round numbers: 40 orders a month from Microsoft in your own order system, at an average order value of £250 and a 40% gross margin, is £4,000 of gross profit. Against £1,500 of ad spend and £500 of management time, the channel returns £2,000 on £2,000 invested: an ROI of 100%, or 1:1. The platform might report the same month as a ROAS of 6.7:1, because it divides £10,000 of revenue by £1,500 of spend. Both numbers are arithmetic, but only one tells you whether to spend more. We apply the same discipline across every paid channel in our digital marketing work, and our guide to data-driven marketing covers how to build the reporting that makes it possible.

One further point specific to Microsoft: the volume is lower, so the confidence intervals are wider. Thirty conversions in a month on Google is a signal. Thirty conversions on Microsoft might be a fortnight of noise. Judge Microsoft performance over quarters, not weeks, or you will keep making changes in response to randomness.

The Microsoft Ads audit checklist, in the order that finds money fastest

Work through this in sequence. The early items find the largest wasted spend, so if you only have thirty minutes, the first two sections are where to spend it.

1. Conversion tracking, consent and goals

• Is the UET tag present and firing on every page, including the thank-you page? Check it in the browser, not just in the interface.
• Is consent mode (or the IAB framework) passing a signal to the UET tag for UK visitors? Without one, Microsoft stops recording UET conversions and filling remarketing lists for UK, EEA and Swiss traffic.
• Are conversion goals set to count correctly (all versus unique) for your business model?
• Which goals have Include in “Conversions” ticked? Only those feed automated bidding, and duplicate goals inflate the numbers and mislead it.
• If sales close offline, is offline conversion import running, with MSCLKID captured on every lead form?
• Do platform conversions roughly reconcile with your CRM or order records for the same period? A gap over about 20% means something is broken.

2. Search terms and negative keywords

• Pull the search terms report for the last 90 days, sorted by cost, and read the top fifty.
• Add negatives for every irrelevant query, and apply them at the right level: shared lists for account-wide exclusions, campaign level for anything more specific. Negatives also govern where your ads can appear in Copilot.
• Check your negative match types. An exact negative blocks only that exact query, so a broad-sounding exclusion may be doing far less than you think.
• Look for queries triggering the wrong ad group. Cross-matching within an account is a quiet waste.

3. Campaign structure, distribution and match types

• Which ad distribution option does each ad group use, and what does the Website URL (publisher) report say about the partner sites spending money? Exclude the ones that do not convert.
• Are audience ads carrying spend inside your search reporting? Break them out so they are judged on their own results.
• Are broad match keywords running without a strong negative list behind them?
• Are budgets distributed by opportunity or by whatever the Google import suggested?
• Are there imported keywords with zero impressions over 90 days? They add noise and management overhead. Remove them.

4. Bids, budgets and device splits

• Check the desktop, mobile and tablet performance split, then set adjustments to match reality rather than habit.
• If a campaign is limited by budget on a profitable ROAS, raise it. If it is spending in full on a losing ROAS, cut it.
• Review the bid strategy. Microsoft's manual strategies are now limited to formats such as cost per sale and manual CPM, so on Search the choice is between Enhanced CPC, where you still set keyword bids, and fully automated strategies. Since 4 August 2025, target CPA and target ROAS for new campaigns are set as optional targets inside Maximize Conversions and Maximize Conversion Value. On thin conversion data, Enhanced CPC often holds up better than a CPA target the account cannot yet support.
• From 1 October 2026 the Max CPC cap is no longer available on new non-portfolio campaigns using Maximize Conversions with a target CPA, Maximize Conversion Value with a target ROAS, or Maximize Clicks. Existing campaigns keep it, and portfolio bid strategies still offer it, so check before you rebuild a campaign that relies on a cap.
• Check the ad schedule against when your conversions actually land, which for B2B is usually working hours.

5. Ads, assets and landing pages

• Are there at least three responsive search ads per ad group, with pinned headlines only where genuinely necessary?
• Are sitelinks, callouts and structured snippets present at campaign level? Missing assets cost impression share cheaply, and your text and images are what Microsoft uses to build ads in Copilot.
• Do the landing pages match the ad promise, and do they load quickly on desktop? Our guide to landing page design covers the build side.
• Is the same landing page working for both Google and Microsoft traffic, despite the audience being different? Often it should not be.

