A $6.20 Google click can cost $2.40 on Bing and still be the worse buy. I used to tell clients that gap was free money. I was wrong: a cheaper click tells you about the auction, not what happens after someone lands on the site.

Comparisons put Microsoft averages 30 to 50% below Google, around $1.50 to $2.80 versus $2.85 to $5.26. Fewer bidders help explain the gap. But Bing is a thinner auction serving a different searcher mix, with partner traffic and identity data Google cannot copy. Those mechanics decide whether the discount becomes revenue or just a smaller invoice.

I will build the model one layer at a time, using a B2B software client spending $20k a month on Google Search. The question is not whether Microsoft can sell that client a cheaper click on project management software. It can. The question is how many useful clicks it can sell, from whom, and under what conditions.

First, fewer bidders change the price and the ceiling

On Google, project management software might face eight serious bidders with decent Quality Scores. On Bing, it might face three. Both platforms use an auction in which Ad Rank affects position and the price paid. With less competition for the same query, the click can clear for less. That is a mechanical explanation, not a verdict on the person clicking.

It also has a second consequence: there are fewer clicks to buy. One cut of the data puts Bing at 12 to 17% of US desktop search, against 8 to 9% of search overall. Desktop is where Bing has more reach. Mobile is where it has much less.

Say our client buys 3,200 Google clicks a month on that exact keyword at $6.20. On Bing, the keyword might clear at $2.40 but bring in only 400 clicks. The lower CPC does not tell us the CPA yet, and 400 clicks cannot carry a volume forecast built for 3,200. Price tells you what a click costs; auction size tells you how many you can buy.

Three bidders in an auction room facing rows of empty chairs

Next, the searcher changes what that click is worth

A keyword is not a person. Bing skews older, wealthier and more desktop-heavy than Google. The mix I plan around is ages 35 to 65, averaging near 45, and roughly 65% desktop against 35% mobile. Google runs closer to the reverse device split. Some Bing traffic arrives through Edge defaults, corporate Windows installs and the Windows search box rather than a deliberate choice of search engine.

Now put our software buyer into that picture. Someone searching project management software at a desk on a work machine may be a better fit for a B2B offer than someone making the same query on a phone. That is a reason to test, not a conversion guarantee. For a mobile app install or a late-night impulse purchase, the same skew may work against the advertiser.

This is why I do not take Google’s conversion rate, multiply it by cheaper Bing clicks and call the result a forecast. The device and workday mix changes the pool of people behind the keyword. Match the searcher to the offer before celebrating the CPC.

Then, split the traffic before judging performance

Not every Microsoft Ads click comes from Bing itself. Owned-and-operated search includes Bing, Yahoo and AOL; syndicated partners extend reach through sites such as DuckDuckGo, Swagbucks and Qwant. Those are different pipes feeding the same account. Reward searches, toolbar queries and unusual placements can behave differently from core search, even when the keyword looks familiar.

Two labeled pipes, one clean and one murky, feeding the same bucket

Microsoft lets advertisers choose Owned-and-Operated only or Owned-and-Operated plus syndicated traffic. Since July 2024, they cannot set a campaign to syndicated-only. That still gives our B2B client a clean way to learn: start owned-only, establish a baseline, then add syndicated reach as a separate test.

The comparison needs to reach past clicks. If the owned traffic produces qualified pipeline and the extra partner traffic does not, a blended CPC hides the problem. If partners do produce pipeline at an acceptable cost, keep them. Separate the pipes before you judge the water.

The identity layer changes which clicks you bid harder for

Google can respond to the query and its surrounding signals. Microsoft has an additional B2B lever: LinkedIn profile data. Advertisers can adjust bids by Company, Industry and Job Function, with Job Seniority added June 16, 2026. On Search, that targeting is bid-only. It can raise a bid for a matching profile; it does not exclude everyone else.

I used to describe this as a filter. It is not. It is a price control layered over the keyword. Our client can keep bidding on project management software while bidding more aggressively when the profile fits a target company, industry or role. Everyone else can still enter the auction at the base bid.

A common B2B setup uses high-intent exact keywords with +50% to +200% modifiers for valuable profiles. If the base bid is $2.40, a +100% modifier makes it $4.80 for a matching profile. It does not mean every matched click costs $4.80, or that cost per SQL automatically falls. The test is whether the additional spend buys enough qualified pipeline to justify it. Buy the keyword, then decide which identities warrant a higher bid.

Copilot changes where the ad meets the question

The next layer is placement. Search, Shopping, Multimedia and Performance Max can be eligible for Copilot ads without a new campaign. An ad may appear below an AI answer with a Sponsored label, matched to a conversational question rather than only to a short search query.

For our software client, a person asking an assistant to compare project tools gives more context than a person typing project management software. That context may help an existing asset match the question. One tracker reports 73% higher CTR and 16% stronger conversion than traditional search, though those figures are not a forecast for this client.

