Original Conversion Value Reaches the Google Ads API (August 19, 2026): The Column Michigan Advertisers Should Have Added Nine Months Ago

Google shipped version 25.1 of the Google Ads API on Wednesday, August 19, 2026, and one of its six additions is a metric named original_conversion_value — the value of a biddable conversion before Google’s own value adjustments are applied. The same figure has been available as an “Original Conversion Value” column inside the Google Ads interface since November 2025, and most accounts in metro Detroit have never switched it on.

The API release is engineering news. The column is the part a three-person marketing team can act on before Friday.

What did Google Ads API v25.1 actually add on August 19, 2026?

Six features, two locked behind an account allowlist. Google called it a “drop-in upgrade” for v25 — no breaking changes, no removed resources — in a Developer Blog post credited to Anash P. Oommen of the Google Ads API Team, with a live walkthrough held August 20 at 10 a.m. ET.

• Read-only LiftMeasurementConfig and LiftMeasurementFlight resources, 24 new Conversion Lift metrics, winner score metrics, and Brand Lift dimensions for age range, campaign, device, gender and video — allowlist only.
• A category_filter parameter in BenchmarksService, plus aggregate cost, views and share_of_voice metrics.
• BrandSentimentInsight in ContentCreatorInsightsService — allowlist only.
• Segments.loyalty_membership for reporting, and a LOYALTY_MEMBERSHIP primary dimension for conversion value rules.
• Metrics.original_conversion_value.
• parental_statuses targeting in ReachPlanService.

PPC Land’s Luis Rijo put the sharpest point on it: across all six additions, the pattern is retrieval rather than control. Lift studies can be inspected but not created; benchmarks scoped but not configured.

What is original conversion value, and why does the gap matter?

Original conversion value is what a conversion was worth before conversion value rules or customer lifecycle goal adjustments rewrote it. Your regular “Conv. value” column is the number after that rewrite. The difference is not an error — it is the measurable size of Google’s editorial hand on the ROAS you report to your owner or CFO.

Most advertisers have never calculated that difference, because until November 2025 there was no supported way to see the pre-adjustment number at all. The nine-month lag before it reached the API is exactly the gap Google’s January 2026 shift to monthly releases — four major and eight minor versions a year — was meant to shorten.

How do conversion value rules change what Smart Bidding actually buys?

A value rule multiplies a conversion’s recorded value based on an impression dimension — audience, location, device, now loyalty membership. The multiplied figure, not the original, is what Target ROAS optimizes against. That single fact creates a loop worth tracing.

Set a rule multiplying conversions from a given ZIP code radius by 1.4. The bidder now sees richer returns there, bids harder and shifts budget toward it. More conversions arrive from the boosted segment, each inflated by the same 1.4. Reported ROAS climbs. Your deposits do not. Nothing malfunctioned; the system did what you configured it to do, at a scale you did not anticipate when you typed the multiplier.

The rule of thumb this gives you: any value rule keyed to a dimension the bidder can target will compound, because the bidder can chase the boost. A rule keyed to something the bidder cannot steer toward — a lifecycle goal applied after the fact, for instance — adjusts the reported number without redirecting spend. Apply that test to a device-level rule you set last year and you can predict the outcome without running it.

None of this makes value rules wrong. A multiplier encoding a genuine margin difference is good practice. The failure mode is that almost nobody revisits it. It was a reasonable estimate on the Tuesday it was created and has steered bids ever since, through a supplier price increase and two rounds of menu pricing.

Why does this hit Michigan’s fitment catalogs and trade contractors hardest?

Because value adjustment does the most damage where a single conversion action spans an enormous range of real transaction values — which describes a large share of the region’s advertising base.

An aftermarket parts seller in Warren or Roseville fires one “Purchase” action across a $14 cabin filter and a $2,400 turbo kit. A location rule written when the business was mostly counter sales keeps inflating in-state orders long after the mix shifted to freight-heavy out-of-state shipments with thinner margins. A Tier 2 supplier in Sterling Heights gets one “Request a Quote” form worth either $400 or $400,000, and patches the gap with a geographic multiplier because the CRM cannot return a real value inside the attribution window. An HVAC contractor covering Livonia and Rochester Hills books a $180 diagnostic call and a $9,400 system replacement through the same “Lead” action.

In each case the rule stands in for data the business already owns but has not connected. The stakes are concrete: a contractor running a 400% tROAS target against values inflated 18% is really running roughly 340%, buying leads that would have failed the stated threshold, and will not learn it from any report on screen. That is the reconciliation problem we work through with Detroit-area Google Ads accounts whenever reported returns and bank deposits stop agreeing.

