WORX: Reading an Intent-Led Advertising Case

Put the reported result in context

This is an analysis of a public Microsoft Advertising case, not a Mangione Group client engagement. The June 2026 story describes WORX using Performance Max, CRM inputs, automated bidding, and creative optimization for an unnamed client. Microsoft reports a 420% year-over-year increase in quality leads and a 77% decrease in cost per acquisition. The public account does not fully disclose baseline counts, spend, qualification criteria, or a controlled comparison. These are vendor-reported outcomes for one client, not independently audited causal estimates or a forecast for another advertiser. Microsoft Advertising: WORX Intent-Driven Advertising Case Study

The interesting business question is how an advertiser teaches its campaign what a valuable lead means. A system can optimize an observed action, but the action may be a poor representation of the outcome the business actually needs. If the website records every form submission as equally valuable, the reporting can improve while the sales team becomes busier with unsuitable requests. Better automation starts with a better definition of success.

This article builds an original evaluation framework around that problem. It explains how to read a success story, design quality feedback, review creative and landing pages, and pilot a change without overstating the evidence. The hypothetical examples are not claims about WORX's undisclosed client. They illustrate decisions another business would need to make before treating the case as a reason to change its own advertising program.

Translate percentage claims into questions

Large percentage changes are informative only when the underlying quantities are understood. A quality-lead increase could represent a small absolute change from a low baseline or a substantial increase in an established program. A cost reduction could reflect better performance, different conversion definitions, a changed mix of campaigns, or several factors together. A public case may be accurate while still lacking the detail needed for a forecast.

Ask for the unit, denominator, comparison period, and population behind each claim.

• What counted as a lead in each period?
• Were duplicates removed?
• Did the qualification rule change?
• Was cost limited to media spend, or did it include creative, software, and management?
• Were results compared across the same markets and offers?

These questions are not objections to a success story. They are the bridge between inspiration and a responsible business decision.

Be careful with arithmetic. A 100% increase means twice the baseline quantity, while a 100% index simply represents the baseline itself. An indexed chart can explain a reported percentage change without inventing absolute counts, but it must be labeled clearly. It should not look like a chart of actual leads if those counts were not published. The visual in this article uses that distinction explicitly.

Most importantly, do not put a public result into a budget model as an expected multiplier. Use it to identify a mechanism worth investigating, then build a model from your own costs, volume, sales process, and uncertainty. The case can justify asking a better question. It cannot supply the answer to a different company's economics simply because the campaign type has the same name.

Define quality in terms sales can apply

A useful quality definition identifies why a request is worth pursuing. It might consider service fit, location, project scope, timing, authority, or a confirmed need. The exact criteria depend on the business. The definition should be specific enough that two reviewers can apply it with reasonable consistency, while allowing a place for uncertainty rather than forcing every record into a premature yes or no.

For a hypothetical implementation firm, an eligible lead might be a real business seeking a supported service within the company's delivery area. A qualified opportunity might require a completed conversation confirming a relevant project and an agreed next step. These stages should not be collapsed. The first can often be judged from the inquiry; the second requires information that becomes available later.

Write examples of accepted, rejected, and uncertain records. A request for a service the company does not offer is different from a relevant project that is not ready this quarter. A duplicate inquiry is different from a new inquiry by another person at the same organization. A spam submission is different from a legitimate person who lacks sufficient information. These distinctions help the team improve marketing rather than simply reject records without learning from them.

Review disagreements between salespeople. If one person marks a lead unqualified because they did not answer a call and another keeps the same kind of lead open, the feedback is inconsistent. The campaign may then receive labels reflecting staff habits instead of buyer quality. A modest, shared qualification standard is a prerequisite for meaningful feedback, regardless of how sophisticated the advertising technology becomes.

Understand what Performance Max is optimizing

Microsoft describes Performance Max as a campaign type that uses business goals, creative assets, audience signals, conversion information, and automated optimization across its inventory. It also states that audience signals are starting inputs rather than limits restricting delivery only to the supplied audience. Microsoft Advertising: Performance Max That distinction matters when an advertiser hears the phrase intent-driven targeting.

