Historic stone courthouse with American and state flags on the lawn, illustrating the federal ad tech remedies order now governing Google Ads routing

On September 16, 2026, the full memorandum opinion in the Justice Department’s ad tech case against Google became public, with Courthouse News posting the PDF the same day. It does not order Google to sell AdX or its publisher ad server. It orders six years of behavioral restrictions, a third-party Monitor, and — in the part almost nobody quoted — routing and nondiscrimination rules on the Google Ads buy side.

That last item is the only piece of the ruling that reaches a Michigan advertiser’s account, and it is the piece the “Google avoids breakup” headlines skipped. Here is what the document obligates, what it does not, and what is worth doing in an account in Warren or Grand Rapids before the Q4 display ramp.

What did the court actually order Google to do?

Four things: no divestiture, behavioral injunctions, outside oversight, and a six-year clock. Judge Leonie Brinkema declined to force a sale of AdX or the publisher ad server, writing that divestiture was “neither realistic nor needed” and casting the DOJ’s push as resting on “a lack of trust that Google will comply.”

Per Search Engine Roundtable’s reading of the memo, the operational mandates are three: basic data-sharing provisions, an injunction against discriminatory bidding that favors Google’s own products, and mandatory integration of Google’s ad exchange with rival publisher ad servers. Compliance is supervised by a third-party Monitor and a Technical Committee rather than left to Google. On the term, the judge took Google’s six years over the longer window enforcers wanted — Search Engine Roundtable reported the DOJ sought more than 12 years, Courthouse News coverage cited 15 — while keeping authority to extend. None of it disturbs the April 2025 liability finding that Google illegally monopolized the publisher ad server and ad exchange markets.

Why does the buy-side piece matter more to advertisers than the breakup that didn’t happen?

Because the buy side is where advertiser money enters the system, and it is the only part of the order that changes what a Google Ads campaign can reach. Jason Kint, who has followed the docket closely, noted on September 16 that the judge reconfirmed her view that Google’s advertiser demand — the old AdWords buy side — is the “golden goose” creating the monopolies, and put nondiscrimination and routing restrictions on it, despite Google arguing the buy side sat outside the markets she found illegal.

The mechanism predicts effects nobody has written about yet. When a Display, Demand Gen, or Performance Max campaign bids, Google’s stack decides which auction path the budget travels down, and that routing historically favored Google’s own exchange: demand reachable only through Google Ads hit AdX inventory first. A routing and nondiscrimination obligation means the same budget must reach rival exchanges on comparable terms.

Widen the set of auctions a fixed budget can enter and three things move together: the inventory mix changes, effective CPMs drift as the bidder finds cheaper paths, and frequency across sites flattens. Any campaign type with a display component inherits all three. That is the rule to hold onto — it will tell you what to expect in a campaign type this article never mentions.

Does this change anything in a Google Ads account this quarter?

Not yet, and probably not in October. Behavioral remedies move on the schedule of a final judgment, a Monitor who has to be appointed, and engineering work nobody outside the Technical Committee will see. Treat anyone selling an urgent restructure off this ruling as selling you something. What changes this quarter is your ability to detect the effect later: if placement mix shifts in 2027 and you have no 2026 baseline, you will read the shift as a performance problem and pull levers that were never the cause.

Who in Michigan is most exposed to a display inventory shift?

Multi-location service businesses across the metro, and it is a timing problem more than a strategy problem. A furnace contractor covering Warren, Sterling Heights, and Troy spends the bulk of its year’s budget between October and February, much of it through Performance Max, where the display arm is what the advertiser controls least. Same shape for a dental group in Rochester Hills or a roofer in Livonia: seasonal concentration, one shared budget, a placement mix audited roughly never.

Here is the specific way that costs money. Say the Warren contractor runs $8,000 a month through Performance Max from October through February — $40,000 of the year’s acquisition spend. If display share starts sourcing from exchanges the account has never seen, with no account-level exclusions and no content-suitability tier set, a real slice of that budget lands on unvetted inventory during the only five months that matter. The loss never shows up as a spike, but as a quietly worse cost per booked job that gets blamed on the weather. Michigan’s Tier 1 and Tier 2 suppliers have the mirror image: long B2B cycles make the shift harder still to attribute. For a second read, our Detroit paid search team runs this as a standing audit.

