ArticleWhat To Do When Ads Don’t Work
Despite the fact that we’ve had more than a year of stuff happen since January – new administration, trade wars, trade deals, AI developments, unlikely alliances, drama (writ large) and, yes, a glorious Philadelphia Eagles Superbowl win – this year still has a lot more to go. There’s no two ways about it: this has been a roller-coaster year, and it is showing no signs of slowing.
The sheer velocity of stuff doesn’t just impact news and headlines, either – it has a real, tangible effect on advertising. Perhaps more now than at any point I can remember since the height of COVID more than 5 years ago, I am hearing a constant refrain from senior marketing leaders, CMOs, executives, investors and founders: our ad account isn’t working anymore. Our ads aren’t converting like they did in the past. Our site isn’t performing as it once did. Something’s wrong.
On one hand, this isn’t uncommon to hear from various verticals and certain segments of the industry that things aren’t working. That’s business as usual. Something is always broken for someone. What’s unusual is that, at least according to the 100+ people I’ve spoken with since May 1, most things are broken for most people.
The data doesn’t quite bear this out, at least according to our client portfolio. But, when the data and the anecdotes disagree, I side with Jeff Bezos over Robby Mook – I trust the anecdotes. If enough people are saying something, there’s probably a reason, and I (and our clients) are better served by my being curious about it vs. dismissing it.
This all brings me to the topic for today’s issue: what do you do when it seems (right or wrong) that your ads aren’t working? I’ve been in this position too many times to not have a playbook to get through it.
The first – and most difficult – step is this: recognize that 90% of what happens inside your ad account(s) is a direct result of everything outside of it. The macro environment, your product, your sales team, your customer support/customer success teams, your competition, your offer, your business data, your messaging, you name it. The worst-kept secret underlying remarkable marketing that everything matters. This is not a cop-out; it’s not abdicating responsibility or an invitation to blame other stuff for your account not performing. To the contrary, this is a wake-up call for you – the marketer, the founder, the investor, whatever – to recognize that you must take a step back and look at all of it. The answers you must find likely live outside of Google Ads or Facebook Ads or LinkedIn Ads.
Once you’ve accepted that, it’s time to get to work.
Step #1: Clearly, Succinctly & Simply State The Problem
There’s an old saying that a problem clearly & simply stated is half solved. While I may have my doubts as to whether it’s truly half-solved, the fact remains that the actual problem with clarity and specificity is a remarkable accelerant toward identifying a solution. When I’m on calls and this topic (ads not working, writ large) comes up, I begin asking a lot of questions:
· What, specifically, has changed with your marketing performance?
· When did you first notice the change? Has it been consistent or sporadic?
· How long has this been happening?
· Have any consumer-level metrics (i.e. AOV, LTV, CAC, MQL:SQL ratio, SQL:close ratio) changed during this time vs. the preceding period? If so, which ones and how much?
· Did this coincide with any other changes to product, price, offer, etc.?
· Are certain products/services more impacted than others? Which ones?
· Can we review your historical performance from the same times in 2023 and 2024?
· Has your competitive set shifted? Did any new competitors enter the space?
· How are your competitors marketing today?
· Do you know of – anecdotally – anyone else in your space that is experiencing similar issues? If so, did they begin around the same time?
· Have there been any changes to what your sales/customer support teams are hearing on their calls?
· Are you aware of any increase in negative reviews or negative customer sentiment? Have any new themes emerged in recent negative reviews vs. past ones?
I’m quite confident this feels about as fun as a colonoscopy for the person(s) on the other end, but the fact is that this level of questioning is essential. Marketing is too complex, too all-encompassing to go on wild goose chase after wild goose chase. The more you can do up-front zero in on the specific issue that is fueling this perception, the better.
Throughout this exercise, you must ask follow up questions. If the person on the other side of the conversation isn’t getting a bit (or quite a bit) annoyed, you’re probably not asking enough questions. Even when you get a seemingly satisfactory answer, probe for more. Ask “why”. Request more examples or anecdotes. Re-state the same question in a different way, compare the answers you’ve received, and if there’s a difference, ask more.
Once you have completed this step, you should be able to state the problem with wonderful clarity: “Since April 10, PPC leads are down 34% vs. the prior period and 41% vs the same time last year. Total Google spend is up 29% vs. the prior period and 31% vs. the prior year. Lead qualification rates (MQL:SQL, SQL:Close) have also declined by 18% and 17%, respectively. This has resulted in 52% fewer new customers from PPC from April 10 – now vs. the prior period, and 57% fewer new customers from PPC vs. the same time last year.”
If you are not at the point where you can articulate something similar to the above, you are not done asking questions. You may need to screen share. More than likely, you’ll need to pull up GA4 or Shopify or HubSpot or Salesforce. Block more time than you think you’ll need, because this is going to be annoying, frustrating and exhausting. It may seem excessive and over the top. But trust me when I say, it isn’t. From a psychological standpoint, it is much more comfortable to be vague about problems, because then both accountability for the problem and responsibility for the solution can be equally obfuscated. If everything is blurry, then nothing is clear – and no-one is to blame, least of all the person who is (ostensibly) trying to “solve” the problem. Is this a bit pessimistic? Absolutely. Is it also right 9-in-10 times? Also absolutely.
The beauty of getting to the clarity above is that it enables you to focus everything you’re going to do next on the core drivers of the problem. No wasted effort. No vague solutions. Continuing with the above example, it’s obvious we have at least four big-picture questions to answer:
1. Are all leads down, or is it just PPC? If all leads are down by similar numbers, that could indicate a broader macro issue, a competitive landscape issue, or an offer/product issue. If it’s the latter, the most likely cause is an issue in the ad account.
2. What is driving increased Google Ads spend? Spend being up ~10x inflation is a red flag – it could be competitive pressure (more bidders in the auction = higher CPCs). It could be wasted spend or low-quality KWs. It could be that you haven’t checked the SERPs and have missed something.
3. What are the reasons for the lower qualification rates? Is sales putting notes into each DQ’d lead? Are there any commonalities or emerging trends in what sales is being told (i.e. rates are too high right now, I’m worried about spending this much right now, I’m talking to 4 different vendors, etc.)? Are leads being DQ’d at a roughly equal rate across all your salespeople, or is it uneven? If possible, pull call recordings, chats and/or email exchanges for DQ’d leads to see what’s going on.
4. Are the lower qualification rates uniform across all leads, or just PPC leads?This is one that most marketers (especially PPCers) ignore – but it’s actually one of the most critical questions to ask. We tend to assume that if there’s one PPC problem (i.e. fewer leads) that other problems are also PPC problems. Our brains are evolutionarily hard-wired to create these kinds of assumptions, whether or not they’re based in reality – because better safe than sorry. Back in the caveman days, most of the people who got too curious were quickly rendered unable to contribute further to the gene pool. Fortunately, curiosity about marketing has a fantastically low fatality rate.
Step #2: Confirm It’s A Real Problem
There’s a pronounced tendency among certain kinds of people (particularly marketing people) to jump to the conclusion that something is wrong, when it might not actually be wrong, or it might just be statistical noise.
That’s why you should have a forecast. Once you understand exactly what’s going on, your immediate next step is to validate not simply that it is happening, but it is an actual issue. A material deviation from the expected. In general, the shorter the timeframe where the issue is occurring, and/or the fewer incidents, the more likely it is to be noise.
This is particularly evident with brands that deal in high-ticket sales (i.e. $5,000 chairs) or lower lead volumes (i.e. full-home renovations or a particular niche procedure or a certain sub-set of professional services) – when you only get 20 sales/leads a month, volatility is likely to be higher, and the impact of that volatility is more severe.
Note: just because something appears to be noise does not mean it isn’t a problem – and you should still investigate it. But there’s a material difference between a brand that ordinarily gets 20 sales a month getting 14 (more than likely represents noise, normal variance and/or idiosyncratic factors) and a brand that typically sells 20,000 units only selling 14,000 (highly unlikely). In both cases, the percent decline is the same, but the sheer volume of the magnitude of the change in the latter case makes it much more likely that the decline is a result of material operational, supply and/or demand issues.
Step #3: Work Backwards From The Data Through The Experience
This is a weird thing to do, but it is effective precisely because it breaks the pattern. Most of the faults I’ve found in traditional diagnostics/audits are a result of either confabulation or preceptive completion. If you’re unfamiliar with either term:
Confabulation: our brains simply invent stuff that’s not there, with no intention of deceit (i.e., I could have sworn I added those new KWs or excluded those audiences or updated those ads).
Preceptive completion: when a person’s brain unknowingly fills in gaps using either specialized or readily-available knowledge. The most common example is when your brain “spell corrects” as you read – you don’t even notice the misspelled word because your brain has near-instantaneously adjusted the letters to their correct positions. Where this becomes dangerous is in the case of specialized knowledge – most marketers have been through their customer’s journey 1,000x. They know their messaging and sales pitch to the syllable. They know exactly how every element – from the ad to the lander to the CRM – is supposed to work. And that is exactly what creates the problem – as they’re reviewing (whatever), their brain fills in the gaps using the specialized knowledge they have. But, when a customer/client arrives at the same point in the experience, he or she doesn’t have that same specialized knowledge to fill in the gap – so s/he gets stuck, gives up, or tries another method (i.e. calling or emailing).
The way to get around this is to go in the opposite direction of the normal flow, because then your brain doesn’t have the ability to do any of this. Coincidentally, this is why it is exponentially easier to forge a signature if you turn it upside down, then copy the lines. Any subconscious thoughts about how letters should look or connect together are deactivated by the mere act of inverting the signature, so none of them creep into the copy (legal disclaimer: don’t forge signatures).
From a marketing standpoint, this means beginning at the end. Create a list of DQ’d leads during the time in question. For each one (or a sampling, if there’s too many), review everything they did: how they contacted your organization; what pages they visited prior to submitting the lead form; what keyword(s) they searched; which Meta Ads they saw. All of it.
Next up, do exactly what they did in the immediately preceding step – if they submitted a lead form from an Android phone, go find yourself a Samsung phone, head on over to the lead form, fill it out in the same exact way the DQ’d lead did, and submit it. Then go through the experience that they did between the end result (being DQ’d) and this action (i.e. submitting the lead form on the site). This likely entails your scheduling and attending a demo/sales call, along with receiving a few emails. Do all of it (and yes, it helps to have a few burner gmail accounts).
As you progress through this process, you’re going to simultaneously validate a few other things:
· Data Accuracy – once you’ve submitted the lead form (and while you’re waiting for Stan the Sales Dude to reach out), head back to the CRM and confirm the form has appeared and is correct. Pull up G Ads, Meta Ads + GA4 (and whatever else) and confirm that a form submission event has been received. As you progress through the journey, continue to check back into the CRM and confirm each touchpoint is recorded as it should be (e.g., emails from Stan the Sales Guy should be populating, call notes from Stan should align with the actual content of your conversation, etc.). If that’s all true, then you can feel fairly confident that there aren’t any data gaps (note: you’ll likely need to test a few before you can come to this conclusion).
· Messaging Fidelity – There are few things more counterproductive to converting prospects into customers than that prospect being “sold” something on a landing page, only to talk to a real person who tells a completely different story. This is why we’re going to talk to Stan the Sales Guy. Bonus points if you can pull the actual call log from the DQ’d lead and give Stan similar answers to that person.
· Timing – Note the time you submit the form and the time the next thing happens (for instance, you get an email from Stan offering to set up a call). Compare that timing to the timing in the CRM. If it’s much faster (and, make no mistake, speed to lead is a real thing), then investigate further. Are salespeople slower in responding to PPC leads vs. direct/organic/referral leads? If so, why? When did this begin? Is it all salespeople, or just some?
· Product – if you’re working with an eCommerce brand, actually ship yourself the product (assuming it’s practical to do so). When you get it, inspect the packaging. Use the product. Make sure that everything the customer is getting is aligned to what they expect based on the ad, lander, etc. The same is true for a SaaS business – one of the biggest drives of churn is a clunky, non-intuitive or otherwise messed-up experience.
Assuming all of this checks out, go back to the stage preceding this – review the preceding website pages for any conflicts, oddities, etc. Assuming that all checks out, search for whatever KW they searched for, from wherever they searched. Check out the SERP – are there any strange competitors on it? How compelling is your offer vs. the competition?
If you’re doing this right, you’re likely going to find all kinds of weird stuff, some of which you previously had no idea was going on. You might notice that a competitor has a staggeringly better offer than you do, or that an alternative just rolled out a new product feature, or that another firm launched a new site. Note all of this – but keep on this one journey until you can’t go back any further.
Note: I’m aware that the above example is predominantly lead-gen focused, but the same principle and approach applies to eCommerce/retail/SaaS. The only thing that changes is what you’re looking at – start with the evaluate checkout first, then PDP-cart, then preceding pages. Everything else should stay the same.
Again, this may seem odd, but over the years, doing it this way was the only way I found:
· A call center employee selling inbound leads to a competitor (yes, really)
· A sales team that was told to respond to PPC leads last by a CSO
· A customer success group that didn’t want to honor current promos because they made less commission
· A random CRM rule that a previously DQ’d lead who submitted an otherwise qualified inquiry at any subsequent point would never be passed to the sales team
· A post-conversion flow that sent a confirmation for the wrong product in Shopify
· Another post-conversion flow for one specific product set that had the “track” link accidentally swapped with the “cancel” link (gee, I wonder why returns were up and sales were down)
· A website chat that denied the existence of current offers and refused to connect customers to the sales team for [reasons].
And so many more random, conversion-killing issues. Having specifics is what enables you to work backwards effectively; working backwards is what allows you to actually spot the problem, whether it’s hiding in plain sight or lurking in a random corner of Klaviyo.
Step #4: Assess The Landscape + The Macro
Whether or not you’ve found anything wrong, Step #4 is definitely one you should be doing. You’re going (to the extent possible) to see what the problem customers/clients/prospects saw during the time in question. To do that, we’re going to do the following:
· Search for relevant + related KWs from their location (use a VPN or SE Ranking’s Google Location Changer). See what other competitors are coming up, what they’re offering, what’s included on their LP, etc. If you want more info on how to conduct real, effective competitor research, check out Keeping Up With The Joneses.
· Visit both your and your competition’s social media – watch some of your competitors reels on IG and shorts on YouTube. Check out whatever they’re posting on TikTok. Review their LI content. Do the same for your brand. Not only will this help you identify some historical issues, but seeing a bit more of your competition in your feed will allow you to spot future issues faster – especially ones where the competitor has a new product, a better offer, etc.
· Check out Facebook’s Ads Transparency Center, Google’s Ads Transparency Center, the YouTube Ad Library, etc. – see what else your competition had in the market during the time in question.
· Go to Google News + conduct a search for the time in question. See what articles/stories were trending in the local market during that time (in just one example, a chain of mid-Atlantic clinics had a precipitous decline in new female patients right after a major scandal involving a fake doctor taking illicit photos of patients came to light. In another, an eComm brand selling primarily to hobbyists, farmers & craftspeople saw revenue absolutely collapse in mid-late August, specifically around Minnesota, Iowa, Wisconsin & Michigan. They were convinced ads were the issue….right up until we reviewed the news and found that all four states happened to have their State Fairs going on). All marketing is local marketing – and sometimes, the reason why leads or sales are down is a random local story that large corporations simply ignore.
· Once you’ve checked the local news, do the same for the national news (and X). This is also a great use-case for Sparktoro’s consumer information – fire up the wayback machine and go visit the news sites that your primary target audience actually trusts. Was Fox News posting non-stop about something related to your industry? Did the Washington Post run an investigative report on how roofers scam people just before your inbound pipeline started struggling?
In many cases, you’ll find that there’s not just one thing that’s leading to the issues you’re facing – it is (more than likely) an amalgamation of multiple issues – a faulty site or sales experience, some negative local or economic news, and some suboptimal performance from your ad accounts – that’s ultimately causing the issue. Most marketers stop once they find a single problem that can explain the issue at hand. But, by going deeper, you can often find much, much more that will allow you to not simply resolve the issue, but to avoid it again in the future.
Depending on the situation, it is also often worth examining reviews for your brand (is a new negative review trend popping up or an issue you’ve never heard of suddenly included?), along with those of your competition (do people suddenly love their new product, or is there a new flaw with their product?). If you don’t have reviews, pull customer service/customer support tickets – are customers calling in about a problem more often than before? Are they referencing a particular pain point more than before (i.e. price, asking for discounts, saying “I’ll have to check with [significant other] after getting price information)?
These things are often connected in weird ways – perhaps there’s a niche issue with your product. Your competition has picked up on it and is running pseudo-positive ads directly addressing it (a pseudo-positive ad is something like “99% Uptime Guaranteed” when your competition has recently had an outage – it’s positive on the surface, but anyone who is in the market or knows what’s going on in that particular market is well aware of what you’re doing). Looking at either of these in a vacuum (the niche issue and the competitor ad) might not explain a massive drop in sales, but put them together and you have a recipe for significant impact.
Step #5: Review The Ad Accounts
To this point, we’ve focused on stuff that’s outside of the ad account – the sales team, the CRM, the website, competitors, macro, etc. Each of those has a disproportionate impact on the performance of the ad account (Google, Meta, whatever), but the performance can never be entirely explained by outside factors. There’s still the 10% that is entirely controlled inside the account – so we need to review that, too.
Here’s a top-line on exactly how I conduct Google + Meta Ads audits:
1. Data, Tracking & Attribution – you’ve already validated some of this, but it’s time to go deeper. Ensure that all relevant events are tracking, the correct “primary” events are added to each campaign, the right attribution windows are in effect, and that the right counting mechanism is in place (i.e. don’t count every form submission, count only the first). Confirm that all enhanced + offline (Google) or Conversions API (Meta) events are firing correctly and pushing back the correct data + values.
2. SERP + Feed Audit – actually search for your keywords or browse through a potential customer feed. While this is imperfect (ad platforms now tailor content to individual user preferences), it will help you see how your creative looks in the wild, along with what other competitors/alternatives are offering. This is more critical than ever for Google Ads given the roll-out of both AI Overviews and AI Mode – these features are absolutely cannibalizing search ad placements in the short-run. One example: Google Ads stopped serving ads for “truck accident lawyer” across the US following the terrorist attack in New Orleans way back in January, through about May (they’re back now). Performance for a number of PI forms cratered when it happened, but most had no idea why – Google continued matching other searches (like Truck Wreck Lawyer or 18 Wheeler Injury Attorney) to “truck accident lawyer” – so the volume dipped, but didn’t collapse.
3. Account Structure – I’ve seen far too many accounts that are massively over-complicated, resulting in fractured data, limited insights and far too many unintended consequences. In our current age of automation, simplicity tends to perform best for most brands – aim for maximum sustainable (i.e. how many keywords can I put into this ad group before the marginal benefits from segmentation become too overwhelming to ignore) instead of minimum viable (i.e. could I just run an ad group with this one KW in it?). On Meta, the same principles apply – focus on one campaign per angle/offer, use multiple ad sets (interest stack, LAL, broad) with relevant creative variations in each.
4. Bidding Strategy – I tend to be a major proponent of automated strategies for MOST campaigns (with exceptions – like Manual CPC for branded). Whenever possible, I avoid “highest volume” or “maximize clicks” because they optimize for the wrong thing (i.e. spending my money or getting clicks). My go-tos are Cost Caps/Bid Caps (on Meta) or tCPA / tROAS (on Google), with the primary optimization event being the thing that is most important to the business and the target being aligned to the forecast. Adjust tCPA/tROAS/Cost Cap/Bid Cap on an Ad Set / Ad Group level.
5. Control The Machines – ensure that auto-apply settings, including automated creative optimization, auto-apply recommendations (Google), url expansion (Meta + Google), audience expansion (primarily Meta, but also Google) are under control. The single-largest source of waste in most accounts is platforms using automated updates in ways never intended by the brand/user.
6. Isolate Brand vs. Non-Brand – don’t allow branded search to creep in to your PMAX or non-branded search campaigns – this conflates performance and obscures incrementality. Use brand exclusions and brand exclusion lists to minimize the presence of brand in your campaigns, and negative out any brand searches that squeak through.
