Hacker News new | past | comments | ask | show | jobs | submit login
We wasted $50K on Google Search Ads and the insights we got on the way (indiehackers.com)
201 points by igordebatur on Jan 14, 2019 | hide | past | web | favorite | 72 comments

Surprised that keyword selection wasn't mentioned. Any time you're dealing with programming-related keywords, you're going to catch A LOT of people who just want free answers or code samples to copy-paste.

They're going to click on the first search result automatically, and if that happens to be your ad, then you just wasted $0.50-$5.00.

Another issue is keyword match types. If you used the keywords exactly as in the spreadsheet, without any modifiers, then you're giving Google way too much leeway in how they interpret and match search queries to your target keywords. For example, your keyword might be UPLOADCARE REVIEW but your ad will show for the search "uploading review" (I'm just guessing).

You might think "well even if our ad shows for an irrelevant keyword, surely the user would see the difference and not click on it!" Wrong. People click the first X results automatically.

Case in point: A data science platform company I consult once targeted python programmers. Instead they were getting tons of clicks from snake enthusiasts, before I got involved. Oops.

Another company was selling monitoring software for mobile apps but was paying Google (inadvertently) for searchers of free mobile apps. Oops.[1]

Anyway, the other lessons mentioned are still valid, but I feel the biggest lesson (mistake?) was overlooked.

[1] Quick tip if you're running Google Ads now: Go to your Keywords Report and click on the Search Terms tab. You'll see the actual search terms people used before they saw your ad. If they vary wildly from your target keywords, you need to fix your keyword targeting.

PS - More fun "Oops" stories about Google Ads: https://www.gkogan.co/blog/selling-software-snake-charmers/

Was going to comment the same. I've managed a wide variety of campaigns over the years and the best money for jam in terms of optimization is always the search term report. For non-G-Ads enthusiasts, it's the actual list of queries that a user searched for when your ad was clicked. The problem is usually rooted in too many keywords (top of funnel, as their awareness ladder says) or use of broad match keywords--exactly as you said gk1.

My favorite point: "Lesson #1: Test and trim keywords sets before hiring an agency to scale things" - the number one indicator of a successful campaign and client relationship for me is whether the internal team has legitimately tried to use the marketing channel before. I think you can successfully test channels with freelancers but I think a simpler process that likely has a higher rate of success is to test internally before you move it to an external resource.

The core benefit is knowledge acquisition for your team. You'll make better decisions with that knowledge and leverage the time of your freelancer or agency better than if you had to ask about every detail you didn't understand about their work.

Thanks for pointing this: "The problem is usually rooted in too many keywords (top of funnel, as their awareness ladder says) or use of broad match keywords--exactly as you said gk1." Our case is definitely too many keywords, we used Exact Match and Modified Broad Match techniques for different ads in a campaign. Another lesson that we learned is that it's better to prepare content and campaigns from the bottom of the Awareness Ladder, and then shifting up.

There's a subtle point here that I want to emphasize, as it's counter to their "Lesson #2: Focus on page quality and CTR when doing paid tests."

The best way I found to qualify leads was, for things I was charging money for, to put in the price. That way people didn't click if they were looking for free. This actually worked pretty well.

The problem was that this means my ads were showing and not getting clicked. This reduced my quality score (although, in my mind, it provided a better experience for the user because they weren't wasting time going to a site that they did not want).

So Google's Quality Score is "quality" as measured from their point of view (high click/view), not yours nor necessarily your customers.

It's [supposed to be] quality as measured from their users' point of view, which may be a large superset of your customers. If the user wanted a free code snippet to copy & paste and they get an ad with a price in it, that's a bad result, even if they never click on it; that space on the page could've been devoted to a link that they would actually visit. It's good (from the users' POV) for that ad to never get shown again on that query, because it makes their results more relevant.

The right solution here is to refine your target market segmentation & keywords so that you're buying ads on searches that only people in your target market make. Then you get a good CTR & quality score and a good conversion rate. Optimize the funnel all the way down to the searcher.

That might be the idea, but my quality score goes up if the user clicks, even if it's not what they want- and Google gets paid. My quality score goes down if I am forthcoming and pre-qualify the clicks. (There are probably more ways to do this than price.)

If the user doesn't use the word "free" in the search then there's no way to know a priori that they wanted free. By me putting a price in my ad, I am making the user's experience of the search page better by not leading them astray.

