By November 2028, at least 10 agencies on a fixed 20-provider watchlist selling combined Google and Microsoft Ads management will explicitly describe cross-channel lead deduplication. Not platform totals side by side. One person counted once.
That is my most concrete bet on a bigger change: buyers will stop accepting two platform lead totals as proof of two separate contributions.
The arithmetic is easy to flatter. A shopper clicks your Google ad on Monday, clicks your Bing ad on Thursday, buys on Friday, and both platforms can claim the same order. Add those dashboards together and you can end up reporting revenue that never existed in your bank account. Two totals look like two sources of new customers. They are not.
Say you spend $20k a month on Google and add $3k on Microsoft. Microsoft reports 40 leads at a tidy cost per lead. Your CRM shows total leads flat month over month. That does not establish what Microsoft added. Some demand may have moved between channels; the dashboard alone cannot settle it.
That distinction drives all three predictions here: I expect buyers to demand additional qualified customers, not merely another platform claiming existing demand.
Count the person once before arguing about who added them
Each ad platform grades its own homework. Google can count a conversion when its click sits in the path. Microsoft can do the same for its click. Neither platform’s total, on its own, reconciles the overlap with the other. That is how one person becomes two reported conversions, and how two attractive cost-per-lead numbers can sit beside one flat sales total.
I draw a hard line between attribution and incrementality. An attributed platform touched the journey. An incremental contribution added a customer you would not otherwise have had.
Deduplication fixes the headcount. It does not establish causation. Matching CRM records by email or phone tells you whether the same person appears twice. It does not tell you whether that person would have bought without the Microsoft ad.
If Microsoft claims 40 leads and your backend count rises by 12, you have 12 more recorded leads than before. You have not yet proved Microsoft caused all 12. That is why I would judge the test on total business leads in the CRM, not platform totals, then use a test to investigate what the extra spend actually contributed.
The sequence matters: get one reliable count, separate the demand being claimed, then test what changes when the spend changes. My predictions follow that order.
Prediction 1: Combined contracts will make deduplicated reporting explicit
The mechanism is straightforward: match records before adding totals. Google reports 60 leads, Microsoft reports 40, and the slide says 100 prospects. Match those records using email, phone, and timing, collapse repeat submissions, and the count might fall to 73 unique people.
That is an illustrative count, not a result from a client account. The point is the gap between platform claims and people. One person submitting twice should not become two new prospects.
The machinery is already part of the offer. Agencies describe GA4, Enhanced Conversions, UET, and offline conversion imports as components of cross-platform measurement. Those connections can feed business outcomes back into the platforms. They do not, by themselves, produce a deduplicated cross-channel report. The agency still has to reconcile the records and show the result.
That is where I expect buying pressure to land. Once a buyer asks why 100 reported leads became 73 people, another dashboard is a poor answer. The reporting method becomes something the agency needs to explain before the contract, not after the discrepancy.
My forecast: by November 2028, at least 10 of the fixed 20 providers will explicitly describe cross-channel deduplication on their offer pages or in sample reports.
I will publish the baseline and definitions before tracking starts. To count, the material must name CRM identifier matching and counting a person once across channels. A footnote warning that double-counting exists does not qualify.
What proves me wrong is equally plain: I check the same 20 providers under the same rules in November 2028, and nine or fewer qualify.
Buy the reconciled count. Do not mistake it for proof of added demand.
Prediction 2: Microsoft brand searches will get their own reporting line
Searches for your business name and searches for the problem you solve answer different questions. Someone typing your name is already looking for you. Someone searching for the service still has to choose a provider.
Blend those searches into one Microsoft Ads total and the cheap brand leads can make the whole account look ready for more budget. The buyer sees a low average cost per lead. The non-brand campaigns get to hide inside it.
That is the mechanism behind my second prediction: once buyers challenge the combined total, they will want to know which demand the agency is claiming.
The distinction is already explicit in advice to report brand campaigns separately from non-brand from day one. Evidence about when branded search is worth the investment also points to the problem with treating every branded conversion as an added sale. People who type your name may have found you anyway.
My forecast: by November 2028, at least 10 of the same 20 providers will publish or offer separate Microsoft brand-search reporting as standard. For this watchlist, that means a distinct budget and lead line, not a vague promise to monitor branded keywords.
The split will not prove incrementality. Neither will a non-brand label. It will make the next question harder to dodge: which part of the account earns more money when its budget grows?
What proves me wrong: fewer than 10 providers meet that standard in November 2028.
I would ask for the split now. A blended average should not decide what gets scaled.
Prediction 3: More budget will require more than a cheap platform lead
A low reported cost per lead is easy to flatter. Keep searches for your own name inside the total, let near-brand queries into the mix, and the average can fall without adding a customer.
That is why I expect the next budget conversation to move beyond CPL. The question becomes: what happens to total qualified leads when Microsoft spend stops in one place and keeps running in another?
