Tag: Positive Sparks News
Ever feel like Google’s Performance Max gets all the hype… but your next best customer is actually browsing (and buying) somewhere else? What if we told you Microsoft Ads can be one of the cleanest ways to find net-new customers, without fighting in the same crowded auction pool all day?
Microsoft’s Performance Max (PMax) gives e-commerce brands an AI-powered way to show up across the Microsoft Advertising Network and convert buyers in real time. And here’s the kicker: Microsoft’s own reporting suggests Performance Max can drive 3x higher ROAS and a 32% lower CPA compared to running Search ads alone (results vary, but the direction is hard to ignore).
So how do we actually use Microsoft Ads PMax to target new customers, without burning budget, confusing the algorithm, or measuring the wrong thing?
Let’s walk through the 5-step playbook we use with growth-minded e-commerce teams.
Step 1) Are we clear on what Microsoft PMax is really doing (and why it finds new customers)?
Are you thinking, “Isn’t PMax just another black box?” Fair. But bear in mind: it’s a goal-driven system, not a channel-specific campaign.
So what does Microsoft PMax actually optimise across?
Microsoft PMax is designed to find and convert customers across the Microsoft Advertising Network, using automation to decide:
- Where your ads show (multiple surfaces, placements, contexts)
- Who sees them (based on signals and predicted intent)
- Which creative combination gets served (assets are mixed and matched)
- When to bid and how much (automated bidding strategies)
What makes it good for new customer targeting?
Because it’s not stuck in the “capture the last click” mindset. If we give it the right inputs (signals + conversions + values), it can:
- Expand reach beyond pure branded search
- Identify patterns among high-LTV customers
- Find similar buyers earlier in their decision journey
Actionable takeaway: Before touching campaign settings, write a one-sentence goal:
“We want Microsoft PMax to acquire new customers at or below £X CPA (or at £X ROAS), while protecting profitability.”
If we can’t define “new customer” and “profitable” in plain English, the AI can’t magic it into existence.
Step 2) Are we tracking the right conversions (or are we training the AI to chase noise)?
Let’s face it: most PMax problems are measurement problems wearing a “platform issue” disguise.
What’s the minimum tracking setup we need?
Microsoft’s foundation is Universal Event Tracking (UET):
- Create a UET tag in Microsoft Ads
- Install it sitewide (ideally via GTM)
- Configure conversions for the key actions (purchase at minimum)
If you’re using Maximize Conversion Value, you’ll also need revenue/value parameters flowing correctly, otherwise you’re telling the algorithm “all orders are equal,” which is rarely true in e-commerce.
Primary vs secondary conversions (this matters)
Are you currently feeding PMax micro-conversions like “Add to Cart” as a primary goal? That’s a common leak.
A clean structure:
- Primary conversion goal: Purchase (with correct revenue)
- Secondary conversions: Add to Cart, Begin Checkout, Email signup (helpful for diagnostics, not for bidding)
A quick real-world example
We worked with an e-commerce brand that was “crushing it” on reported CPA, until we noticed the campaign was optimising heavily for Begin Checkout. Purchases were flat. Once we set Purchase as the primary conversion and demoted the rest to secondary, volume initially dipped for a week (learning period), then rebounded with higher conversion value per click.
Actionable takeaway checklist:
- UET installed and firing on all pages
- Purchase conversion is firing once per order (no duplicates)
- Revenue values match your backend (spot-check 10 orders)
- Purchase is Primary; micro events are Secondary
If you want help validating tracking end-to-end, this is exactly the kind of thing we handle as part of performance + measurement work at Positive Sparks: https://positivesparks.com/google-analytics-agency
Step 3) Do we have enough high-quality assets (or are we starving the machine)?
Are you expecting PMax to “just work” with two headlines and a logo? You’re not alone. But PMax is creative-hungry. The algorithm needs variety to match the right message to the right shopper.
Microsoft PMax uses asset groups (not ad groups)
Think of an asset group as a mini creative library + audience intent pack:
- Headlines
- Descriptions
- Images
- Logo(s)
- Audience signals (we’ll cover this next)
- Optional: search themes
You can create up to 100 asset groups per campaign, but most e-commerce brands do well starting with 3–8 asset groups based on product categories, margins, or customer intent.
What “rich assets” look like in practice
For each asset group, aim for:
- 8–12 headlines (mix benefit-led + offer-led + category-led)
- 4–6 descriptions (different angles)
- 10+ images (lifestyle + pack shots + UGC-style)
- 1–2 logos (light + dark versions)
And yes: enable auto-generated assets and final URL expansion when it makes sense. If your site experience is strong and your product pages convert, this can unlock extra scale. If you’ve got messy collections or out-of-date landing pages, keep a tighter leash.
A simple creative framework that sells (without sounding salesy)
When we build assets for “new customer” growth, we rotate these angles:
- Problem → solution (“Struggling with X? Here’s Y.”)
- Outcome (“Get results in weeks, not months.”)
- Social proof (“Trusted by 10,000+ customers.”)
- Offer (“Save 15% on your first order.”)
- Differentiator (“Clean ingredients / 2-year warranty / made in the UK.”)