Here's an example: Suppose somebody searched "fix my carburetor." The likely return will have services that fix carburetor for you, youtube videos showing you how, and for-pay pdf instructions. Let's say I'm selling the for-pay pdf instructions. How am I supposed to exclude my page? I could limit my keywords to "fix my carburetor pdf for sale" and hope someone types it into the search box, but I find that unreasonable.

Maybe you can walk me through how I'm wrong. But I submit that Google is working on the hypothesis that more clicks = better user experience, and I believe that hypothesis is false.

Not wrong, just incomplete. It's better for you to put a price on the ad than not put the price on the ad and get the bogus click. However, it'd be even better to fix the targeting & keywords and not show the user the ad in the first place if they're not going to click. Google's trying to align your incentives with its users; after all, you're the only one who knows your business well enough to understand if there're subtle differences between what your prospective customers search for vs. what window-shoppers in your industry shop for.

There are often other keywords that can indicate purchasing intent - just thinking back to my own search habits, "buy", "product", "cost", "pricing", "comparison", etc.

And it's true that more clicks doesn't always = better user experience, but it's also true that more clicks in the absence of other information usually = better user experience. Google works off the data it has - for many cases, that's only CTR, bounce rate, and time on site. I suspect (I don't have any inside information how Ads Quality works) that when they have information about conversions - like when you've chosen to link your Google Analytics account with AdWords and set up conversion funnels in the former - they use that information too for determining quality score.

I agree with you. I always try to disqualify traffic with price.

Many people optimize for CTR but that's not usually good when you're doing leads and/or sales.

It's easy to write ads that get higher CTR. Then your CPC will usually go down but that doesn't mean you'll get a higher ROI.

So best to focus on CPA not CTR.

Good point, we now have the target CPL we use to qualify paid ads hypotheses.

Agree on that, the problem was about target market segmentation & keywords. The insight is that we shouldn't focus on a broad market (and test hypotheses for different product parts separately).

That's a good point. We didn't trigger that mechanic because we have a Free plan. However, by doing that we weren't targeting our "ideal customer" which is definitely a paid one.

On the other hand, if we would lower the CTR by defining the price, we would have higher CPCs or lowered the number of impressions due to a low Quality Score. Even if we WOULD get lower CPLs, that won't bring ROI since users still won't activate on our platform.

The problem wasn't about Google Search Ads settings: we hired skilled SEMs and everything. The problem was about the product-market fit understanding and segmenting potential customers based on related value props and keywords.

Got it. Great piece, by the way.

Traffic source matters too. You could end up with a bunch of junk click from the search network while Google.com is doing just fine. If things are really unfavorable, google might start sending you youtube clicks. The more you are able to segment every possible variable, the more likely you are to catch problems and isolate out what is actually working.

I think the most important things are realizing that there is a ceiling on search traffic. For some markets, it is a very, very low ceiling once you’ve optimized out all of the garbage. The giant burst of clicks in the very beginning may be misleading.

VC backed startups often are able to dump a tremendous amount of money in to paid acquisition. If you take an already narrow search channel, the “winner” may end up being whoever raised the most money.

Definitely, traffic source matters. We always analyze the funnel broken down by referrals. However, the case was largely about wrong target market segmentation & keywords.

Yes, it seems like they didn't handle that part of the job properly. From what I read, it looks like they provided an SEM agency with a pre-chosen list of keywords. That doesn't seem like a good idea when you have little experience with paid marketing. In fact, I'm surprised the agency didn't push back to suggest they do that research.

There was a pre-chosen set of keywords and the research as an ongoing process involving the agency. We would quickly trim the set, exclude irrelevant words, etc. The insight we got was that we should advertise product value props rather than test keywords at scale. The research that wasn't in place was the one matching value props with different keyword sets. I mean we scratched the surface of it, but scaling things up was too early in the process.

As someone trying to wet my feet in SEO, what kind of a process would you apply to determine the "best" keywords for a project. I can list a load of terms I'd like to rank highly in, but they're most likely too competitive to actually be pursuable.

Is it possible to figure out what your competitors are doing, obviously without access to their Google analytics? I'd like to see how they've solved their problems, learn from it and know how to differentiate from their approach.

you can execute searches based on your "best" keywords and see which competitor content appears in the search results. that won't give you info on their spend, but will definitely help you understand which type of content converts for the keywords. you will then be able to structure the "content" column of your Awareness Ladder accordingly

Regarding your question specifically, you can gather the stats on your competitors using:

https://seranking.ru/ https://serpstat.com/

The keywords in the Awareness Ladder are just examples of what we used. We did craft the keyword list and the SEM agency helped us filter it. We would dump the irrelevant keywords too. The second point is we used both Exact Match and Modified Broad Match for different ads in a campaign. Here's an example: http://prntscr.com/m7erv2 (CTR/CPC are decent)

I manage 6 figures a year of profitable Google Ads spend, and while this article isn't bad, it's a bit vague. Its strongest point is "don't hire an agency to scale." Its second strongest point should have been keyword selection, but that's buried at the bottom of the article.

