Performance Max for Microsoft Ads - Done Properly
About this service
Microsoft rolled out Performance Max (PMax) for Microsoft Ads in 2024 - the all-channel, asset-group-based, machine-learning-driven campaign type that also exists on Google Ads. On Microsoft it converts well for some accounts and stalls for others, most often because the asset groups were built too broad, the audience layer carried no exclusions, or PMax was allowed to cannibalize the existing search campaigns. This retainer is for buyers who want PMax set up properly on Microsoft: isolated against cannibalization, asset groups structured around real product and intent clusters, audience exclusions applied, and a real Target CPA driving the bid strategy instead of the platform default. Honest disclosure upfront: PMax on Microsoft is newer than Google's PMax, has smaller inventory (Microsoft Search plus Microsoft Audience Network), and is not always the right campaign type for every account. The 4.3-star average on this listing reflects that unevenness - some clients have seen 25-40% lower CPA than their equivalent Google PMax on B2B, finance, and legal offers; others saw flat volume and we moved them back to manual search plus Audience Ads.
This is management only. You bring the active Microsoft Ads account, the spend budget paid by you directly to Microsoft, and the creative assets. I run the strategy, the asset-group architecture, the audience exclusions, the bid-strategy calibration, the negative-keyword isolation, and the steady optimization. Your money stays in your account, building your history, leaving with you when you leave.
1. What PMax on Microsoft actually is - and is not
Before hiring this engagement, it is worth stating plainly what Performance Max for Microsoft Ads is and is not, because most accounts I inherit have been set up against incorrect assumptions. The table below maps the dimensions that matter.
| Dimension | PMax on Microsoft (2024 launch) | PMax on Google (reference) |
|---|---|---|
| Inventory surfaces | Microsoft Search + Microsoft Audience Network | Search, Display, YouTube, Discover, Gmail, Maps |
| Audience targeting | LinkedIn profile targeting integration (unique to Microsoft) | Customer Match, similar audiences |
| Asset-group structure | 1 campaign → N asset groups | 1 campaign → N asset groups |
| Typical volume ceiling | Lower than Google PMax on most query sets | Highest of any single Google campaign type |
| Typical CPC range | USD | USD |
| Typical CPA vs Google PMax (B2B, finance, legal) | 25-40% lower at parity budgets | Baseline |
| Maturity | Newer (2024), less battle-tested | Mature (2021) |
| Cannibalization risk | High if not isolated by brand/exact-match negatives | Same risk, same fix |
The honest takeaway: PMax on Microsoft is not a strict upgrade over manual search plus Audience Ads. It is a different campaign type with different strengths. For accounts with high-intent B2B, finance, legal, or automotive offers, PMax on Microsoft frequently delivers lower CPA than equivalent Google PMax. For accounts that need volume at scale, or where Microsoft inventory is thin for the query set, PMax on Microsoft underperforms manual search plus Audience Ads. The first job of this engagement is to determine which category your account is in - before any PMax campaign is built.
2. What you receive
Phase 1 - Fit assessment and PMax build (days 1-7):
- A written fit-assessment matrix scoring your account on six dimensions (query intent, asset availability, audience-match strength, budget level, conversion-data integrity, competitive density) with a clear go or no-go recommendation - and a refund of month one if the recommendation is no-go
- UET tag repair where applicable: goal configuration, Enhanced Conversions for Web, server-side event validation, deduplication, documented test-event flow - because every PMax decision depends on honest conversion data
- PMax campaign architecture: one campaign, two to five asset groups structured around real product or intent clusters (not "all products in one group" - the most common error I see)
- Asset-group build: headlines, descriptions, images, logos, videos where available, and long-headlines, each pinned to the asset-slot rules Microsoft actually rewards; minimum 5 headlines, 2 descriptions, 4 images, 2 logos per group
- Brand and exact-match negative-keyword isolation across the account, so PMax does not cannibalize your existing brand and exact-match search campaigns - the single most common cause of "PMax stole my search conversions"
- Audience exclusions at the campaign and asset-group layer: competitor audience lists, low-intent remarketing segments, and geography and language exclusions as appropriate
- Bid strategy calibration: PMax launched with Target CPA at the reverse-engineered ceiling, computed from your AOV, gross margin, and contribution-margin inputs - not on Maximize Conversions with no ceiling, which is the platform default and the second most common cause of waste
- Day-7 baseline document: launch state, asset-group inventory, isolation map, target CPA versus Microsoft's suggested CPA, and the 30-day plan
Phase 2 - Ongoing PMax management (from month 2):
- Weekly asset-group performance review: which groups are converting, which are spending without return, with a Friday note
- Audience-exclusion maintenance: new competitor segments added, low-intent remarketing pruned
- Creative refresh cadence: 2-4 new assets per asset group per month (the Microsoft PMax learning model rewards fresh creative - fewer assets than that and the model decays)
- Monthly A/B test on one asset-group variable (headline set, image set, audience exclusion, or final URL), with hypothesis and contribution-margin threshold pre-registered in writing
- Monthly report in plain English: spend, conversions, CPA, asset-group-level breakdown, contribution margin per conversion, and the three actions I recommend next
- Account health monitoring - disapprovals, policy flags, UET outages, asset rejections flagged within one business day
- Quarterly strategy call (30 minutes) to re-evaluate whether PMax on Microsoft is still the right primary campaign type for the account - and to recommend pivoting back to manual search plus Audience Ads if the data indicates it
3. The four pitfalls I correct before any PMax campaign goes live
These are the four errors I see most often in PMax implementations on Microsoft, and the engagement is structured around preventing each one at build.
