---
title: "WordStream’s Google Ads Grader, Read by Claude: A Grade Is Not a Change Order"
description: "A teardown of WordStream’s Google Ads Performance Grader: what its scores reveal, what they miss, and what Claude should verify through groas before any change."
url: "https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/378e6704-8f39-460f-a26a-c58d92576cf0.png"
published: "2026-11-22T05:41:29.942Z"
modified: "2026-11-22T05:41:29.991Z"
---

November 22, 2026 · 10 min read

# WordStream’s Google Ads Grader, Read by Claude: A Grade Is Not a Change Order

[](https://groas.com/author/alexander-perelman)

[Alexander Perelman](https://groas.com/author/alexander-perelman)Personally Tested

Head Of Product @ groas

[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![Cartoon of a contractor about to sledgehammer a sound kitchen wall because a report card taped there says C-, while the homeowner rushes in to stop him.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/378e6704-8f39-460f-a26a-c58d92576cf0.png)

In this article

1. [The opening promise: a score plus next steps](#the-opening-promise-a-score-plus-next-steps)
2. [The credentials: scale is not account-level proof](#the-credentials-scale-is-not-account-level-proof)
3. [Quality Score: a review queue, not a rewrite order](#quality-score-a-review-queue-not-a-rewrite-order)
4. [Impression share: missing exposure is not a bid instruction](#impression-share-missing-exposure-is-not-a-bid-instruction)
5. [Wasted spend: the dollar figure needs cross-examination](#wasted-spend-the-dollar-figure-needs-cross-examination)
6. [Account activity: movement is not management](#account-activity-movement-is-not-management)
7. [The missing layer: can the conversion data support any of this?](#the-missing-layer-can-the-conversion-data-support-any-of-this)
8. [From report to live change: require a ticket](#from-report-to-live-change-require-a-ticket)
9. [The verdict: keep the questions, reject grade-chasing](#the-verdict-keep-the-questions-reject-grade-chasing)

WordStream’s [_Google Ads Performance Grader_](https://www.wordstream.com/free-google-ad-grader) promises an overall score, individual grades and next steps. I chose it for this teardown because that tidy package invites an expensive shortcut: treating a grade as permission to change an account. **A grade is not a change order.**

The page describes 10 key metrics, with wasted spend, Quality Score, impression share and account activity among its headline concerns. Those are useful places to start looking. They are not enough to approve a negative keyword, raise a bid or move a budget.

I’m going to read the artefact in the order its pitch unfolds: the promise, the credentials, the highlighted metrics and the report’s limits. For each, I want two things kept separate: what the grade suggests, and what Claude must verify against connected account data through groas before execution becomes possible.

## The opening promise: a score plus next steps

The [grader page](https://www.wordstream.com/free-google-ad-grader) frames its output as an overall grade plus individual grades produced by a proprietary grading algorithm. Its report contents extend beyond the four headline areas to include benchmarks, best practices, best and worst text ads, CTR, account structure, a performance tracker, and display and shopping.

That breadth is the appeal. An advertiser gets a manageable report instead of another afternoon wandering through Google Ads menus. An independent [PaceAds review](https://paceads.com/blog/wordstream-review-pricing-grader-and-alternatives) describes a free, roughly 60-second letter-grade report, including a monthly wasted-spend estimate, Quality Score distribution, CTR benchmarks, impression-share losses and a dormancy check.

My verdict on the promise: useful compression, provided nobody mistakes compression for explanation. **The report can identify a question without establishing its answer.** “Why is this campaign losing impression share?” is a question. “Increase its budget” is a decision. There is quite a bit of account between the two.

## The credentials: scale is not account-level proof

Next comes the reassurance. The page cites 6,536,213 free reports, $37.8 billion in spend analyzed and 318 million improvements suggested, alongside Premier Google Partnership status.

It also traces the tool’s history through a 2011 launch, a $3 billion spend milestone, and a relaunch featuring a Performance Tracker that rechecks every 30 days. Mobile Readiness Score and refreshed benchmarks appear in that history too.

I read those details as context about the tool’s age and scale, not validation of its next recommendation. **Millions of reports cannot establish whether this account records purchases correctly.** Nor does an “improvement suggested” tell me whether someone implemented it or whether it helped.

The historical tracker also deserves to stay separate from the initial report. A [grader comparison](https://paceads.com/blog/google-ads-grader-tools-compared) characterizes WordStream’s free grader as a one-time, read-only snapshot, not live account control. That is the boundary I carry into the metric sections: this is an audit artefact, not an operator.

## Quality Score: a review queue, not a rewrite order

Quality Score is one of the first sections the page spotlights. Its FAQ frames the metric around its effect on ROI, rank and cost; the independent review describes a 1-to-10 distribution compared with industry norms.

A distribution is useful because it shows where a closer review might start. It does not tell me which ad to rewrite or which landing page to replace. **A low Quality Score earns an inspection, not an automatic edit.**

For this section, I want Claude to preserve the report’s observation, then retrieve the corresponding account data through groas. The request needs boundaries:

> Label the pasted grader excerpt REPORT. Retrieve the corresponding account data through groas for the same campaigns and reporting period, and label it ACCOUNT. Identify agreements, conflicts and missing context. Separate observations from proposed changes. Make no changes.

I use _REPORT_ and _ACCOUNT_ as evidence labels, not competing authorities. If they disagree, Claude should first check the date range, scope and metric definition. A fresh number from a different campaign is not a rebuttal. It is a different number wearing a confident expression.

My verdict: keep the Quality Score distribution as a queue for ad-relevance and landing-page review. Do not let it write the brief before that review happens.

## Impression share: missing exposure is not a bid instruction

The next highlighted section concerns impression share: how often ads appear for relevant searches. The independent review describes losses split between budget and rank. That distinction matters, but neither label establishes that buying more exposure will be profitable.

Google defines impression share as impressions divided by estimated eligible impressions, based on auctions where the ad was competitive. Its [documentation](https://support.google.com/google-ads/answer/2497703?co=ADWORDS.IsAWNCustomer%3Dtrue\&hl=en\&oco=0/) also notes a 24-to-48-hour processing lag and missing values for some low-traffic or new keywords.

**Lost impression share identifies a constraint, not necessarily an opportunity worth funding.**

Say an account spends $20,000 a month and the report flags impression share lost to rank. I would ask Claude for the campaign-level figures, the reporting period, and supporting context such as absolute top rate, click share and recent auction movement where available. Missing data should stay marked as missing, not be filled with a plausible explanation.

Then comes the business question: do the recorded conversions justify chasing that exposure? If tracking is unreliable, or the campaign’s results do not support expansion, the grade has not earned a bid increase.

My verdict: use this section to locate the constraint. Verify the economics before paying to remove it.

## Wasted spend: the dollar figure needs cross-examination

The grader frames wasted spend around negative-keyword use. The review describes a monthly dollar estimate, which makes the section read like found money. I read it as an invoice that has not been checked.

**A zero-conversion search term is not automatically waste.** It may be irrelevant. It may also involve research, a long sales cycle or an incomplete conversion window. The dollar figure cannot make that distinction on its own.

Here is the Claude check I would attach to this part of the report:

> For each flagged term, retrieve spend and conversion history through groas. Show conversion lag where available and assess relevance against the supplied offer. Separate exclude, investigate and retain, with evidence for each classification. Mark anything that needs business context. Make no changes.

The underlying account check is the search-terms report under Campaigns > Insights and reports > Search terms. It connects actual queries with their triggering keywords and supports review by cost and impressions, as this [practitioner guide](https://surfsideppc.com/blogs/google-ads/how-to-use-search-terms-report-negative-keywords) explains.

I would treat the proposed groups as decisions to review, not a vocabulary blacklist:

- **Exclude candidates:** clearly irrelevant queries, with their spend and conversion history attached.
- **Investigate:** ambiguous intent, including competitor research or modifiers such as “free,” “job,” “tutorial” and “DIY,” checked against the actual offer.
- **Retain:** relevant queries, including early-funnel terms tied to later closed deals.

The same guide explains why match type belongs in that review. Negative broad requires all the terms to appear, in any order; negative phrase requires the phrase in order; negative exact blocks the matching query. Adding negatives from the report defaults to negative exact.

“Block free stuff” is not an executable instruction I would approve. The exact text, match type and scope matter. My verdict: keep the waste flag, interrogate the estimate, and inspect the proposed exclusion before it touches live traffic.

## Account activity: movement is not management

The page’s fourth highlighted area is account activity, framed in its FAQ as time spent in the account. The independent review describes a dormancy check.

That can surface a useful question: is anyone paying attention? But **activity is evidence of activity, not evidence of good decisions.** A change log full of edits does not explain whether those edits were warranted. An account should not receive busywork because a report dislikes the quiet.

For this section, I want Claude to compare the flag with account activity and performance over the relevant period. What changed? What stayed stable? Is there an unresolved issue that requires action, or merely a lack of recent edits?

My verdict: keep dormancy as a prompt to inspect stewardship. Reject “make more changes” as the remedy. The account does not need exercise rings.

## The missing layer: can the conversion data support any of this?

After those highlighted sections, the report’s wider contents offer more review prompts: CTR against norms, best and worst text ads, structure and best practices. I would use them in the same way: locate an issue, inspect the account evidence, then decide whether it deserves action.

But there is a missing layer underneath all of them. The [PaceAds review](https://paceads.com/blog/wordstream-review-pricing-grader-and-alternatives) says the grader does not check conversion tracking, pacing accuracy, query-level waste, cross-platform allocation, or audience and bidding fit. The broader [comparison](https://paceads.com/blog/google-ads-grader-tools-compared) likewise identifies conversion accuracy and attribution context as common gaps in free graders.

**A report built on recorded conversions does not establish that those conversions are trustworthy.**

Before approving a bid, budget or exclusion based on performance, I would ask Claude to inspect the conversion-action settings and reporting available through groas:

1. Which actions are Primary, and which are Secondary?
2. Which actions feed bidding and the conversion figures under review?
3. When were those actions created, and did reporting change during the period?
4. Do the recorded outcomes match purchases or qualified leads, rather than mixing them indiscriminately with soft events?

Google’s [conversion setup documentation](https://support.google.com/google-ads/answer/16560108?visit_id=639282647458406229-1138916675\&rd=2) explains the Primary/Secondary distinction and notes that GA4 key-event imports can take 24 to 48 hours to propagate. If page views sit alongside purchases in the outcomes being optimized, a reported CPA may not mean what the business thinks it means.

That does not make broken tracking an explanation for a low Quality Score. These are separate checks. I want both inspected, not folded into one convenient diagnosis.

There is also a limit to the connected-account review. Claude through groas can inspect what Google Ads records and how it is configured; that does not prove every website event fires correctly. I would still submit a test lead, check that it reaches the intended conversion action with the documented reporting delay allowed for, and check for duplicate counting. The [10-point audit framework](https://groas.com/post/how-to-audit-google-ads-account-2026-10-point-framework-agency-secrets) covers the wider conversion review.

My verdict: unresolved tracking questions pause performance-driven changes. A cleaner grade can wait.

## From report to live change: require a ticket

This is where the grader’s role ends and groas’s role becomes useful. WordStream supplies the audit prompt. Claude uses connected account data through groas to investigate it, then proposes a specific action for approval. The report does not authorize the write.

**No ticket, no write.** I want each proposed change to contain five things:

1. **Exact change:** the keyword or negative text and match type. `negative exact [free proposal template]` is specific; “remove bad traffic” is not.
2. **Scope:** the named campaign and, where applicable, ad group. Never account-wide by default.
3. **Evidence:** spend, conversions and date range from ACCOUNT, plus the REPORT observation the change addresses. Any unresolved limitation stays visible.
4. **Approval:** an explicit yes to that ticket, not a standing invitation to “fix waste.”
5. **Check-back:** what Claude will retrieve in 7 to 14 days to judge whether to keep or revert the change.

Once approved, Claude can execute the change [through groas](https://groas.com/mcp). Every change is logged with its reason and can be undone in one step. Round-the-clock autopilot remains off until switched on.

Those controls do not make a weak recommendation strong. They make the boundary between analysis and execution explicit, and leave a record of what crossed it. That is more useful than another reassuring color on a report.

## The verdict: keep the questions, reject grade-chasing

I would keep WordStream’s grader as a free triage artefact. CTR against norms can start a creative review. Quality Score distribution can start a relevance and landing-page review. Impression share can identify a capacity constraint. Dormancy can prompt a check on whether anyone is steering.

An [audit-tool roundup](https://www.reportingninja.com/blog/best-ppc-audit-tools) places WordStream in that same role: a free, one-time health check against benchmarks, with detailed fixes and client reporting handled elsewhere. **Useful for triage, useless as an optimization target.**

I would reject the leap from a grade to a live instruction. The independent review cautions against treating the report as a to-do list and notes its route toward a paid upgrade or LocaliQ call. That commercial context does not invalidate its questions. It does make an independent account-level check worth keeping between the recommendation and the write button.