How should a small team baseline placements this week?

Six steps a three-person team can finish in an afternoon.

  1. Export a 90-day placement report for every Performance Max, Demand Gen, and Display campaign. Save a dated file, not a saved view — saved views regenerate and the snapshot is gone.
  2. Record display-share CPM by month for the trailing twelve months. Drift is only visible against a line.
  3. Set account-level placement exclusions now if you have none. Most accounts we open have an empty exclusion list.
  4. Pick a content suitability tier deliberately instead of accepting Standard inventory, and write down which tier and why. In twelve months you will not remember.
  5. Isolate display inside Performance Max by asset group where structure allows, so a channel-level change is legible instead of averaged away.
  6. Set a quarterly reminder to re-pull the report and diff it against the September 2026 baseline. Fifteen minutes, four times a year.

If you also sell ad inventory on your own site — plenty of Michigan trade publications and association sites do — add a seventh step: ask your ad server vendor in writing what mandatory integration changes for them, and on what timeline. Their answer, or its absence, tells you how seriously to take their roadmap.

Is six years of oversight actually long enough?

Probably not, though not for the reason the DOJ argued. Enforcers wanted longer because they distrust Google’s compliance. The better objection: six years is long in calendar time and short in architecture time.

Consider what the order regulates: exchanges, publisher ad servers, and the routing of demand between them — the open-web display auction roughly as it has existed for fifteen years, and the part of the market advertiser spend has been steadily leaving. If commercial attention keeps migrating onto AI answer surfaces over the term, the Monitor and Technical Committee will spend 2031 supervising an auction that matters far less than it did in 2026, and none of these rules reach the surfaces where the spend went.

Second weakness: behavioral remedies only work when violations get noticed, and the parties best positioned to notice discriminatory routing are small advertisers and small publishers — who have no visibility into the auction, no access to the Monitor, and no standing to complain. The Technical Committee closes some of that gap. Whether it closes enough is the question worth watching, not the breakup-or-not argument that filled September.

What dates should be on the calendar?

  • April 2025 — Liability ruling: Google illegally monopolized the publisher ad server and ad exchange markets. Undisturbed.
  • Early September 2026 — The court declines to order divestiture of AdX or the publisher ad server.
  • September 16, 2026 — Full memorandum opinion published; Courthouse News posts the PDF. Six-year term, Monitor and Technical Committee, three operational mandates.
  • October 1, 2026 — Michigan heating season. HVAC, plumbing, and roofing budgets concentrate. Baseline before this, not after.
  • Entry of final judgment — the six-year clock runs from here, not the opinion date. I have seen no entry date reported as of September 22; treat any specific date as unconfirmed until the docket says otherwise.
  • January 2, 2027 — First quarterly placement-report diff against your September 2026 baseline.

Frequently asked questions

Does this ruling mean my Google Ads costs will go down?

Nobody can promise that. A fixed budget gaining access to more auction paths points toward lower effective display CPMs over time, but it also points toward a different inventory mix, and cheaper impressions on worse inventory is not a win. Judge the change on cost per qualified lead across a full season, not CPM in isolation.

Should I pause Performance Max until the remedies take effect?

No. There is no effective date to wait for, the order requires a final judgment and an appointed Monitor before anything moves, and pausing a seasonal campaign to dodge a change that may take quarters to appear costs you the season. Baseline your placements and keep running.

Does any of this affect Search campaigns or organic rankings?

Not directly. This case concerns the open-web display stack: publisher ad servers, ad exchanges, and how advertiser demand routes between them. Google Search ads and organic results were the subject of a separate antitrust case with its own remedies and appeal. Keep the two straight when coverage blurs them.

Is Google going to appeal?

An appeal is widely expected given the April 2025 liability finding Google has disputed throughout, but as of September 22, 2026 I have seen no filing reported, so treat any appeal timeline as unconfirmed. It changes little either way: an appeal delays implementation, and the useful work is the baseline.

The short version

Google kept AdX. It did not keep the freedom to route advertiser demand wherever it prefers, and that restriction — not the absent breakup — is the part with a line into your account. Acting on it means an export, an exclusion list, and a reminder. Do that and you will read 2027 correctly; skip it and you will spend next year debugging a change that was never yours. To have the baseline built and documented, start here.

Category
Tags

Comments are closed