The method is not hypothetical. A branded-search test can pause campaigns in one area for two to four weeks while keeping a control live. Compare total paid-plus-organic outcomes, rather than celebrating whichever dashboard retains the conversion claim.
For a test of branded campaigns, pause branded campaigns. For a claim about the whole Microsoft account, the test has to address that broader spend. A brand pause cannot establish the contribution of everything else still running.
My forecast: by November 2028, at least five of the same 20 providers will explicitly offer holdout or staged-budget tests tied to additional qualified leads, with the method named on the page.
A holdout strips the argument down to something useful: pick two areas that usually produce similar lead volume, keep the relevant Microsoft campaigns running in one, pause them in the other, and compare backend outcomes. If total customers hold steady where spend stops, that is evidence against the claim that those ads were adding customers.
Low-volume accounts need patience. With 15 leads a month, two weeks proves little. I would rather run longer and accept a less certain read than pretend a dashboard settles it. A staged increase also needs an agreed outcome measure; spending more and receiving more platform-attributed leads is not enough.
What proves this prediction wrong: four or fewer providers offer a qualifying test in November 2028.
Before approving more budget, ask what result would make the agency recommend less.
What I would buy today: an owner, a reconciled count, and an exit rule
If I were hiring for both engines this month, I would not buy a reach pitch. I would buy a plan to establish added customers, in writing, before the first dollar of Microsoft spend.
Three questions do most of the work:
-
Who checks tracking and reconciles leads across channels? I want a named person who verifies UET and GA4 on the relevant conversion pages, matches platform conversions to CRM records weekly, and reports unique qualified leads rather than stacked totals. A dashboard link is not an owner. I also want that person to explain what the reconciled count can establish and what still needs testing.
-
Who manages Microsoft separately after the Google import? The default failure is campaigns imported once and left to drift, with paused ads still running. Negatives, feeds, and conversion goals need attention on Microsoft, not an assumption that the import completed the job. My comparison of who should actually run your Microsoft Ads goes deeper. The short version: one owner for both is fine. One unattended copy for both is not.
-
What evidence increases, reduces, or stops the budget? I want the thresholds agreed before spend starts. At the 60-day review, what result keeps Microsoft running, what triggers a rebuild, and what shuts it off? If volume is too low for a useful read, what longer test do we agree to? “Lower reported CPL” does not answer any of those questions.
The third question keeps the first two honest. Clean tracking and attentive account management matter, but they are not the business outcome. They make it possible to judge the outcome without arguing over whose dashboard deserves credit.
A serious budget answer names the backend measure and the test. It might separate brand search for four weeks or stage the next $2k behind a target of 15 additional qualified leads. Those are proposed decision rules, not universal thresholds or proof that a test has enough volume. Agree on what the account can support before attaching certainty to a number.
That is how I would keep the reporting focused on customers, not claims. Agree on the cutoff while everyone is calm, not after the invoice arrives.

Where groas fits in that hiring screen is straightforward: combined Google and Microsoft management as one owned system, with tracking checked on both sides, Microsoft managed for its own auction, and leads reconciled to unique people in the backend.
The stated fee is $1,499 per month for combined management. Media spend stays separate, and results are not guaranteed. The point is accountable management, not two totals dressed up as growth. I would rather see a trial lose on a flat backend count than a retainer win on duplicated claims. If the evidence does not show added customers, pay less attention to the dashboards and more to the exit clause.
Put the forecast on the calendar, then change what you buy
I will publish the fixed 20-provider watchlist with baseline coding before tracking starts: 20 agencies publicly selling combined Google and Microsoft management, linked and dated, with three yes-or-no codes per provider.
The rules will stay visible:
- Deduplication: the page or sample report names CRM matching by email or phone and counting one person once across channels.
- Brand separation: Microsoft brand search gets its own budget and lead line apart from non-brand.
- Budget testing: the page names a holdout, geo pause, or staged increase tied to additional qualified leads in the backend, not merely a lower platform CPL.
Same list, same rules. No substitutions when someone rebrands.
I will recode the list in November 2027 and November 2028 and publish the sheet, hits and misses together. A prediction that reports only wins is a sales deck.
Prediction 1 passes at 10 providers. Prediction 2 passes at 10. Prediction 3 passes at five. Below those thresholds, each fails. If I miss two out of three, I will write plainly that the thesis was early or wrong.
These are tests of what agencies explicitly offer, not an audit proving every agency delivers it. That is a deliberate limit. My bet is that buyers will make these requirements visible enough to become part of the sales offer.
If I am right, do not wait until 2028 to buy differently. Demand one backend count of unique qualified leads as the starting point for renewal decisions. Demand brand search on its own line. Then agree on the test and budget rule that distinguish additional customers from reassigned credit.
Two platform totals will keep looking like two sources of customers until buyers stop paying for the arithmetic. Stop paying for it now.