The search-ad analogy breaks here: you can choose a keyword, but you cannot choose the user’s prompt. You can control the assets available to match it. A specific commercial page gives the placement more to work with than a vague homepage. Treat Copilot eligibility as another context for the same offer, not a separate pool of guaranteed demand.

Why a Google import starts drifting

Once those layers are visible, the copy-and-paste problem is easier to diagnose. An import moves structure quickly; it does not make Google’s volume, traffic mix or bid economics follow it across. Google-sized budgets, automated bidding that does not translate cleanly, scheduled imports that overwrite Microsoft edits, overbidding and missing extensions can all undermine the copied account.

Microsoft also has its own allocation rules. Since December 2025, exact-match Search keywords have absolute priority over Performance Max for the same query, regardless of Ad Rank. A structure that relies on PMax to cover certain queries on Google can distribute them differently on Microsoft when an exact keyword is present.

A photocopied blueprint curling over a different, misaligned machine

Picture our client importing a $650 daily Google campaign into a Microsoft account with far fewer available queries. The daily cap no longer describes a comparable opportunity. Bids built for a crowded auction may be too high for a thin one. Then a scheduled import can wipe out the bid cuts made the day before. None of that makes importing useless. It makes importing a starting point. Import the structure once; rebuild budgets, bids and exact-match coverage for Microsoft.

Forecast the B2B client from the mechanics, not the discount

Here is how I would put the layers together before an agency promises this $20k client anything. Start with Google exact search terms that already produce SQLs. Then check where those conversions happen. If 70% come from desktop, Bing might plausibly add 10 to 15% volume at similar intent. If 70% come from mobile, the forecast drops to low single digits: Bing holds 12.82% of US desktop search but 1.81% of mobile. That device split does more work than a headline CPC comparison.

Then price the available traffic without confusing a model with a result:

  1. Start with proven search terms, not every Google keyword that generated a click.
  2. Estimate the smaller reachable pool from the client’s device mix and Bing’s thinner volume.
  3. Model the base CPC, then account for higher bids on valuable LinkedIn profiles. If $6.20 becomes an estimated $3.10 base CPC, a +50% bid modifier makes $4.65 the modified bid in this simplified example, not a promised click price.
  4. Start owned-and-operated only. Measure cost per SQL and pipeline before testing syndicated reach.
  5. Make suitable assets Copilot-eligible, then judge the resulting traffic on the same business outcomes.

This forecast may still be wrong. Its advantage is that you can see where it is wrong: volume, searcher fit, partner quality, profile value or placement. “Bing clicks are cheaper” gives you nowhere to look when the SQLs fail to arrive.

The partner test: do they rebuild or merely import?

If an agency owner asks me how to choose a white-label Microsoft Ads partner, I ask four questions before opening the pricing page:

  • Do they separate owned-and-operated traffic from syndicated reach at the start?
  • Do they reset budgets for Microsoft’s volume rather than copying Google’s daily caps?
  • Do they use LinkedIn bid modifiers where Company, Job Function or other available profile signals fit the offer?
  • Do they build exact-match coverage knowing exact now takes priority over Performance Max on Microsoft?

A partner whose whole plan is “keep both platforms in sync” has not answered those questions. Syncing can preserve the very mismatch the client hired them to fix. The wasted spend is less dramatic than a broken campaign; it arrives as partner clicks nobody separated, bids nobody reset and a monthly report that calls the lower CPC a win.

That is why I run Bing inside groas Search Ads rather than treating it as a copied side account to babysit. The service covers Google, Bing and ChatGPT Ads for one flat fee, with bids, budgets and keywords worked around the clock rather than reviewed once a month. The operating model here is the one I have just laid out: establish an owned-only baseline, use LinkedIn modifiers where the B2B economics support them, build exact coverage around Microsoft’s priority rules and keep suitable assets eligible for Copilot. I have also ranked Microsoft Ads management options by whether they run Bing or copy Google and examined who should actually run your Microsoft Ads.

Hire the operator who rebuilds for the thinner auction, not the one who mirrors the thicker one faster.

One more client, one different answer

Take a garage door company spending $8k a month on Google. Eighty percent of its booked jobs come from mobile calls after 5pm. Run the same model without changing its rules.

  1. Auction and volume: The thinner auction may lower CPC, but Bing’s small mobile share limits how many relevant emergency searches it can supply.
  2. Searcher mix: The workday desktop skew points away from a homeowner calling after hours.
  3. Traffic source: Syndicated reach could add clicks, but those clicks would need to prove they become booked jobs, not just make the account look busier.
  4. Identity: LinkedIn profile bidding offers little help for a homeowner with a stuck door.
  5. Placement: Copilot may suit someone comparing opener brands better than someone who needs help at 9pm.

My call is to skip Bing for that client and put the test budget into call coverage. The keyword may still be cheaper on Microsoft. That does not make it the right purchase. For the B2B client, desktop reach and profile bidding give the smaller auction a plausible job to do. For the garage door company, the same mechanics work against the demand it needs to capture. No CPC discount fixes a mismatch between the traffic you can buy and the buyer you need.