An audience signal does not mean the platform has identified a particular person who definitely intends to buy your service. It is an input into a prediction and delivery system. The business still needs an appropriate offer, accurate creative, suitable destinations, and outcome feedback. The system's sophistication does not remove uncertainty about an individual buyer or guarantee that every impression reaches someone ready to act.

Ask the campaign manager to explain the controls in plain language. Which settings are firm restrictions, which are suggestions, and which are optimization goals? Which actions count toward bidding? Which placements or destinations are eligible? Which parts can change automatically? This explanation should be specific to the chosen platform and campaign configuration, not borrowed loosely from another product with a similar name.

Then decide whether that operating model fits the business's readiness. A company with unclear services, unreliable forms, and no consistent lead disposition may need foundational work first. A company with clear offers and dependable quality records may be better positioned to evaluate a broader automated campaign. Readiness is not a judgment about company size. It is a practical assessment of whether the inputs and feedback can support the intended decision.

Build the conversion definition before the dashboard

Microsoft's Universal Event Tracking supports configured website conversion goals. Microsoft Advertising: Conversion Tracking The commercial task is choosing which actions should be observed and which should influence optimization. A page visit, resource download, form submission, and qualified inquiry can all be useful measurements, but they should not automatically carry the same meaning or value.

Start with the desired business outcome and work backward to observable events. If the goal is a suitable appointment, identify what makes an appointment suitable and where that information becomes available. The website may know only that a booking occurred. The CRM may later know whether it was accepted, attended, and relevant. Both observations can be useful if their roles are clearly defined.

Create a measurement dictionary before implementation. For every event, write the plain-language meaning, trigger, source system, counting rule, and owner. Include examples of events that should not fire. A form validation error should not count as a successful inquiry. A thank-you page refresh should not create a new person. A test submission should be distinguishable from live demand. These details determine whether a dashboard represents reality.

Review the dictionary with the people who use the report. An analyst may interpret conversion as a recorded technical event, while a sales director hears a completed commercial outcome. Neither interpretation is inherently wrong, but the mismatch can lead to poor decisions. Use names that make the stage clear and keep the relationship between early actions and later outcomes visible.

Connect later outcomes without erasing their timing

Microsoft documents offline conversion goals, including count and bidding-related configuration. Microsoft Learn: Offline Conversion Goal Such mechanisms can connect later business outcomes with advertising measurement when implemented appropriately. The presence of an import feature, however, does not guarantee that the business has trustworthy outcomes to import or that every record can be matched.

Decide when a lead becomes qualified and preserve the time that event occurred. Do not rewrite history by treating a later qualification as if it happened at the original form submission. The distinction matters when evaluating recent campaigns whose leads have not had time to progress. A new campaign can look weak merely because its cohort is younger than the one used for comparison.

Define who updates the qualification record and how quickly. A weekly manual review may be adequate for some programs, while others need more frequent operational feedback. The business should understand the delay and interpret reports accordingly. Missing disposition is not the same as rejection. If staff have not reviewed a lead, the record should say unknown or pending rather than silently becoming low quality.

Keep the feedback loop useful for people as well as the platform.

• A rejection reason such as wrong service can inform the landing page and creative.
• A reason such as outside coverage can reveal a targeting or page-clarity issue.
• A reason such as timing uncertain may suggest a nurture path.

Quality feedback is most valuable when it improves the whole acquisition process instead of disappearing into an automated bidding setting.

Prevent duplicate measurement across systems

Microsoft's Conversions API documentation describes shared event identifiers for deduplicating browser and server records of the same conversion. Microsoft Learn: Conversions API Integration This illustrates an important measurement risk: adding another tracking path can increase reported events without increasing business activity. More complete instrumentation should not mean counting the same action twice.

Map which systems send each event. A website tag, form platform, CRM integration, and server process may all be capable of reporting a submission. Decide which route is authoritative or how equivalent records are matched. The implementation should be documented well enough that a later integration does not accidentally create a second independent count of the same event.

Test a complete record from click to disposition using approved test procedures. Inspect the event time, identifier, stage, and source. Then test a repeated page load and a duplicate submission. Compare the reporting with the underlying business record. This is especially important before changing bidding objectives, because an inflated event count can become an input to automated decisions rather than merely an inaccurate chart.