7. Review The Search Terms – related to the above – check your search terms report. As Google has gotten more fickle in how they match queries to keywords, you may find that the queries you think you’re appearing for and the queries you’re actually appearing for are quite different. A simple example of this is competitors – Google is matching non-branded keywords (i.e. best retinol serum) to branded queries (i.e. Clinique, Drunk Elephant, L’Oreal). I’ve seen it go as far as matching a competitor’s office phone number to a non-branded keyword. Check your search terms.
8. Examine Your Ad Creative – Particularly on Meta, creative is king. The best ad accounts have three things in common: (1) a robust diversity in style, content and medium (i.e., a mix of statics, videos & carousels, showing UGC, lifecycle, graphics, comparisons, founder stories, etc. – if you’re not sure what you should include, check out this issue.); (2) a healthy rotation – ads introduced each week (or more frequently), winners iterated on in rapid succession, false positives pruned. The simple fact is that the half-life of most “hit” ads is now a few weeks to a few months. Yes, there are exceptions (I have an account where the #1 best performing creative has been going strong for nearly a year, despite ongoing and sincere efforts to de-throne it) and (3) authenticity – consumers today can detect BS a mile away. They know when they’re being sold a bag of baloney. If your creative doesn’t feel real and relatable to your audience, it doesn’t matter how much you make or how often you update the account – it’s not going to work.
9. Landing Page Optimization – this isn’t exactly in the ad account, but it counts nevertheless. If your landers don’t connect to your creative and effectively close the deal, it doesn’t matter how wonderful the ad creative or clever the keywords. Review your landers. Make sure they’re relevant to the ad and they connect with the audience.
10. Audience Management – audiences are still wonderfully overpowered levers on both Google + Meta; smart media buyers can use them to reach more of the audience they want and avoid most of the audience they don’t. For Google, make sure audiences are added as “observe” for campaigns – this puts more data into the view and gives G more signals to consider when optimizing. For Meta, ensure that customers/site visitors are segregated into their own ad sets (it’s never going to be perfect), so that prospecting ad sets can serve to net-new people. Adjust your tROAS / Target Return (Meta) up (tCPA / Cost Cap / Bid Cap down) for remarketing/current customers, so that platforms don’t over-deliver to existing customers. Take advantage of Google’s “bid for new customers” feature where practical.
11. Ad Schedules + Location Settings – I’ve audited so many accounts where these basic settings – when ads ran, where they ran, what devices ads showed on – were misaligned to the brand. It does not make sense to run ads saying “call now” when the call center is closed. It does not make sense to run ads for an iOS-only App on Android. It does not make sense to run ads for a service business in an area that the business does not serve. Is it basic? Yes. Does it often explain hundreds (or thousands, or sometimes millions in wasted/underperforming spend)? Also yes.
12. Experimentation – any agency or freelancer that wants to “test everything” is probably an idiot – there are plenty of established, low-risk/high reward practices that don’t need to be tested in an account. What you want to see are strategic, high-reward tests being run on a regular basis – new landers, new offers, bold changes to messaging, new audiences, etc. If that kind of structured, big swing testing is happening, that’s a good sign. If it isn’t….not so much.
13. Use Rules/Scripts – These are oldies, but they’re goodies. Most successful Google Ads accounts still have some old-school scripts rolling to serve as checks on the machines. This can be as simple as “stop loss” rules that turn campaigns off should their spend exceed a certain amount, or as complex as triggers that adjust pricing or targets based on inventory. If the latter, test the script to confirm it’s working as intended. If the former, confirm it has worked / fired in the past.
The above is the quick version; if you want a full deep-dive on my approach to account audits, check out this issue.
Step #6: Get To Work Fixing It
This one is self-explanatory: work like hell to fix the issue. Avoid blaming and finger-pointing; just solve the problem. There’s no shortage of time once performance has reverted to normal to address the underlying issues and ensure they don’t happen again – but until that time comes, work the problem. I’ve been on far too many teams where attitudes, finger-pointing and blame-games have subverted sincere efforts to address the root causes of the problem.
That’s all for this week!
If you have any specific questions, reply to this email or DM me on X. I hope this week’s issue was helpful to you. Next week, we’re going to talk about AI – more specifically, about how the hype + the on-the-ground reality have diverged, and what you can do about it.
Until then, enjoy that summer sun!
Cheers,
Sam