| Pitfall | Symptom in the account | Fix applied in Phase 1 |
|---|---|---|
| 1. PMax cannibalizing existing search | Brand and exact-match query volumes drop 20-50% the week PMax launches; CPA looks better but total conversions flat | Negative-keyword isolation: brand terms and high-intent exact-match keywords added as account-level negatives to PMax; PMax budget capped relative to search |
| 2. Asset groups built too broad | All products or all intents dumped into one asset group; the learning model cannot optimize because every signal is mixed | 2-5 asset groups structured by real product or intent cluster, each with its own creative set and final URL |
| 3. No exclusions on the audience layer | PMax spending on competitor-name remarketing, low-intent lookalikes, or out-of-geography impressions | Audience exclusions at campaign and asset-group layer; competitor segments excluded; geography and language scoped tightly |
| 4. Default bid strategy without CPA target | Maximize Conversions running with no ceiling; spend paces to budget, not to efficiency | Target CPA at the reverse-engineered ceiling, computed from your AOV and contribution margin; never the platform default |
4. How the engagement runs
- Day 1: Access and inputs. You grant read-and-analyze access (or partner access if you prefer; billing stays yours either way). I send the structured margin-input questionnaire.
- Days 2-3: Fit assessment. I run the six-dimension fit matrix on your account, query set, and assets. If the recommendation is no-go, month one is refunded and I recommend the alternative engagement.
- Days 4-6: Build. PMax campaign and asset groups built, isolation negatives applied, audience exclusions applied, bid strategy set at the reverse-engineered Target CPA.
- Day 7: Launch and baseline. First learning period begins. Day-7 baseline document delivered.
- Month 2 onward: Steady optimization. Weekly asset-group review, monthly creative refresh (2-4 new assets per group per month), monthly A/B test, monthly report, quarterly strategy call.
5. You bring / I run - the division of responsibility
This is a management-only engagement, and the division is deliberate.
You bring:
- The active Microsoft Ads account, with billing you control
- The monthly ad spend budget, paid by you directly to Microsoft - I never bill or fund media
- Access to your website, analytics, and any landing-page tooling needed to verify tracking
- AOV, gross margin %, refund rate, and (if available) contribution margin per conversion - or a finance or ops lead who can sit with me for 30 minutes to derive them
- Creative assets you already have (images, logos, videos, headlines) - I can write copy but I do not shoot your product photography
- A 48-hour response habit on approvals and questions - the single biggest predictor of a smooth engagement
I run:
- Strategy, asset-group architecture, audience exclusions, bid-strategy calibration, creative testing
- UET configuration and measurement hygiene
- The PMax-specific reporting layer (asset-group-level breakdown, cannibalization-isolation monitoring, contribution-margin column)
- Communication and the steady work of making PMax perform - or recommending a pivot away from PMax if the data warrants
6. Who this engagement fits - and who it does not
A strong fit if:
- You spend between USD 1,500 and USD 20,000 per month on Microsoft Ads
- Your offer is B2B, finance, legal, automotive, or any high-margin product or service where Microsoft's older, affluent, B2B-skewed user base is structurally good
- You have at least 30 conversion events in the last 60 days (PMax needs conversion data to learn; below 30, manual search plus Audience Ads outperforms)
- You want reporting that shows contribution margin, not just CPA
- You can act on recommendations: approve changes, answer questions, keep the site live
A poor fit if:
- Your monthly budget is under USD 1,200 - PMax cannot learn at that volume, and I will say so before you pay
- Your conversion volume is below 30 in 60 days - PMax's machine learning has insufficient signal; the engagement will recommend manual search plus Audience Ads instead
- You need volume at scale on a query set where Microsoft inventory is thin (some consumer ecom categories) - PMax will underperform
- You expect a guaranteed ROAS figure - nobody honest guarantees specific returns on PMax on any platform, especially on the newer Microsoft variant
- You want spend managed on your behalf from my accounts - that is a different service model entirely
7. Reporting and communication
A concise weekly Friday note - what changed, what it did - and a monthly report with the asset-group-level breakdown, contribution margin per conversion, and three recommended next actions. One 30-minute strategy call per quarter. Communication via the platform's message system; I respond within one business day, and urgent items (asset disapprovals, tracking outages, spend anomalies) are flagged proactively, not discovered by you first.
8. Why this engagement
Three points worth stating plainly. First, the fit assessment comes first: I will not run PMax on an account where the data indicates PMax will underperform - and the engagement refunds month one if the fit matrix returns no-go. Second, the four-pitfall framework in section 3 is what separates a working PMax implementation from a wasted one; each pitfall is prevented structurally in Phase 1, not discovered later. Third, you own everything - the account, the asset-group architecture, the creative assets, the negative-keyword isolation corpus. If we ever part ways, you keep a cleaner and more honestly-instrumented account than the one you brought me.
What is included
- Base Bing Ads campaign · Your ad account, your spend
- Revisions · 2 rounds included
- Delivery · 7 to 9 days
Frequently asked questions
Industries served
- B2B Industrial Advertising
- Personal Finance
- Law Government and Politics
Search and social, mostly Google and Bing
- Completed orders
- 267
- 1 in the last 30 days
- Rating
- 4.894 reviews
- Active listings
- 7
- Ad platforms
- 6
More from Ben Harris
View storefrontGoogle AdsGoogle Shopping Feed Rebuild - Merchant Center Optimization Done Right7 to 9 days$299.00
Quora AdsI Will Manage Your Quora Ads for B2B SaaS Pipelines5 to 7 days$199.00
Discord AdsI Will Revive Your Dead Discord Server and Bring It Back to Life5 to 7 days$139.00
Google AdsDone-For-You Google Ads Launch - Ad Spend Included, Zero Setup Stress5 to 7 days$749.00