Quality Score KPI

Quality Score is a critical performance indicator that reflects the relevance and quality of ads in relation to user queries.

A higher score can lead to improved ad placements, lower costs per click, and enhanced overall ROI.

This KPI directly influences advertising effectiveness and operational efficiency, making it essential for maximizing marketing budgets.

Companies that leverage Quality Score effectively can achieve better strategic alignment with their target audiences.

Tracking this metric allows for data-driven decision-making, ensuring that marketing efforts yield optimal business outcomes.

How Quality Score Connects to Your Strategy

Quality Score sits in four of KPI Depot's KPI groups, and the four rosters read very differently, which is the first useful thing to know about it. In Advertising it ranks roughly mid-roster in a priority order of about fifty metrics, behind the group's headline set of Reach, Impressions, Click-through Rate (CTR), Cost per Click (CPC), Cost Per Thousand Impressions (CPM), Cost Per Acquisition (CPA), Conversion Rate and Return on Investment (ROI). In Overall Marketing Department it sits in the lower half of a longer roster led by Cost per Acquisition (CPA), Return on Investment (ROI), Customer Lifetime Value (CLV), Customer Acquisition Cost (CAC), Conversion Rate, Lead Generation, Customer Retention Rate and Customer Churn Rate. In Digital Marketing it ranks near the bottom of a roster whose lead positions go to Customer Lifetime Value (CLV), Return on Investment (ROI), Cost per Acquisition (CPA), Conversion Rate, Lead Conversion Rate, Marketing Qualified Lead (MQL) Conversion Rate, Sales Qualified Lead (SQL) Conversion Rate and Customer Retention Rate on Digital Channels. And in Advertising & Marketing Services, the industry group, it again ranks low, under Click-Through Rate (CTR), Conversion Rate, Cost Per Acquisition (CPA), Return on Ad Spend (ROAS), Customer Lifetime Value (CLV), Customer Acquisition Cost (CAC), Cost Per Click (CPC) and Engagement Rate.

Its balanced scorecard placement is internal process, and it is close to alone in that. Almost every co-metric named above sits in the customer or financial perspectives. So this is the one input-side metric in a set of outcome metrics, and it is genuinely leading rather than nominally leading: the score is an input to the auction that prices a click, which makes it mechanically upstream of Cost per Click (CPC) and, through it, of Cost Per Acquisition (CPA) and Return on Investment (ROI). Movement here shows up in those metrics later, not the other way around.

The tension in Advertising is with the group's two lead metrics, Reach and Impressions. Both reward wide keyword coverage and loose match types, and loose matching is exactly what pulls ad-to-query relevance down. A team told to grow Reach and Impressions is being pushed toward the keyword set that depresses this score, while the group's cost metrics reward the opposite. Note also that Cost Per Thousand Impressions (CPM) buying largely happens on placements where no such score exists at all, so part of this group's spend is invisible to the metric.

In Overall Marketing Department the tension is with Lead Generation. The cheapest way to lift this score, and to lower Cost per Acquisition (CPA) at the same time, is to concentrate on high-intent branded queries where the ad, the keyword and the landing page all obviously match. That harvests demand somebody else created, and lead volume flattens while every efficiency metric in the group improves. This is also a department-wide group covering email, events and organic work, where the score does not exist, so a strong reading here can coexist with a department that is not growing.

In Digital Marketing the tension is with Marketing Qualified Lead (MQL) Conversion Rate and Sales Qualified Lead (SQL) Conversion Rate. The score's expected click-through component rewards ads that get clicked, and the modifiers that reliably attract clicks attract people shopping for free or cheap. Those clicks convert into leads that fail qualification downstream, so this metric and the group's two qualification rates can move in opposite directions for the same reason. That the group ranks this KPI near the bottom is consistent: the group is organised around Customer Lifetime Value (CLV), and the score knows nothing about what a customer is worth after the click.

In Advertising & Marketing Services the tension is unusual and easy to miss: Click-Through Rate (CTR) is the group's top-priority metric, and expected click-through rate is one of the score's own components. An agency reporting both to a client is reporting the same signal twice, once raw and once as an ingredient in a vendor composite. The same double count runs through Cost Per Click (CPC) further down that priority order, since the score's effect on auction pricing already lands there. Treating the two as independent evidence of a campaign improvement overstates the case.