6. Audiences and targeting layers

• Are remarketing lists built and applied? They are frequently missing entirely on imported accounts, and they stay empty for UK visitors if consent is not configured.
• Are in-market audiences layered on with bid adjustments?
• If LinkedIn profile targeting is applied, does the campaign target any market where the data is actually available? On a UK-only campaign it is doing nothing.

Microsoft's own help centre covers the mechanics of each setting if you need the exact steps, and its guidance on automation and ROI is worth reading before handing more control to automated bidding.

What an audit should hand you at the end

An audit that ends with a list of observations is half done. Whether you run it yourself or pay for it, the output should be something you can act on in order:

1. A measurement verdict. Is the conversion data trustworthy enough to optimise on, including consent for UK visitors? If not, everything else waits.
2. A wasted-spend figure. Search terms, partner sites and audience placements that took budget over the last 90 days without converting, with the cost attached.
3. A settings list. Import sync, distribution, bid strategy, device adjustments and targeting layers, each marked keep, change or remove.
4. A prioritised plan. The three changes that will move ROI most in the next 30 days, and what to measure to prove they worked.

Repeat the full audit at least twice a year and after any large Microsoft product change. Given how much changed in 2025 and 2026, from consent enforcement to bid strategy rules, an account last audited two years ago is being run on assumptions that are no longer true.

Microsoft Ads versus Google Ads: a UK budget allocation framework

The question is rarely which platform is better. It is how much of a fixed budget belongs on each. A workable starting split for UK advertisers:

• B2B, professional services, software, recruitment: start at 20 to 25% of paid search budget on Microsoft. The desktop, working-hours audience genuinely matches, even without LinkedIn data for UK users.
• Considered consumer purchases (finance, home improvement, healthcare, travel): 10 to 15%. Worth running, worth watching.
• Fast-moving consumer ecommerce, youth-skewing brands, mobile-first apps: 5% as a test, or skip it. The audience is not there in volume.

Treat those as hypotheses to test, not rules. The decisive evidence is your own cost per acquisition on each platform after 90 days, calculated with real margin. If Microsoft's blended CPA comes in materially below Google's, shift budget towards it until the advantage closes, which it eventually will as you exhaust the cheap inventory. Current UK cost benchmarks by sector are tracked in our UK PPC Index, and our guide to what Google Ads costs in the UK covers the other side of the comparison.

One trap worth naming: do not fund Microsoft by cutting a Google campaign that is currently limited by budget and profitable. Fund it from waste, or from genuine incremental budget. Our Google Ads management approach starts by finding that waste, and the same audit logic applies on either platform.

The five most common reasons Microsoft Ads campaigns underperform

1. The account is a Google clone. Same keywords, same budgets, same bids, same landing pages, different auction, often with a scheduled import still overwriting every change. It will underperform and the platform will take the blame.
2. Conversion tracking was never verified. The tag is installed but not firing on the confirmation page, or it is firing without a consent signal, so UK conversions go unrecorded and the account optimises towards clicks. This is the most common single fault we find.
3. Partner and audience traffic is buried inside search reporting. Syndicated and display-style placements drag down the blended numbers and nobody notices because they are not broken out.
4. Judgement is being made on too little data. Decisions taken weekly on a channel producing a handful of conversions a week are decisions taken on noise. This is the same failure mode that makes automated bidding misfire when the underlying signal is thin, which we cover in more depth in our analysis of why Performance Max fails on poor data.
5. Nobody owns it. Microsoft is treated as a secondary channel, so it gets reviewed quarterly at best. The account then drifts, and the drift is read as proof the platform does not work.

In-house or a Microsoft Advertising agency: how to decide

Microsoft Ads is genuinely easier to run in-house than Google Ads, because the account is smaller and the auction is less volatile. If you have someone competent already managing Google, adding Microsoft is a few hours a month rather than a new hire.

Bring in outside help when one of three things is true: the account is spending enough that a few percentage points of efficiency exceeds the fee, the conversion tracking or consent set-up is broken in ways nobody internally can diagnose, or nobody has looked at it in six months. That last one is the most common, and the cheapest to fix.

What to ask any agency before signing: how they will report ROI rather than ROAS, whether they will break out partner and audience traffic, what they will change in the import settings, how they will verify consent mode for UK visitors, and what they will do in the first thirty days. Vague answers to the first question are disqualifying. The same questions apply to server-side and offline conversion work on other platforms, which our guide to the Meta Conversions API covers for Facebook and Instagram. Background reading on the wider channel mix is in our guides to getting more from Bing Ads and how PPC advertising works.

Frequently asked questions