Monitor unexpected changes after releases. A sudden conversion increase coinciding with a form replacement deserves investigation before celebration. It may represent a real improvement, but it may also reflect a changed trigger. Keep technical release notes beside campaign changes so the team can distinguish market behavior from measurement behavior. A reliable program makes that distinction routine rather than embarrassing.

Treat first-party data as a responsibility

The phrase first-party data describes a relationship to the collection process; it does not eliminate obligations about use, access, disclosure, or retention. A CRM may contain useful commercial information, but not every field belongs in an advertising integration. Decide what information is necessary for the intended purpose and review how the chosen tools handle it.

Microsoft's consent guidance describes signals that affect tag behavior for relevant users and regions. Microsoft Advertising: Providing User Consent Signals Technical settings should be aligned with the business's actual privacy choices and applicable requirements. Installing a consent feature is not a substitute for understanding the data flow. The team needs to know what is collected, which systems receive it, and how a changed preference is respected.

Separate operational notes from optimization inputs. A salesperson may record sensitive context that helps serve a customer but has no appropriate role in an audience upload. Broadly exporting CRM fields because they are available creates unnecessary risk and complexity. A focused integration with documented purpose is easier to explain, maintain, and review.

Assign ownership for access and changes. If an agency, software vendor, and internal team all touch the data flow, clarify who can modify it and who verifies the result. Review access when roles or vendors change. The goal is an effective feedback system with a clear boundary, not a growing collection of exports whose purpose and recipients become difficult to reconstruct.

Distinguish audience information from certainty

Microsoft offers audience targeting options based on different kinds of information. Microsoft Advertising: Audience Targeting For a business evaluating intent-led advertising, the important question is what each input actually represents. A past customer list, a recent website visit, and an inferred interest are different kinds of evidence. They should not be described as equally strong proof of a current buying decision.

Use that distinction in creative planning. Someone encountering the company for the first time may need a clear explanation of the service. A returning visitor comparing options may need a practical distinction or credible evidence. A known customer may need an appropriate expansion offer. The campaign can be designed to accommodate these contexts without claiming perfect knowledge of the person behind every impression.

Also consider who is absent from the data. A customer list may overrepresent the kinds of clients the business historically acquired and underrepresent a new market it wants to serve. A campaign optimized only around past patterns might need careful evaluation when the strategy changes. This is a reason to define the commercial objective clearly, not a reason to discard historical data.

Ask for reporting that makes available audience and placement information understandable, while recognizing the limits of what the platform reveals. Do not demand a level of individual certainty the system cannot provide. Instead, combine aggregate evidence with lead-quality review and controlled changes where feasible. The business should learn whether the overall program serves its intended market, not pretend that it can read every buyer's mind.

Give automated bidding a coherent objective

Microsoft's bidding documentation describes strategies tied to selected campaign objectives. Microsoft Advertising: Automated Bidding The business implication is that the chosen objective should reflect what the company values and can measure reliably. A target expressed in the advertising interface is not meaningful if the underlying conversion represents an inconsistent or low-value action.

A hypothetical professional services firm might initially measure all inquiries while building a reliable qualification process. Once later outcomes are sufficiently consistent, it can evaluate whether a more meaningful objective is appropriate. That transition should be planned, documented, and reviewed rather than treated as a cosmetic setting change. The old and new reports may not be directly comparable if the definition changes.

Avoid assigning arbitrary monetary values merely to make a value-based strategy available. A defensible value can come from a transparent business model with stated assumptions, but it should not be confused with actual revenue. If a qualified inquiry receives an estimated value, explain the estimate and update it when evidence changes. Overconfident inputs do not become more accurate because software uses them automatically.

Give the campaign a change log and a decision cadence. Frequent simultaneous adjustments to goals, budgets, creative, and destinations make interpretation difficult. There may be urgent reasons to change several things, but the team should record them and narrow its claims about what caused the outcome. Automation needs informed oversight, not constant interference or complete neglect.

Prepare creative that can survive different placements

Microsoft's display and native advertising documentation describes visual assets used across eligible placements. Microsoft Advertising: Display and Native Ads A campaign therefore needs more than one attractive image cropped repeatedly without review. Headlines, descriptions, logos, and photographs should work in the combinations and dimensions the platform may use, while preserving a coherent offer.

