Skip to content
whitead

Fix: Target ROAS running without real conversion value tracking

finding google ads updated 2026.05.27 8 min read
// contents · 6 sections
  1. Why this matters
  2. How to verify the issue
  3. How to fix it
  4. How to confirm the fix worked
  5. Related rules + concepts
  6. Sources

Target ROAS is the strategy you set, but the auction is optimising against a value signal that does not exist. If conversion values are blank or every conversion is recorded as the same flat amount, tROAS silently degrades to a noisy Max Conversions — the ROAS column in your reports becomes arithmetic, not optimisation [1][2].

Why this matters

Target ROAS is a value-maximisation strategy with a return-on-spend ceiling: it raises bids on auctions where predicted conversion value per dollar exceeds the target and suppresses bids where it falls below [1]. The mechanism is differential — it needs to distinguish a $25 trial from a $2,400 annual purchase to bid sensibly. Strip that signal away and the differential collapses.

Two failure modes look identical in the campaign list:

No conversion value at all. The conversion action was created with "Don't use a value for this conversion action," which the Google Help page itself flags as "not recommended" [2]. tROAS still runs technically, but the value-per-conversion is null for every event the strategy ingests. There is no gradient to climb.

"Same value" placeholders. The conversion action was created with "Use the same value for each conversion" — typically $1.00 or some internal placeholder — and that placeholder is the only value the bidder sees [2]. The math becomes pathological: ROAS = (count × $1) / cost, which reduces to (1 / CPA). The bidder is now optimising for cost-per-conversion through a tROAS wrapper, just with extra noise from a fake denominator and no ability to weight high-value events.

In both cases the UI still displays a "Conv. value / cost" column. That column is a report — it does not feed back into bidding when the underlying signal is flat. The cost is double: you lose the bid intelligence you thought you were buying, and the dashboard makes the campaign look like it is hitting its ROAS target by construction. Practitioner analyses of large bid-strategy samples find Max Conversions plus Target CPA still dominate adoption, partly because value-based bidding fails silently when the value layer is missing [3]. Google's automated-bidding overview groups Target ROAS with Max Conversion Value as the strategies that "get as much conversion value as possible at the target return on ad spend" [4] — the product intent is value-aware optimisation; the audit finding is that the prerequisite is missing.

How to verify the issue

  1. Open Tools → Conversions → Summary. For every conversion action attached to a tROAS campaign, look at the Value column and the Value setting column.
  2. If Value setting = Don't use a value for this conversion action, the finding fires (no value) — this is the most severe variant.
  3. If Value setting = Use the same value for each conversion and the value is a placeholder ($1, €1, or an internal flat number that is not your real average order value or lead value), the finding fires (placeholder value).
  4. Open Campaigns → filter by Bid strategy type = Target ROAS (or open the campaign and check Settings → Bidding). Cross-reference against step 1-3: any tROAS campaign whose primary conversion action falls into either failure mode is the audit target.
  5. Open Segments → Conversion action on the campaign for the last 30 days (or pull the same breakdown via Report Editor with Conv. value aggregated). If 100% of conversion volume sits in a single value bucket — or every action carries a placeholder $1.00 — the bidder has no differentiation, finding confirmed.
  6. For Shopping / Performance Max: check that the Merchant Center feed sends price and the conversion tag reads transaction_value. A broken feed produces flat values even when the action is technically "Different values."

How to fix it

The sequence is non-negotiable: turn on real value first, then let tROAS recalibrate. Switching tROAS targets before value is flowing makes the problem worse — the learner builds on the new (still flat) signal.