What the differing ranks are telling a customer is worth saying plainly. The closer a KPI group sits to the media buying desk, the higher this metric ranks; the further out it goes, toward department-wide or lifetime-value framing, the more it becomes a diagnostic rather than a target. Read it as the explanation for movement in Click-through Rate (CTR) and Cost per Click (CPC), not as a line on an executive scorecard. And because the rank differs this much, a figure quoted inside one of these groups should not be carried into another group's review without restating what it is doing there.

Measuring Quality Score in Practice

Start by settling which quantity you are actually reporting, because this KPI's name travels badly. The definition and formula carried on this page specify one thing: a rating assigned by a search engine in paid search, computed from the relevance of the ad, the keyword and the landing page together with click-through expectations, and used by the platform to influence ad position and the price of a click. The formula field says the rest out loud, that there is no standard formula. You cannot recompute this metric. You can only read what the vendor publishes. That makes it unlike every co-metric it sits beside in these KPI groups, all of which you can rebuild from your own logs and defend line by line.

Which leads to the first fork, and it is a fork about the name rather than the method. Quality Score is a generic label, and it gets attached to internally defined quality ratings as readily as to the platform metric defined here: an agency's own scoring of creative, a service team's quality rating of delivered work, a review panel's rating of leads. When a marketing department dashboard and a channel report both show a line called Quality Score, there is a real chance they are showing two unrelated quantities, one vendor-computed at keyword level and one assigned internally by people. A customer comparing those two figures is not comparing performance across contexts, they are comparing different metrics that happen to share a name. Establish the owner of the calculation and the unit of analysis before anything else, and if the answer is a human rating, most of what follows still applies but the calibration problem gets much larger.

Where the data lives and how to join it honestly. The platform's reporting interface or API is the only source, and it exports the score at keyword level. Join on the platform's keyword identifier together with match type, ad group and campaign, never on the keyword text. The same text in two match types, or in two ad groups pointing at different landing pages, is a separate scored entity with a separate score, and text-level joins silently merge them. Two more join hazards. The landing page component is assessed at page level and is shared by every keyword pointing at that page, so those rows are not independent observations, and any statistic that assumes they are will look far more precise than it is. And the score does not exist across the whole account: search partner inventory and display placements are not scored, so a join that treats the account as one population is quietly mixing scored and unscored spend.

The composite has hidden weights. The definition names the components. It does not name their weights, and platforms do not publish them. So when the overall reading moves you can usually see which component changed, but not how much of the movement it caused, and platforms generally expose the components as coarse labels rather than as continuous values. That has two direct consequences for reporting. Component labels can sit unchanged while the overall reading moves, and the overall reading can sit still while a component quietly degrades toward its next label boundary. Never present a movement in the composite with a causal explanation attached unless you can show the component change and the account change that produced it, on the same dates.

The scale is ordinal, so do not average it. This is the trap that ruins most reporting of this metric. The score is a position on a small bounded integer scale defined by the platform, and the distance between adjacent positions is neither constant nor published. Arithmetic on ordinal positions is not meaningful: a mean of those positions is a number you can compute and cannot interpret, and a change in that mean cannot be read as an amount of improvement. The honest summaries are distributional. Report the shape of the distribution, and report the share of impressions, clicks or spend sitting in each band, because spend weighting is what connects the metric to money. A simple average across a keyword list answers a question nobody asked, since it weights a keyword that took a handful of impressions the same as the keyword carrying the campaign.

What gets scored is not a random sample of your account. Platforms only assign a score once a keyword has accumulated enough recent activity, so new keywords and thin ones sit in a placeholder state, and paused keywords hold whatever value they had when they stopped serving. The scored population is therefore endogenous: it is the set of keywords that ran, at the budgets they ran at. Two consequences worth being blunt about. Pruning low scorers raises every account-level summary immediately without changing anything about the account's relevance, which makes the metric trivially gameable by deletion rather than by improvement. And a keyword throttled by a shared budget entered few auctions, so its score rests on a thin basis while appearing in your reporting with exactly the same weight as a keyword that served all day.