Start with the message hierarchy.

• What must a viewer understand if only the headline is noticed?
• What useful detail does the description add?
• What does the image communicate about the service or situation?

Each component should be accurate on its own and compatible with the others. A strong headline paired with an unrelated image can produce attention without useful understanding.

Review combinations for unintended claims. A statement that is accurate with one description may become misleading beside another. A photograph can imply a service setting or customer type the business does not actually serve. Generated assets need the same factual and brand review as manually created assets. Efficiency in production should not reduce responsibility for what the ad communicates.

Test legibility on a phone and in small placements. Avoid relying on tiny text embedded in an image to explain essential conditions. The visual should retain its purpose when cropped within the supported formats. The brand consistency guide explains how a shared system can support variation without making the campaign look like several unrelated companies.

Make the landing destination part of the campaign

Microsoft's implementation guidance describes asset groups and configurable final URL expansion. Microsoft Learn: Performance Max Campaigns Before permitting broader destination selection, review which pages are suitable for paid visitors and which should be excluded from the intended journey. A website can contain useful information that is still an unsuitable destination for a particular offer.

Audit the eligible pages for clarity, current claims, working forms, and consistent next steps. An outdated event, a thin archive, or a service the company no longer offers can create confusion if it becomes a campaign destination. The campaign manager and website owner should agree on what is eligible rather than assume that every published page is equally ready for advertising.

Maintain continuity from the ad's promise to the page. If the ad offers a discussion about a particular service, the page should make that service recognizable immediately. If a qualification condition matters, explain it early enough to help unsuitable visitors self-select without making suitable visitors work too hard. The goal is a better conversation, not simply the largest possible form count.

The landing-page design guide explores this in more detail. For campaign evaluation, record material destination changes alongside ad changes. A new form or clearer explanation can influence the quality of inquiries independently of the audience strategy. Treat the page as an active part of the system rather than a fixed backdrop that receives traffic.

Evaluate cost at the stage that matters

Cost per acquisition is only interpretable when acquisition is defined. Cost per form submission, cost per qualified inquiry, cost per attended appointment, and cost per new customer are different measures. A program can improve one while worsening another. Choose the stage that supports the decision and keep the others visible enough to explain changes in the journey.

In a hypothetical comparison, Campaign A spends two thousand dollars and produces one hundred inquiries, twenty of which qualify. Campaign B spends the same amount and produces sixty inquiries, thirty of which qualify. A has the lower cost per inquiry, while B has the lower cost per qualified inquiry. Neither example predicts your outcome; the arithmetic shows why the denominator changes the interpretation.

Then consider costs outside media. Creative production, landing-page work, management, tools, and sales handling can affect the business case. They do not all need to be allocated perfectly to every record, but ignoring them can make an operationally expensive program appear efficient. Use a clearly stated cost view appropriate to the decision, and avoid switching views between periods without explanation.

Include capacity. If a campaign creates more qualified conversations than the team can serve promptly, the immediate constraint may be scheduling or delivery rather than advertising. Increasing budget under those conditions can waste opportunity and damage the experience. A useful performance review connects acquisition efficiency with the business's ability to respond and fulfill its offer.

Design a pilot with an honest comparison

Microsoft's optimization guidance discusses conversion, audience, and asset inputs to Performance Max. Microsoft Advertising: Increase Conversions with Performance Max Those inputs are areas to review, not evidence that a particular configuration will succeed for every advertiser. A pilot should state the specific question it is intended to answer and the conditions under which the answer will be considered useful.

Choose a scope the business can evaluate. That might be one service, one region, or one clearly defined campaign group. Establish the current measurement and qualification process before making the change. Record the baseline period and any demand factors likely to differ during the pilot. If seasonality, a new offer, or a sales-team change overlaps, include that context rather than treating the pilot as isolated.

Where a controlled comparison is feasible, plan assignment and outcomes before launch. Where it is not, describe the result as an observational comparison and avoid causal certainty. A before-and-after review can still inform a decision, especially when combined with detailed lead review, but it should not be presented as stronger evidence than it is. The label should match the method.