The same structural problem appears in [audits treated as action plans](https://groas.com/post/why-google-ads-audits-dont-work-structural-problem-diagnosis): identifying a symptom is not the same job as choosing a remedy.

So I would keep the grader’s questions, carry them into Claude, and verify each one against connected account data through groas. Then approve a named change with proof attached. The grade starts the work. It never signs it off.

## Frequently Asked Questions

### Can I just make the changes my WordStream Google Ads grade recommends?

Treat the grade as a set of questions, not a change order. The report can identify a question without establishing its answer, so verify each flagged issue against the account's own data before approving a bid, budget or exclusion.

### Do WordStream's millions of reports and billions in analyzed spend prove its advice works?

No. Those figures are context about the tool's age and scale, not validation of its next recommendation. Millions of reports cannot establish whether a specific account records purchases correctly, and a suggested improvement does not show whether it was implemented or helped.

### Does a low Quality Score mean I should rewrite my ads right away?

A low Quality Score earns an inspection, not an automatic edit. The distribution shows where a closer review might start, but it does not say which ad to rewrite or which landing page to replace. Compare the report's observation with the account data for the same campaigns and period before deciding anything.

### Should I raise bids or budget when my impression share is low?

Not automatically. Lost impression share identifies a constraint, not necessarily an opportunity worth funding. Check the campaign-level figures, reporting period and supporting context first, then ask whether the recorded conversions justify chasing the exposure before any bid increase.

### Are zero-conversion search terms always wasted spend I should exclude?

No. A zero-conversion search term is not automatically waste. It may be irrelevant, but it may also involve research, a long sales cycle or an incomplete conversion window. Review each flagged term's spend and conversion history, and classify it as exclude, investigate or retain with evidence before blocking it.

### My report flags low account activity. Does that mean I need to make more changes?

Activity is evidence of activity, not evidence of good decisions. A dormancy flag is a useful prompt to check whether anyone is steering and whether an unresolved issue needs action, but it is not a reason to make changes for their own sake.

### Does the WordStream grader check whether my conversion tracking is accurate?

No. The grader does not check conversion tracking, pacing accuracy, query-level waste or attribution context, so a report built on recorded conversions does not establish that those conversions are trustworthy. Inspect which conversion actions are Primary or Secondary, when they were created, and whether outcomes match purchases rather than mixing in soft events before approving performance-driven changes.

### What should be in a change request before letting an AI assistant edit my Google Ads account?

Require a ticket with five things: the exact change including keyword text and match type, the named campaign scope, evidence such as spend and conversions for a stated date range, an explicit approval, and a check-back plan to retrieve results in 7 to 14 days to keep or revert the change. No ticket, no write.

## Run Your Marketing On groas

Plug ChatGPT, Claude or any AI agent into groas and let it do the work of a whole marketing team: $19 a month flat with every feature included, and autopilot whenever you’re ready.

[Start For $19](https://groas.com/pricing)

[Or have groas run it for you](https://groas.com/done-for-you)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","description":"groas is the marketing infrastructure for your own AI: it connects your AI to all your marketing and gives it proprietary marketing data collected 24/7. Plug Claude, ChatGPT or any AI agent into groas and run all your marketing from it, for a tiny fraction of the usual cost, with the groas app there whenever you want to see more. It is the layer ChatGPT, Claude and any AI assistant run a business’s online marketing on. Through groas an assistant can do anything you could do natively in Google Ads, Bing Ads, Meta Ads and ChatGPT ads, on your website and in AI search, and, where the owner switches it on, can run connected campaigns by itself, so AI agents do the work a whole marketing team used to. groas also gives those agents large-scale proprietary marketing data it collects around the clock, on any schedule the owner chooses: every ad competitors run on Google and Meta, their landing pages and full funnels with the offers and prices in them, Reddit customer research, AI search answers from ChatGPT, Google AI Overviews, Gemini, Claude and Perplexity, and SEO data from Semrush and Ahrefs, backlinks included ($19 a month, flat, with everything included, for unlimited teammates on one shared workspace and unlimited ad accounts and websites, delivered as an MCP server). Separately, its done-for-you service has senior people and proprietary quantitative systems run paid ads, SEO and AI search for businesses, and white label for agencies, from $5,000 a month.","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a#webpage","url":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a","name":"WordStream’s Google Ads Grader, Read by Claude: A Grade Is Not a Change Order","description":"A teardown of WordStream’s Google Ads Performance Grader: what its scores reveal, what they miss, and what Claude should verify through groas before any change.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/378e6704-8f39-460f-a26a-c58d92576cf0.png"},"breadcrumb":{"@id":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a#breadcrumb"},"dateModified":"2026-11-22T05:41:29.991Z","mainEntity":{"@id":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"WordStream’s Google Ads Grader, Read by Claude: A Grade Is Not a Change Order","item":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a#article","url":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a","mainEntityOfPage":{"@id":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"WordStream’s Google Ads Grader, Read by Claude: A Grade Is Not a Change Order","description":"A teardown of WordStream’s Google Ads Performance Grader: what its scores reveal, what they miss, and what Claude should verify through groas before any change.","datePublished":"2026-11-22T05:41:29.942Z","dateModified":"2026-11-22T05:41:29.991Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/378e6704-8f39-460f-a26a-c58d92576cf0.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"account, claude, grader, review, google ads, ads, google, change","wordCount":2082,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"My background includes Goldman Sachs and Stanford Computer Science.","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/wordstream-s-google-ads-performance-grader-read-by-claude-a#faq","mainEntity":[{"@type":"Question","name":"Can I just make the changes my WordStream Google Ads grade recommends?","acceptedAnswer":{"@type":"Answer","text":"Treat the grade as a set of questions, not a change order. The report can identify a question without establishing its answer, so verify each flagged issue against the account's own data before approving a bid, budget or exclusion."}},{"@type":"Question","name":"Do WordStream's millions of reports and billions in analyzed spend prove its advice works?","acceptedAnswer":{"@type":"Answer","text":"No. Those figures are context about the tool's age and scale, not validation of its next recommendation. Millions of reports cannot establish whether a specific account records purchases correctly, and a suggested improvement does not show whether it was implemented or helped."}},{"@type":"Question","name":"Does a low Quality Score mean I should rewrite my ads right away?","acceptedAnswer":{"@type":"Answer","text":"A low Quality Score earns an inspection, not an automatic edit. The distribution shows where a closer review might start, but it does not say which ad to rewrite or which landing page to replace. Compare the report's observation with the account data for the same campaigns and period before deciding anything."}},{"@type":"Question","name":"Should I raise bids or budget when my impression share is low?","acceptedAnswer":{"@type":"Answer","text":"Not automatically. Lost impression share identifies a constraint, not necessarily an opportunity worth funding. Check the campaign-level figures, reporting period and supporting context first, then ask whether the recorded conversions justify chasing the exposure before any bid increase."}},{"@type":"Question","name":"Are zero-conversion search terms always wasted spend I should exclude?","acceptedAnswer":{"@type":"Answer","text":"No. A zero-conversion search term is not automatically waste. It may be irrelevant, but it may also involve research, a long sales cycle or an incomplete conversion window. Review each flagged term's spend and conversion history, and classify it as exclude, investigate or retain with evidence before blocking it."}},{"@type":"Question","name":"My report flags low account activity. Does that mean I need to make more changes?","acceptedAnswer":{"@type":"Answer","text":"Activity is evidence of activity, not evidence of good decisions. A dormancy flag is a useful prompt to check whether anyone is steering and whether an unresolved issue needs action, but it is not a reason to make changes for their own sake."}},{"@type":"Question","name":"Does the WordStream grader check whether my conversion tracking is accurate?","acceptedAnswer":{"@type":"Answer","text":"No. The grader does not check conversion tracking, pacing accuracy, query-level waste or attribution context, so a report built on recorded conversions does not establish that those conversions are trustworthy. Inspect which conversion actions are Primary or Secondary, when they were created, and whether outcomes match purchases rather than mixing in soft events before approving performance-driven changes."}},{"@type":"Question","name":"What should be in a change request before letting an AI assistant edit my Google Ads account?","acceptedAnswer":{"@type":"Answer","text":"Require a ticket with five things: the exact change including keyword text and match type, the named campaign scope, evidence such as spend and conversions for a stated date range, an explicit approval, and a check-back plan to retrieve results in 7 to 14 days to keep or revert the change. No ticket, no write."}}]}]}
```