Calibration drift, and revisions after the fact. Because the vendor computes this, the vendor also recalibrates it. Component definitions change, scoring windows change, and reporting changes. A step change that lands across the whole account on one date is almost always a platform change and almost never your campaign. Date-stamp the series and keep an annotation log of platform announcements beside it, or you will eventually present a recalibration as a result. The revision problem is separate and just as damaging: the score is recomputed against a trailing activity window, so the platform reports the current score rather than the score that was in force during the auctions you are analysing, and pulling the same historical date twice can return two different answers. If you need a defensible time series, snapshot the score yourself on a fixed cadence and report from your snapshots. Anyone comparing a report from last quarter against a fresh pull is comparing two computations, not two periods. Where the metric is a human quality rating instead, the equivalent exposures are rater variance and drift: the same work scored by two raters, or by one rater months apart, gets different scores, which needs periodic calibration sessions and a re-scored held-back sample to measure rather than assume.

Segmentation that actually matters. Branded against non-branded is the split that changes conclusions, and the tracked sources make the point for us by scoping to branded keywords in the first place. Report the two separately, always. Then match type, since exact and broad matching produce structurally different relevance. Then device, because the landing page component is evaluated on the experience the platform sees, and a page that reads well on desktop and badly on a phone shows up here as a mystery. Then landing page, since one page's problem propagates to every keyword pointing at it. And separate mature keywords from new ones, because a new keyword's score is mostly a platform estimate rather than a measurement of your account.

Instrumentation traps specific to this metric. The landing page component reacts to changes another team made, so a site release, a consent banner, a tag manager container that slowed the page, or a redirect chain will move this metric with no marketing change at all, and the movement will arrive weeks after the release. A page that is blocked from crawling or geo-restricted will register as a poor experience even while converting fine for the traffic that reaches it. Page speed as the platform's crawler measures it is not page speed as your real-user monitoring measures it, and when the two disagree the platform's view is the one that moves the score. Finally, watch for the reporting artefact of your own account hygiene: a quarter in which the team consolidated ad groups and paused underperformers will show a jump here that is entirely composition, and if that jump is presented next to a Cost per Click (CPC) improvement caused by the same consolidation, the causal story writes itself and is wrong.

Common Pitfalls

Many organizations overlook the importance of Quality Score, focusing solely on ad spend without considering relevance.

• Failing to conduct regular keyword audits can lead to outdated targeting. This results in ads that do not resonate with current user intent, diminishing performance over time.
• Neglecting ad copy testing prevents continuous optimization. Without experimentation, companies miss opportunities to refine messaging and improve engagement.
• Ignoring landing page quality can hurt scores significantly. If users find landing pages irrelevant or confusing, they are less likely to convert, negatively impacting overall metrics.
• Overlooking negative keywords can waste budget on irrelevant searches. This leads to lower click-through rates and diminished Quality Scores, as ads are shown to users who are not in the target audience.

Improvement Levers

Enhancing Quality Score requires a strategic focus on relevance, user experience, and continuous optimization.

• Regularly update keyword lists to reflect changing market trends. This ensures that ads remain aligned with user intent, improving click-through rates and engagement.
• Conduct A/B testing on ad copy to identify high-performing variations. Testing different headlines and descriptions can reveal what resonates best with target audiences.
• Optimize landing pages for user experience and relevance. Ensure that content matches ad messaging and provides a seamless transition for users, which can boost conversion rates.
• Utilize negative keywords to filter out irrelevant traffic. This helps to focus ad spend on users who are more likely to convert, enhancing overall campaign effectiveness.

KPI Depot is trusted by consulting, strategy, finance, and analytics teams at leading organizations worldwide, including those listed below.

Quality Score Benchmarks

We have 2 relevant benchmarks in our benchmarks database.

Browse the Top Benchmarked KPIs in Advertising

Reading the Benchmarks for Quality Score

Two sources are tracked for this KPI, HawkSEM and WordStream via KPI Depot, and both records are classified the same way: threshold. That single word does most of the work in this section. A threshold is a cutoff somebody recommends, a line above which they say an account is healthy. It is not an observed average, not a median, and not a distribution. Reading a threshold as though it described what accounts actually do is the most common mistake made with this metric, and the classification on both records says plainly that neither figure is a central tendency.

