A quick primer
The official diagnostic tool for separating these causes is the set of Quality Score components at the keyword level: Expected CTR, Ad Relevance, and Landing Page Experience. Each answers a different question. Expected CTR — how likely a click is on this ad when shown for a specific query, relative to other advertisers — is closest to a pure creative measure (how compelling the copy is). Ad Relevance — how closely the ad matches the intent behind a specific keyword — is about the match between copy and targeting/semantic structure, not the quality of the writing itself. Landing Page Experience is a separate factor about the page, unrelated to ad copy at all.
Google's official guidance for low Ad Relevance is direct: bring the wording closer to the language of the actual queries, and split ad groups with keywords covering mixed meanings into narrower groups — meaning a low Ad Relevance problem is usually fixed by restructuring, not rewriting headlines. Conversely, if Ad Relevance is "Above average" while specifically Expected CTR is "Below average" alongside weak overall performance, that's a more direct signal pointing toward the creative: the ad technically matches the query, but the phrasing isn't compelling enough against the competitors showing on the same queries.
But a single snapshot of Quality Score components is often not enough on its own — it reflects "right now," not "where" or "how it's been trending." A useful, complementary way to localize the cause is to look at variance across segments (which segments specifically dropped — evenly or locally) and trend over time (has it always been low, or is this a reversal from an earlier improvement).
Where the segment and trend breakdown matters: if you take a depressed conversion rate for an ad group and break it down separately by keyword, landing page, and region, the shape of the variance tells two very different stories. If the drop is roughly even across almost all keywords, pages, and regions at once (no standout outliers), that points toward the creative — the one variable common to all those segments is the ad copy itself. If the drop is concentrated in specific segments — say, only 2–3 landing pages out of ten are dragging the metric down, or conversion only dropped in specific regions — that's a structural, not a creative, problem: a local page issue, a specific geography, or a narrow keyword cluster masquerading as a group-wide dip when you only look at the aggregate. On a small sample, the variance between individual keywords or pages can simply be noise rather than a systematic pattern — so any read needs enough volume behind it before it's trusted.
The same logic applies to the time axis. An overall conversion figure for the ad's whole lifetime can mask the shape of the trend: comparing two stable half-periods can reveal that early improvement gave way to a later decline — that is, the ad isn't "consistently weak," it went through a growth phase (novelty effect, an early tCPA/tROAS learning phase, or a temporarily narrow, high-precision set of queries) followed by a reversal (a smart-bidding exit-from-learning shift, broadened reach into less-targeted traffic through broad match/AI features, or copy fatigue after a long stretch with no updates). Telling these two scenarios apart — "consistently bad" versus "was good, got worse" — matters because they call for different fixes: the first calls for fixing the copy or structure right away; the second calls for first identifying what changed at the point of reversal (a bid strategy change, broadened targeting, the novelty effect wearing off) before deciding whether to refresh the creative at all.
What to check before you touch anything
- The campaign's status — is it flagged Limited by budget/target/search volume? Under those statuses, low absolute numbers are explained by a shortage of impressions, not by copy quality.
- The Ad Relevance, Expected CTR, and Landing Page Experience values for the group's keywords — three separate diagnoses instead of one blanket conclusion.
- The variance in conversion rate across the group's keywords, landing pages, and regions — even (pointing toward the creative) or localized to a few segments (pointing toward targeting/structure/a specific page).
- The trend across two stable half-periods (aligned on full weeks) — consistently weak performance, or a reversal after a period of growth.
- If a reversal shows up — what changed in the account around that point: a bid strategy change, broadened semantics/match types, how long the copy has run unchanged, seasonality.
- Whether there's enough data by individual segment (keyword, page, region) to reliably tell a systematic pattern from statistical noise on a small sample.
Possible approaches
- If the drop is even across all segments and stable over time — work on the copy: test new headlines, diversify messages by function.
- If the drop is localized to specific segments — fix those specifically (the particular page, geography, narrow keyword cluster) without touching the ad copy, which is performing fine elsewhere.
- If a "growth then decline" reversal shows up — first line up the reversal date against the account's change history before deciding whether it's copy fatigue or an external factor (a bid change, broadened targeting).
- If the reversal coincides with a long stretch without any copy updates on otherwise stable settings — that supports refreshing the creative (a planned headline rotation, not a one-off edit).
- If a small data volume doesn't support a reliable segment breakdown — don't draw conclusions from 2–3 keywords with a handful of clicks each; either wait for more data to accumulate, or aggregate adjacent segments for a more stable read.