How should a small team audit its adjusted conversion values this week?

Entirely inside the interface, in about ninety minutes — no developer token, no allowlist, no API.

1. Add the column. Columns → Modify columns → Conversions → Original Conversion Value. Place it directly beside Conv. value and segment by conversion action.
2. Pull 90 days and subtract. Calculate (Conv. value − Original conv. value) ÷ Original conv. value per conversion action. Anything over roughly 10% means a meaningful share of your reported return is arithmetic, not revenue.
3. Inventory every rule. Tools → Conversions → Value rules. Record the multiplier, dimension, creation date and owner. Any rule older than twelve months with no documented source is a guess that has been compounding.
4. Check each multiplier against real margin. Pull actual gross margin for that segment from your POS or CRM over the same 90 days. If multiplier and margin disagree by more than a few points, the rule is training the bidder toward the wrong customers.
5. Test by removal, on one campaign. Pause a single suspect rule for 21 days on one campaign — never account-wide — and compare the original conversion value trend. Judging the test on adjusted value defeats the purpose.
6. Move the tROAS target at the same time. Your target is denominated in adjusted value. Remove an inflating rule without touching the target and you have raised the bar overnight, which reads as a sudden volume collapse. Lower the target by roughly the inflation percentage from step 2, on the same day.
7. Report both numbers upward, permanently. One line for adjusted, one for original. The delta becomes a standing metric instead of a twice-yearly discovery.

Does the allowlist gate make the rest of v25.1 useless for most advertisers?

For lift measurement and creator sentiment, effectively yes. For benchmarking, no. Both lift resources and ContentCreatorInsightsService require account-level allowlisting arranged through a Google representative, and Google has published no figures on how many accounts hold that status.

The lift gate stings more than it looks, because the financial barrier already came down. Google lowered the minimum budget for incrementality experiments to $5,000 in November 2025 and recommended a minimum study duration of 14 days, noting that shorter studies showed drops of up to 17% in Absolute Lift for businesses with longer conversion lags. Read that against a Michigan B2B cycle running 60 to 90 days from RFQ to purchase order: a 14-day study is structurally too short for that business regardless of who holds the allowlist.

The category filter in BenchmarksService is the quiet winner. Unscoped benchmarks compare a Royal Oak retailer against the entire advertiser population, which teaches nothing. Scoped to a matched Product & Service Category — discoverable through ListBenchmarksSources — the comparison acquires real diagnostic value, and it reaches whatever reporting platform you already pay for without any code.

What dates should be on your calendar right now?

• August 31, 2026 — Google removes the enable_local opt-out for Shopping campaigns. On v25.1 and later, setting the field to false returns ContextError.OPERATION_NOT_PERMITTED_FOR_CONTEXT rather than being ignored. If a vendor upgrades you to v25.1 this month, local inventory ads arrive with it.
• September 2026 — the next minor release falls due under the cadence Google adopted in January 2026. Expected from the schedule, not announced.
• July 2027 — v25 reaches the end of its twelve-month support window, and v25.1 retires alongside its parent.
• Before your next budget review — bring the original-versus-adjusted delta from step 2. Nobody else in the room will have seen it.

Frequently asked questions

Do I need Google Ads API access to see original conversion value?

No. The Original Conversion Value column has been available in the Google Ads interface since November 2025 and is added through Columns → Modify columns → Conversions. The August 19, 2026 release brings the same figure to the API as original_conversion_value, which matters to reporting vendors and in-house engineering teams, not to anyone working in the interface.

Does original conversion value strip out Google’s attribution modeling too?

No. Google defines the metric specifically against conversion value rules and customer lifecycle goal adjustments. It is not a raw transaction feed and does not undo data-driven attribution or credit assignment across touchpoints. Treat it as the value your tracking recorded before the value-rule layer, not the figure your accounting system shows.

Will removing a conversion value rule tank my campaign?

It can, if you remove the rule and leave the target alone. Smart Bidding targets are denominated in adjusted value, so stripping an inflating rule raises the effective bar and volume drops the next day. Remove the rule and lower the tROAS target by roughly the measured inflation in the same change, then hold both steady for 21 days.

The measurement work that survives an algorithm change is rarely the newest feature. It is knowing which of your own numbers you can defend. If you want a second set of eyes on the value rules steering your budget, that is the audit our team at ROSH Media runs before touching a single bid.