The populations then diverge in a way that matters more than it looks. One record is scoped to branded keywords, the other to ads. Those are not two samples of the same thing. Search platforms assign this score at the keyword level, so a figure framed per ad is already an aggregation somebody chose to perform, and the choice of how to aggregate is unstated. Branded keywords, meanwhile, are the highest-scoring slice of any account by construction: the query names the advertiser, the ad matches it, the landing page is the advertiser's own, and click-through expectations are high for reasons that have nothing to do with account quality. A bar set on branded keywords is not a bar for the non-branded terms where the work actually is.

Beyond that, both records are thin in the same places. Both are labelled cross-industry. Neither carries a sample size, a geography, a company size, a time period, or a stated formula. One carries no source date at all. And the second source name, WordStream via KPI Depot, is itself a flag: it is a relayed figure, so the originating publication is where the methodology would have to be read.

Three things to settle before trusting any external figure for this metric, whichever source it came from:

• The unit of analysis, and whether branded traffic is in it. Per keyword, per ad, per ad group, or per account are four different numbers from the same account. Ask also how the aggregation was weighted, because an unweighted average across a keyword list gives a keyword with almost no traffic the same say as the keyword carrying most of the spend, and the two weightings can point in opposite directions.
• Whether it is a recommendation or an observation. Both records here are thresholds. If a figure is a target somebody advocates rather than something measured, the only honest use for it is as an opinion to argue with, and comparing your account against it says nothing about where your account sits relative to other accounts.
• Which platform, and which vintage of its scoring. The score is computed by the ad platform, not by you, and platforms recalibrate their components and their reporting without announcement. A figure from an earlier vintage of a platform's scoring is describing a calculation that no longer exists. The available records give a customer no way to pin this down, which is precisely why the benchmark data here is stored source-attributed with its dimensions attached rather than presented as one number.

OKRs That Use Quality Score

None of the four KPI groups this metric belongs to name Quality Score as a key result in their OKR examples. Advertising, Overall Marketing Department, Digital Marketing and Advertising & Marketing Services all build their examples out of volume, cost, conversion and retention metrics. So the honest framing is not to promote it to an objective of its own but to attach it to objectives those groups already state, as the mechanism behind them.

The closest fit is the Advertising group's objective to maximize brand exposure while efficiently managing advertising spend, which carries Cost Per Thousand Impressions (CPM) and Cost per Acquisition (CPA) among its key results. For the paid search portion of that spend, this score is a direct lever on what the auction charges, so it belongs under that objective as a supporting key result rather than a headline one. Because the scale is ordinal, phrase it as a shift in distribution: increase the share of search spend flowing through keywords in the account's upper score bands, reported separately for branded and non-branded terms, with the group's existing Cost per Acquisition (CPA) key result as the outcome it is meant to move. The group's own guidance to segment metrics by platform and campaign type applies directly, since an account-level reading of this score hides the segments where the work is.

The second framing comes from Advertising & Marketing Services, whose objective to enhance advertising precision and creative impact to increase campaign effectiveness is built from Ad Targeting Accuracy, Ad Creative Effectiveness, Ad Viewability Rate and Ad Recall Rate. That is a relevance objective, and this metric is the search channel's vendor-issued verdict on relevance, which makes it a natural diagnostic beside those key results. A directional key result a team could set: raise the relevance component of the score across the non-branded keyword set while holding the paid search share of qualified pipeline steady, so the gain is not purchased by retreating to branded traffic. Two cautions from the group's own best practices belong with it. That group advises refining Ad Targeting Accuracy before scaling spend, which is the same argument for treating this score as a gate on scaling rather than as a growth target. And it warns against leaning on technical metrics such as Click-Through Rate (CTR) without weighing ad content relevance, which is a reason to pair this score with Ad Creative Effectiveness rather than let it stand alone, particularly given that click-through expectation is already inside the score.

One boundary worth stating for either framing. Digital Marketing's guidance to balance Cost per Acquisition (CPA) against Customer Lifetime Value (CLV) is the check on both of the above: this score can be pushed up by narrowing to easy traffic, and a key result written on it needs a lifetime-value or qualification metric sitting beside it to catch that.

See OKR Examples for Advertising

Free Whitepaper

Want to achieve performance excellence in Advertising? Download our in-depth whitepaper: Definitive Guide to Advertising KPIs.

Download the Free Guide

KPI Categories

This KPI is associated with the following categories and industries in our KPI database: