Article

Solving Common Issues in Google Ads Accounts

Advertising Analytics Economic Analysis

April 9, 2023

Over the past few weeks, I’ve spent a significant amount of time auditing Google Ads accounts for various clients and potential clients.

One of the things that stood out to me during each of these audits is how many of them have the same core set of issues hindering performance.

For most of the brands we work with, addressing these core issues can dramatically improve efficiency + return on investment – something that’s all the more valuable given the current macroeconomic climate. Today’s newsletter dives into the 7 most common issues I’ve seen in ad accounts (large and small) and provides some guidance on how to address them.

Issue #1: Lack of Research & Audience Understanding

Historically, paid search has occupied a relatively static “bottom of the maze” role, focused exclusively on demand capture. That has resulted in two outcomes: (1) intense commoditization of keywords (and commensurately higher CPCs vs. other platforms) and (2) relatively little emphasis on research + audience understanding (after all, why discover new terms when people just search “buy X” or “consultant for Y”?). The end result? PPCers/brands configure their Google Ads account with keywords they think are relevant, add in some ads, and forget about it.

The problem? ~15% to 20% of searches each day are novel.

Search behaviors change significantly over relatively short periods of time. Long-tail terms (those with relatively few searches each month) are often missed in initial account setups, but deliver outsized value relative to standard or commodity terms.

Iterative audience and keyword research allows brands to systematically identify and promote high-opportunity terms – which elevates the account out of the commodity space (where there’s relatively low opportunity for outsized returns). I recommend using a “Query Flywheel” approach, which is included below:

In an ideal world, your audience research would not stop with people; it extends to your competition and your alternative set. Ask yourself: how often do you Google your own keywords? How well do you know each SERP on which you are paying to place an ad? Does your brand, your product, your offer stand out? If not, why not?

Search is unique among digital channels, in that advertisers have an opportunity to place a specific message in front of a specific person, who is asking one of their most trusted advisors (Google) for help, for guidance or for solutions. As far as marketing goes, every SERP is a golden opportunity. And if you don’t understand *who* is searching and *what* they’re seeing, you’re going to struggle to win over the long term.

Issue #2: Poor Account, Campaign & Ad Group Configuration

The second major issue I see in accounts is poor account configuration – this ranges from improper data capture (i.e. using irrelevant primary optimization events) to enabling Auto-Applied Recommendations (AAR) and Google-Created Assets (GCAs) – resulting in Google running the account (which tends to end poorly).

Google has made turning these settings off fantastically confusing, hiding various settings at different levels of the account (i.e. account settings contain some AARs and Optimization Events; Campaigns contain locations (and some location options are hidden by default); Ad Groups contain ad rotation settings + bidding details (such as CPA and ROAS targets).

There are a number of things to check, but here’s a good starting point:

Conversion Settings (Primary & Secondary)
Offline Conversion Import
Conversion Values (especially important if you’re on eCommerce or DTC)
Conversion Attribution Settings & Windows
Campaign Types + Subtypes; Networks (Search, Search Partners, Search + Display)
Value Rules
Budget
Bidding Strategy
Bid for New Customers vs. Existing Customers vs. Balanced
Locations (Inclusions + Exclusions)
Negative Keywords
Excluded Content
Excluded Types + Labels
Time Zone (make sure it’s in the right one!)
Ad Suggestions
Ad Rotation
IP Exclusions
Auto Applied Recommendations (check Campaign + Account Level)
Ad Extensions (Google can create custom ones for you!)

In short: there’s a LOT of places where what initially appears to be a minor setting can result in staggering ($10k+ per month) in wasted spend, or suboptimal content being promoted, or both. It is always worth it to go through (or have someone else go through) your account to ensure you have the correct settings in place. It is always worth it to go through (or have someone else go through) your account to ensure you have the correct settings in place.

Issue #3: Poor Account Structure

I’m always amazed at how often I still see non-viable account structures used in Google Ads – this ranges from Single Keyword Ad Groups (“SKAG”) to 15+ campaigns, each running with miniscule (<$100/day) budgets, to ad group breakouts based on Phrase Match (PM) and Exact Match (EM), to campaigns running manual CPC with bid modifiers. In their time, each of these things were not wrong.

A major evolution in PPC over the past few years has been that search functions more like programmatic or paid social than ever before – and here, as with there, consolidation is your friend. Automation is not as bad as it once was (and quite frankly, I’ve yet to meet the PPCer who can out-trade a machine).

The accounts that consistently out-perform are the ones that have four things in common:

Consolidated Structure – this is a relatively new development, but it works. Fewer campaigns with higher budgets. Only break out campaigns when there’s a fundamentally different conversion action (since optimization happens at the campaign level). Ensure the budget for the campaign is sufficient to drive at least ~25 (preferably 50) conversions at your target CPA (or within your ROAS target). That’s not always possible (especially for niche B2B); in those cases, consider stacking primary conversion actions (i.e. one for lead submitted with a value associated; one for MQL with a value associated; one for SQL with a value associated). This pushes more data into the machine, while providing a structure that enables it to learn what types of users are most valuable to your business.
Leveraging Automated Bidding w/ Sufficient Volume – properly-configured automated bidding (tCPA + tROAS) almost always beats out Max Conversions, Max Clicks or Manual/Enhanced CPC. I prefer using tCPA for most lead gen accounts, which allows me to adjust priority + budget allocation by moving the CPA target (up = higher priority; down = lower priority).
Strategic Segmentation Within Campaigns – segment only where necessary; I recommend single-topic / single-theme ad groups (“STAG”s), which bundle like terms together. If there’s a fundamentally different message, lander or intent behind the search, consider a different ad group. If you’re trying to build ad groups around different ways to say the same thing (i.e. “WordPress Developer” and “WordPress Agency”) – I’ve not found the juice to be worth the squeeze. Lean into consolidation + make your life easier.
Multiple Targeting Levers + Data Flows – this is the proverbial straw that stirs the drink; keywords are an old lever that is poorly suited to running a modern Google Ads account. Instead, rely on a combination of: (i) topics/themes (your KWs); (ii) smart business data (offline conversions; contribution margins per SKU, etc.); (iii) audiences (particularly custom segments) and (iv) CPA/ROAS targets. The more data you can feed into the machine about what makes a great customer for your business, the more effective smart bidding will be and the more opportunities you’ll have to out-perform your competition.

The final point above – multiple targeting levers and data flows – is especially critical.

Every account should have audiences added, either as “observe” or “target” (based on your specific situation and use case). Every automated bidding strategy makes four core assumptions: (i) completeness (i.e. all relevant information is included in the view); (ii) accuracy (all data is accurate); (iii) continuity (what happened yesterday will happen tomorrow); and (iv) incrementality (all conversions are incremental unless told otherwise). The fourth point above is how you (the advertiser) gain an advantage over your competitors – by giving the system data not included in the standard view (conversions + audiences). Without that, you’re in the same boat as everyone else; with that, you’re able to rise above the competition and realized outsized returns.

Issue #4: Poor Negative Keyword Structures

In automation-rich, rapidly evolving environments (like search), exclusions are often more important than inclusions. For paid search accounts, this means proper negative keyword management, based on your match type. Here’s what I look for:

Use of Negative KW Lists – Negative KW Lists (NKWL) are an incredibly powerful tool for account management, because they simplify your life and avoid situations where you have to go through and manually add the same negative KW to multiple campaigns. I recommend building these at the account level (something you NEVER want to show up for, ever) and the campaign level (for sculpting queries into the right campaign or right campaign level). A well-managed account is likely to have hundreds of negative KWs (usually thousands, but it depends). If there’s only a handful, that’s a HUGE red flag.
Proper Negative KW Mapping – in addition to using NKWLs to simplify management, the next thing to review is ad group-level negatives to keep each of your STAGs relatively siloed. Likewise, a non-branded search campaign should always have a branded search exclusion (usually as a NKWL). These are simple tactics that maximize relevance and focus learning around each concept.
Negative Structure Aligned to Match Type – Google’s Match Type changes (from 2022) have created two different “classes” of matching: query-based and user-based. If you’re using Exact Match (EM) or Phrase Match (PM), you’re using query-based matching. Your negative strategy should reflect that – and aggressively include all of the stuff that’s not relevant to your brand/business. But, if you’re using Broad Match (something I’ve seen perform progressively better over the past 6 months), your negative strategy needs to account for the fact that you are using user-based matching. That means focusing a negative strategy around the absolute must-excludes, and playing a wait-and-see approach with the rest. Remember, with Broad Match, Google is serving ads based on the searcher’s historical activity as well as their query (essentially: they’re making decisions based on data that’s not in your view, so only override where you are confident it makes sense).
Regular Search Terms Report Review – This is an easy one. Just review the search terms report regularly to (1) add non-relevant terms (using the approach outlined above) to the corresponding ad group or NKWL and (2) add relevant terms you missed during your KW research/discovery to the appropriate ad group. If I don’t see any activity in the search terms report, it merits further investigation.
Audience + Customer List Use (where possible) – Finally, Google’s offline conversions upload + customer match are *incredibly* powerful features that I want to see every account using (to the extent possible). A simple check: if you have a search campaign geared toward new customer acquisition, you should be bidding for new customers (a campaign setting). Likewise, if you have a churn reduction campaign (something a lot of SaaS / subscription companies should explore), it should ONLY be focused on existing customers. Audiences are incredible tools for both adding data into views AND focusing the machines on what you want them to do.

Issue #5: Micromanaging the Machine

This one will be unpopular with many old-school PPCers – but micromanaging machines doesn’t work. Daily tweaks in accounts just resets learning and lengthens the time required to optimize. As someone who started when daily tweaks to budgets and bids was commonplace, it’s been a challenge to accept this – but the data doesn’t lie.

That doesn’t mean don’t manage the account (please do!); but instead of tweaking things on a daily basis, batch your changes. Instead of making calls based solely on gut intuition, wait for sufficient data to make an informed decision. Give the machine room to do its job and resist the urge to immediately jump in every time something doesn’t seem perfect.

Think of Google as a Roomba and you as a homeowner. Your job isn’t to micromanage how the Roomba vacuums the floor. Your job is to make sure (1) the Roomba has what it needs to do its job (power, place to deposit all the stuff it vacuums up, etc.) and (2) it doesn’t run over poop (and spread it all around your house) or fall down the stairs.

The easiest way to identify this is to review the change log – if your PPC manager is pushing changes into the same campaigns every day or two, they’re probably micromanaging (and costing you significant money). I can’t over-emphasize the importance of stepping back, compiling changings into bulk updates, and pushing those on a regular schedule. Now, if there’s a mission-critical update that needs to be made (i.e. your product is out of stock, or your website is down, or you’ve suddenly stopped offering a service, or a messaging point is incorrect), yes, change it right away. But for non-mission-critical changes, your manager should be making bulk edits.

The bonus? All of the time you’ll save not meddling in the machine’s business can be used on more strategic (and higher-leverage) tasks, like identifying new searches, creating better ad copy, or fixing your account structure. Win-win.

Issue #6: Ignoring Quality Score

This is directly related to account structure + organization – but a huge unlock for many brands I’ve worked with or accounts I’ve audited. At the most basic level, Quality Score is a core component (along with bid) of Ad Rank (which determines if, and if so, where your ad will show on the search engine results page). Quality Score is Google’s “level the playing field” factor, which enables smaller accounts with higher-quality ads and landing pages to effectively compete against much larger advertisers with much larger budgets but generic or poor assets.

Quality Score is calculated on a scale of 1-10 (1 = very poor; 10 = excellent) at the keyword level, and includes three components:

Expected CTR = this is exactly what it sounds like: the probability that your ad will be clicked if it is included in the SERP.
Ad Relevance = how closely your ad matches the intent behind a users’ search + how closely your keyword is related to other keywords in the ad group.
Landing Page Experience = how relevant + useful your landing page is to users searching for your KW.

Put simply, Quality Score can be a fantastically helpful tool for unlocking new levels of performance across your account. A few ways I recommend using it:

Identifying poor performing creative for ad testing (more on this below)
Breaking up large ad groups with low relevance into smaller, more relevant ones
Consolidating ad groups with similar themes into larger ones
Pinpointing low-performing or low-relevance landers to be updated

As a specific example, breaking apart a large ad group with a bifurcated quality score distribution (roughly 50% of the KWs had QS < 3, with the remaining 50% having a QS of > 5) allowed a brand to reduce CPLs by 15% (this was a result of reducing CPCs thanks to more relevant ad copy and sending each new ad group to a different lander.

Issue #7: Neglecting Ad Copy Testing

Let’s kick off the final section of this with a bold statement: most search ads are flat-out terrible. They’re unfailingly generic, chalk full of redundancy and harp on features that no-one (least of all your audience) cares about. This is a symptom of the very first point above – not knowing your audience and competitive set.

This has been exacerbated by Google Ad’s shift to Responsive Search Ads (RSAs) and deprecation of Expanded Text Ads (ETAs). There is plenty of research (both from within client accounts, as well as by software providers who manage billions in spend) showing that Google tends to prefer RSA combinations with the highest Click Through Rate (CTR), even if that same combination has a much lower Conversion Rate (CVR) or Return on Ad Spend (ROAS).

The bottom line: if you’re not doing ad copy testing, you’re likely leaving money on the table and not getting optimal performance from your account.

While I can (and will) dedicate a full issue just to this topic, here’s what I look for in audits and what you should think about when getting started with Ad Copy Testing:

Mix of Global & Local Tests – there are two overarching approaches to testing ad copy in Google Ads: either you can test within a single ad group (typically one with high volume), or you can test messaging patterns across multiple ad groups (or even an entire account). In general, I want to see a mix of both types of testing within an account – preferably with Global Testing used first to identify patterns, and local testing used within high-volume ad groups to maximize performance.
Cohesive Strategy – the second major mistake is shiny object testing – where random tests are run with no overarching strategy or cohesive vision. The end result is learning that does not generalize. The solution to this is to take a strategic approach, focusing on identifying that purpose behind each test and what consideration factor it is intended to influence (brand, trust/credibility, appeal, value, benefits, scarcity, discovery, synergies/bundling).
Different Types of Tests – Just as there are different testing strategies, there are different testing tactics. Examples of this include: Thematic Testing (testing different messaging themes, such as a discount vs. a benefit vs. a credibility/social proof message); Pinned vs. Unpinned (give G an unpinned asset + a pinned asset, based on what you think is optimal); Structural Testing (where we test different ad structures, such as [H1: Brand] | [H2: Benefit/Proof] | [H3: CTA] vs. [H1: Benefit/Proof] | [H2: CTA] | [H3: Brand] – with each component (H1,H2, etc.) given multiple pinned messages; Message Testing (fix everything except a few messages on a specific structural component) and full Control (pin 3 headlines, effectively re-creating an ETA).
Variant Testing – Wherever possible, you should be using Experiments (specifically, variants) to implement your ad copy testing. This allows you to control the impression split and volume, while keeping other variables constant.
Proper Documentation & Structure – finally, I’m always looking for label usage. Adding relevant labels to experiments or different ad types allows you to easily sum results to get results from Global Ad Testing.

I hope this is a helpful starting point for evaluating your Google Ads Account – and if you have any questions, please feel free to drop me a line. Despite recent changes and an increasing push toward automation, Google is still a complex platform.

Related Insights