First, decide whether the rating matters: read the ad's results
Open the ad group and put the rating beside the results for each ad over a period with enough clicks to read: click-through, conversions, conversion rate and cost per conversion. Three situations come up. A new ad with no history and a Poor rating has nothing else to judge it by, so the rating's suggestions are the best available guide. An established ad with a Poor rating and weak results is probably weak for the reason the rating gives. An established ad with a Poor rating and the best results in its group is telling you that the rating measures something other than what you are paying for.
Then read the suggestions attached to the rating, which name the gap: add more headlines, include popular keywords, make headlines more distinct, make descriptions more distinct, unpin assets. Each maps to one cause below. The responsive search ads guide explains how the ad is assembled and how assets should be written; this page decides which ratings to act on and in what order. Where the question is how ad strength relates to the keyword's Quality Score, the comparison of Quality Score and ad strength separates the two ratings.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Suggestion to add more headlines or descriptions | Number of assets against what the format allows | Too few headlines or descriptions | Add distinct assets until the pool covers each message role |
| Suggestion to make headlines more distinct | Whether headlines are rewordings of one or two ideas | Headlines repeating the same idea | Replace near-duplicates with headlines that make different claims |
| Suggestion to include popular keywords | Ad group keywords against the words in the headlines | Ad group keywords missing from the headlines | Write several headlines that use or paraphrase the keyword theme |
| Suggestion to unpin, with several positions fixed | Pins per position; whether each pin is required | Several pins limiting the combinations | Remove optional pins; pin several alternatives to a required position |
| Keyword suggestions conflict, and no headline set fits all of them | The spread of meanings among the ad group's keywords | Ad group mixing intents, so no headline fits every keyword | Split the ad group by intent, then write a pool for each |
Fix, verify and prevent each cause
A pool too thin to combine
A responsive search ad works by testing combinations, and an ad with only a handful of headlines has few combinations to test. The rating flags it, and the flag is fair: the system has little to choose from for different queries and devices.
Fix: add headlines and descriptions until each role in the pool is covered several times, keyword headlines, proof, offer, call to action and brand, writing each to stand alone, as the responsive search ads guide lays out in its table of roles. The help page on responsive search ads documents how many assets the format accepts and how many it shows at once.
Verify: the rating moves up as assets are added, and over the following weeks the asset report rates the new assets rather than leaving them all in learning.
Prevent: draft the full pool before the ad is built, rather than launching with a few lines and promising to add more.
Many headlines, one idea
Many ways of saying "affordable plumbing" are still one headline for the purpose of testing. Near-duplicates fill slots the system could have used to try a different message, and the rating says so. This often happens when headlines are written by varying a template one word at a time.
Fix: group the existing headlines by the claim they make, keep the best phrasing of each claim, and replace the rest with headlines that make different claims: a qualification, a guarantee, a speed promise, a price frame without figures that change.
Verify: the rating's distinctness suggestion disappears, and the combinations report shows a wider spread of headline pairs serving.
Prevent: write the pool from a list of messages, not from a list of phrasings, and have someone other than the writer check it for repeats.
Headlines that never mention what people searched
An ad group built around a keyword theme, with headlines that talk only about the brand or generic benefits, gives the system no relevant assets to match to the query. The rating's keyword suggestion fires, and the same gap tends to show in the ad relevance component of Quality ScoreGoogle's 1–10 keyword diagnostic..
Fix: write several headlines that contain or paraphrase the ad group's keywords in different forms, and check that the landing page uses the same words; the Quality Score guide explains how ad relevance is judged, and the Quality Score help page describes the components.
Verify: the keyword suggestion clears, and over the following weeks ad relevance for the group's main keywords holds or improves.
Prevent: start every new pool with the keyword headlines, before benefits and calls to action.
Pins that fix too much
Pinning forces an asset into a position, and every pin removes the combinations that would have put something else there. Google documents that ad strength takes pinning into account, so an ad with several pins rates lower even when its copy is good. Some pins are necessary: a mandated disclaimer, a brand name that must lead. Many are not; they exist so the preview reads nicely.
Fix: list each pin with its reason, remove the ones without a business or regulatory reason, and where a position must hold a type of message, pin several alternative assets to it so the system still has a choice, as the responsive search ads guide recommends.
Verify: the unpin suggestion weakens or clears, and click-through (CTRShare of impressions that result in a click.) holds or improves over the following weeks.
Prevent: require a reason for every pin in the ad build notes, and accept a lower rating where the reason is real.
An ad group whose keywords want different things
Sometimes the rating cannot be fixed inside the ad, because the ad group holds keywords that want different answers: repair and installation, a product and its finance, a service and a job vacancy. Headlines relevant to one are irrelevant to the others, the keyword suggestions conflict, and any pool that covers everything says nothing clearly.
Fix: split the ad group into intent groups, sets of queries that want the same answer and can share a landing page, as the account structure guide describes, then write a pool for each.
Verify: each new ad group's ad reaches Good or better with fewer assets than the mixed group needed, and cost per click (CPCAverage cost of each click.) on the moved keywords holds or falls as relevance improves.
Prevent: add keywords to an existing ad group only when its current headlines would already suit them.
Prevention: an asset brief per ad group
Most Poor ratings come from ads written in a hurry inside the interface. A short brief per ad group, written before the ad, prevents almost all of them. The brief states the intent the ad group serves, the keyword theme in the words people use, the landing page, and the messages the pool must carry, each with several candidate headlines: theme, proof, offer, action and brand. It lists any mandatory wording and where it must appear, which is the only legitimate reason for a pin. With that in hand, the pool is complete and varied on the first build, and the rating confirms it rather than prompting it.
Pair the brief with a review rule. Build every ad to Good or better as a floor, because a Poor rating on a new ad nearly always means the system has too little to work with. After launch, stop managing the rating and manage the results: read the asset report on a schedule, replace assets rated Low, leave learning assets alone, and keep the responsive search ads guide open for the writing itself. If an ad's rating falls because a required pin was added, record the reason beside the ad so the next person does not undo it. And when the brief cannot be written because the keywords want different things, that is the signal to restructure before writing anything.
An accountancy firm's Poor ads, fixed and left alone
An accountancy firm has two search ad groups rated Poor. The first, "self-assessment tax return", has one ad with 6 headlines, four of them variations on "Tax Return Help", and no pins; it has a 3.1% click-through rate and a £48 cost per enquiry, the worst in the account. The team adds 9 headlines covering a fixed fee frame, accreditation, turnaround and a call to action, and two descriptions with distinct details. The rating moves to Good; over six weeks click-through rises to 4.6% and cost per enquiry falls to £36. The second, "brand", rates Poor because the firm's name is pinned to position one and a regulatory line to position two. It has a 22% click-through rate and the lowest cost per enquiry in the account at £9. The team leaves it as it is and notes the reason for both pins.
Common mistakes
Chasing Excellent with filler headlines. Unpinning a line the business is required to show. Replacing a proven ad to move a rating. Writing variations of one headline instead of different messages. Trying to fix a mixed ad group with copy. Treating the rating as a performance score.