  1. Pick the right value source.
    • Ecommerce (Shopping, PMax, Search-to-cart): dynamic transaction values from the cart layer. The Google tag (or GTM purchase event) reads value and currency from the dataLayer at checkout.
    • Lead-gen, predictable lead value: "Use different values for each conversion" with per-action values (demo, free trial, paid signup), anchored to historical close-rate × average contract value.
    • Lead-gen, variable lead value: Enhanced Conversions for Leads + Offline Conversion Imports so closed-won revenue feeds back to the GCLID/gbraid/wbraid that originated the lead — the only path where tROAS learns from real revenue. See Enhanced Conversions for Leads.
    • Mixed signals: apply Conversion Value Rules on top to adjust by location, device, or audience.
  2. Implement the value source. For dynamic values, deploy the dataLayer push or GTM trigger; verify in Tag Assistant that the conversion event carries value and currency with non-zero, non-uniform amounts on real test transactions. For OCI, capture GCLID at form submit, store in CRM, push closed-won deals back via Google Ads Data Manager or the offline conversion API.
  3. Pause tROAS while you re-baseline. Switch the campaign temporarily to Maximize Conversion Value (no target). Same value signal, no ROAS ceiling — the new value distribution stabilises without delivery throttling. Run 14-30 days.
  4. Re-anchor the ROAS target on real values. Compute achieved ROAS from the last 30 days of value-cost. Re-enable Target ROAS at a number that respects margin: 1 / (margin × buffer) for ecommerce, or the LTV-to-CAC ceiling agreed with finance. Do not pick a target 30%+ above the post-fix baseline — the strategy throttles delivery rather than "stretching."
  5. Freeze the campaign for 14 days after re-enabling tROAS. Budget swings >20%, target changes >15%, conversion-action edits, and new audience signals all retrigger the Smart Bidding learning phase.
  6. Audit value distribution monthly. Segment conversion actions by value bucket. A healthy value-based campaign shows a long tail (p90 ≥ 5-10× the median for ecommerce). A flat or bimodal distribution = downstream tracking issue.

How to confirm the fix worked

  • Value setting on every tROAS-attached conversion action = Use different values for each conversion (or, for OCI, Different values with offline source feeding real revenue).
  • Tag Assistant records non-uniform value and currency parameters on at least 10 live test conversions across different transaction sizes.
  • The Conv. value column in the campaign report shows a value distribution with at least 5× spread between p10 and p90 over the last 14 days.
  • Campaign bid strategy = Maximize Conversion Value (interim) or Target ROAS (post-recalibration), with the strategy badge showing "Eligible" rather than "Learning, limited" or "Misconfigured."
  • Conv. value / cost (ROAS column) tracks within ±20% of the target after the 14-day freeze window; if it sits dramatically above the target, the target is too soft and the strategy is leaving volume on the table.
  • No "Conversion value missing" or "Same value placeholder" warning on the conversion actions in the Diagnostics view.

Blast radius and co-occurring findings. This finding rarely appears alone. Accounts that ship tROAS without real values almost always also fail conversion tracking hygiene checks — Enhanced Conversions disabled, primary action mislabelled, or transactional events under-counted because of consent-mode gaps. For EEA/UK traffic specifically, Consent Mode v2 Advanced must be live or the value layer is structurally undercounted regardless of tag setup — the audit can fire as troas_without_value_tracking even when value is correctly configured at the tag, because the bidder ingests modeled-only conversions for unconsented traffic. Fix conversion tracking first, value setup second (this finding), and only then revisit the bid strategy choice itself. Sequencing matters because tROAS recalibration is expensive: every change to the underlying conversion or value model retriggers the 7-14 day Smart Bidding learning window, and a campaign that re-enters learning three times per quarter spends most of the quarter in exploration. Treating this as critical reflects the silent-degradation property: the campaign keeps spending, the dashboard keeps reporting a ROAS number, and the auction-time bidder is operating on a signal that does not contain the information the user thinks it does.

Sources

  1. Google Ads Help — About Target ROAS bidding. https://support.google.com/google-ads/answer/6268637 (accessed 2026-05-27).
  2. Google Ads Help — Set up the same value for each conversion. https://support.google.com/google-ads/answer/13064107 (accessed 2026-05-27).
  3. Adalysis — What's the most popular Google Ads bid strategy? https://www.adalysis.com/blog/whats-the-most-popular-google-ads-bid-strategy/ (accessed 2026-05-27).
  4. Google Ads Help — About automated bidding. https://support.google.com/google-ads/answer/2979071 (accessed 2026-05-27).
// was this useful?
// anonymous · no personal data stored
← Back to Wiki