Actionable takeaway: Create one asset group that’s purely “best sellers for first-time buyers” and make the messaging beginner-friendly. New customers don’t need the full catalogue: they need confidence.
Step 4) Are we giving strong audience signals (so Microsoft can find new customers faster)?
Are you worried audience signals will “box in” the algorithm? In Microsoft PMax, signals are a starting line, not a cage.
What signals should we feed the system?
Use a blend of:
- Customer Match lists (past purchasers, subscribers, high-LTV segments)
- Remarketing lists (site visitors, cart abandoners)
- In-market audiences (people actively researching relevant categories)
- Custom audiences (based on URLs, interests, intent cues)
Even when the goal is new customers, existing customer data is gold: because it helps the model learn what “good” looks like, then expand to similar people.
Don’t skip search themes
You can add up to 25 search themes per asset group. These are essentially “here’s how shoppers describe what we sell.”
Why they matter:
- They can reduce the learning period
- They help Microsoft connect your products to real-world intent faster
- They act like guardrails when your category is broad
Example search themes for an e-commerce wellness brand:
- “plant-based protein powder”
- “low sugar electrolyte drink”
- “supplements for sleep”
- “best magnesium glycinate”
Quick win: build a “new customer” signal set
Ask yourself:
- What does a first-time buyer search before they trust a brand like ours?
- What adjacent categories do they browse?
- What competitor brands do they consider?
Then feed that into:
- Custom audiences (competitor URLs)
- Search themes (category + benefit phrases)
- In-market audiences (category shoppers)
Actionable takeaway: Start with 1–2 strong audience clusters per asset group (not 10). Too many mixed signals can slow learning.
Step 5) Are we using the right bidding + budget controls (and optimising like a grown-up)?
Do you want more orders… or more profitable orders? The bid strategy decides what the machine values.
Choosing the right bid strategy
Microsoft PMax typically offers:
- Maximize Conversions (optimises for volume)
- Maximize Conversion Value (optimises for revenue)
For most e-commerce brands, Maximize Conversion Value is the better starting point: if revenue tracking is clean.
From there, we can layer:
- Target ROAS (when we have stable conversion volume)
- Target CPA (more common when optimising for leads; less ideal for e-com)
Budget: the underrated bottleneck
Microsoft PMax uses individual budgets (not shared budgets). That matters because you can’t “float” spend across campaigns as fluidly.
Watch for the status: “Performance limited by budget.”
If you see it and performance is strong, a controlled budget increase (say 10–20% per week) is often the simplest growth lever.
Optimisation rhythm (what we actually do weekly)
Instead of random tinkering, use a steady cadence:
Weekly:
- Check conversion value, ROAS, CPA, new vs returning split (if available)
- Review asset performance (swap weak headlines/images)
- Add 3–5 new creatives per top asset group (freshness helps)
Fortnightly:
- Refine search themes (keep winners, replace vague themes)
- Review product/landing pages PMax is expanding into (if URL expansion is on)
Monthly:
- Decide: scale budget, tighten ROAS, or spin out a separate campaign for a category with different margins
Actionable takeaway: Give major changes (bid strategy shift, big asset rebuild) 7–14 days before judging results: PMax needs learning time.
“But how do we know it’s really new customers?” (The measurement reality check)
Are you worried PMax is just taking credit for customers who would’ve purchased anyway? That’s a valid fear.
Here’s how we reduce that risk:
- Use new customer-focused creative (“First order offer”, “New here?” messaging)
- Separate brand vs non-brand reporting where possible (directionally)
- Pair platform reporting with GA4 + attribution tooling so we don’t rely on one view of the truth
If you’re serious about scaling new customer acquisition, investing in better attribution pays for itself fast. We’ve written about that approach here: https://positivesparks.com/unlock-the-power-of-3rd-party-attribution-apps
Where Positive Sparks fits (and why our commission-based model helps)
Do you ever feel like agencies win when spend goes up… even if your profit doesn’t? Same. That’s why we built Positive Sparks to align incentives.
Our commission-based model is designed so we’re motivated by what you’re motivated by: measurable performance. When we help you grow efficiently: especially with new customer acquisition: we’re rowing in the same direction.
If you want to see how we structure fees (no fluff), it’s here: https://positivesparks.com/how-our-fees-work
Your 5-step Microsoft PMax “new customer” checklist (copy/paste this)
- Define the goal: new customers at a profitable CPA/ROAS
- Fix measurement: UET + purchase as primary + accurate revenue
- Build rich asset groups: enough creative variety to scale
- Feed smart signals: customer match + in-market + search themes
- Optimise with intent: correct bidding, controlled budgets, weekly cadence
What are you going to try first?
Are you already running Microsoft Ads and wondering if PMax could unlock a new pocket of demand? Or are you starting from scratch and want a clean setup from day one?
Tell us:
- What’s your biggest blocker right now: tracking, creative, or scaling profitably?
- And if you’re testing Microsoft PMax, are you optimising for conversion volume or conversion value?
If you want us to pressure-test your setup (and spot the leaks), book a quick chat here: https://positivesparks.com/speak-to-us