Specify stop and review conditions. Broken measurement, an inappropriate destination, unusual spam, or a capacity problem may justify intervention. Ordinary short-term variation may require observation rather than immediate reaction. The team should know the difference before the first fluctuation appears. A written pilot brief helps protect the campaign from both unattended problems and impulsive changes that make learning impossible.

Review lead quality with a small, disciplined panel

A useful pilot review includes the people who see different parts of the journey. Marketing can explain the campaign promise, sales can assess the conversation, and operations can identify whether the proposed work fits delivery capacity. Their perspectives should be organized around shared definitions so the discussion produces evidence rather than competing anecdotes.

Review a sample that includes accepted, rejected, pending, and unusual leads. Do not show only the best conversations or the most frustrating ones. For each record, ask what the person expected, what they requested, whether the service fit, and what happened next. Compare those observations with the campaign and page that brought the person in, where that connection is available and appropriate.

Keep a separate category for missing information. If a salesperson has not attempted contact or recorded the result, the campaign should not automatically receive blame for poor quality. Equally, a lead should not remain indefinitely promising simply because nobody completed the review. Assign follow-up responsibility and a reasonable process for resolving pending records.

Look for actionable patterns.

• Repeated confusion about scope suggests a copy or offer problem.
• Repeated out-of-area requests suggest reviewing both settings and page clarity.
• Strong fit with weak attendance suggests examining the booking experience and reminders.

The purpose of quality review is to identify the next useful change across the system, not to decide which department deserves credit for every result.

Make the agency review concrete

When evaluating a proposal inspired by this case, ask the provider to show a sample decision record rather than only a sample dashboard. The record should identify a campaign question, the evidence reviewed, the limits of that evidence, the decision made, and the next observation that could change the decision. This reveals how the provider thinks when performance is ambiguous, which is often more important than how it presents a strong result.

Request an example of a lead-quality disagreement and how it would be resolved. If marketing calls a record qualified but sales rejects it, the process should examine the definition and the underlying information. It should not default to an argument about which team owns the problem. A shared review can reveal an unclear offer, a missing field, inconsistent sales follow-up, or a legitimate difference in business priorities.

Ask who can change conversion goals and how those changes are recorded. An unnoticed goal change can make historical comparisons misleading and alter the behavior of automated bidding. The person approving the commercial objective should understand the proposed change even if a technical specialist implements it. Important configuration decisions deserve a plain-language explanation and a verifiable record.

Finally, agree on the form of the pilot conclusion before work begins. A useful conclusion might recommend expansion, a narrower test, a measurement repair, or discontinuation. It should explain the reason in terms of suitable opportunities, cost, capacity, and evidence quality. A provider willing to recommend a different next step when the facts change is offering a more credible process than one committed to presenting every pilot as a success.

This review does not require access to another client's private data. It requires clarity about the method the provider will apply to your business. Public cases can demonstrate possibilities, while the proposal should demonstrate how your own decisions will be made and checked.

Decide whether to expand, refine, or stop

At the end of the pilot, compare the evidence with the original question. Did the program produce more suitable opportunities at an acceptable cost? Did the team maintain consistent qualification and measurement? Did the audience experience match the offer? Which uncertainties remain large enough to affect the decision? A useful conclusion can be conditional without being evasive.

Expansion should follow a reasoned view of capacity and marginal performance. The next increment of spend may reach a different mix of opportunities than the first. Do not assume that an observed average will remain unchanged at a much larger budget. Scale in a way that allows the team to review quality and operating constraints, with a clear path to adjust if the mix changes.

Refinement may be the strongest outcome when the mechanism looks promising but a specific part of the journey fails. The campaign may need clearer creative, a better qualification explanation, or more reliable CRM feedback. Stopping can also be a rational decision when the economics or fit do not support further work. A successful pilot produces a better decision, not necessarily a permanent commitment to the tested campaign type.

The WORX story is useful when it prompts that level of rigor. It points toward the relationship between intent signals, creative, automation, and business feedback while leaving many client-specific details undisclosed. The Mangione Group's buyer intent work and outreach funnels can be evaluated through the same practical questions: who should be reached, what should they understand, which outcomes matter, and how will the team know whether the program is improving?