Specifically, the things that have gotten us to profitable ad spend in the 6 figure range are:

-- VERY tight campaigns going to tight landing pages. The more specific campaigns you can run, the better. The keywords on your ads AND landing pages need to match the keywords people type in. Most common mistake is driving traffic to a vague landing page. We have 200+ landing pages, all matched to specific keywords. I built out an entire custom framework on top of WordPress to manage this for our company.

-- Manage negative keywords. Every 2-3 days, I log in and add negative keywords. Good ones to start out with: free, diy...I have hundreds now. Maybe thousands. It takes a while to scroll through them.

-- Track sales back and add more negative keywords. There were some items we were getting a lot of clicks on, but didn't make any sales with. I blacklisted a common keyword on one of our campaigns thanks to this insight. Google's recommendation algorithm complained that by blacklisting this keyword, I was getting fewer clicks. YUP. Tracked it for a month, 3-4 clicks per day and 1 sale that entire month. Negative keyword!

-- Start out with 3 ads per ad group minimum. Every week, log in and pause any ad that's not getting good CTR and add a new one. Repeat. After several weeks, you'll be in the "holy ____ I didn't think that was possible" range of CTRs.

It takes a LOT of effort to get to the point where we are now (highly optimized keywords, 12-20% CTR, half the price of most of our competitors.) Still, every couple days I log in and tweak even more things. And I hired a consultant on an hourly basis to get us even better. Totally worth the $ to hire him for a few hours just to see where I could improve!

If you are technical and think you can outsource this to an agency, DON'T. Consistency is key.

One of our partners decided to take the campaigns I'd written and send them to an agency. Their sales dropped 30%. They fired the agency and went back to me. And I'm the business owner, not an agency. No one cares about your money more than you do.

And, being technical, you can do fun things like customizing landing pages based on keywords entered (something most agencies can't or won't do, even though it boosts sales.)

Adding my +1 to the above advice. I ran a high five-digit spend for a product/service I was working on and did exactly the same stuff (although I could have been better with landing pages).

I was amazed at how the biggest RoR was on ads/keywords and not how pretty my site was. Negative keywords were key (free was a big one for me as well) to reducing spend and making ads more efficient.

I was averaging around $3-5 to acquire a customer for a $15 product in a pretty competitive space and it really just took a weekly look at ad metrics to maintain.

Ads aren't going to magically invent demand for your product, but rather magnify it. The smaller the demand for your product, the tougher it's going to be to address the exact segment of people you need to attract to your page, and you may find that more creative solutions are WAY more effective.

> Negative keywords were key

Aren't you using [exact match] making negative keywords not an issue? Or is part of your campaign running both phrase, and broad as well?

For our campaigns, I started with broad match. And then I found that our target market can’t spell. There are also a lot of “ok google” type of keywords in there—not surprising since our market is heavily mobile.

This is going to vary by market, but contrary to articles I’ve read, I’ve found better results from broad match + lots of negative keywords, with an understanding that some small % of ad spend will be wasted.

Interesting. I usually start with broad, then when I feel I've maxed out inventory, I'll expand to "phrase". Typically my clients don't graduate beyond that given how expensive even exact match keywords are.

we used a combination of Exact Match/Modified Broad Match/Negative Keywords. We'll run some tests using Broad Match + Negative Keywords and see how it goes. Thanks for pointing this out

Also adding my +1 here. I've been managing upper 5 figures for a few years on some of the most competitive keywords. My process is pretty similar except I basically swear by BMM (broad-match-modified: +car +accident +lawsuit) keywords only and then the ones that perform exceptionally well I might convert or add an exact match but I can typically fly solely by an entire mix of BMM + negative keywords. This keeps it much tighter and less likely to trip up on wasted spend as all search terms will be close.

Bidding strategy is also incredibly important, as good as Google is getting their recommendation on overall strategy isn't great. Manual CPC (enhanced) has yielded much better conversions and cost per conversion compared to any other strategy. Manually adjusting each keyword, admittingly this does take some time but you set an overall max CPC and then increment where needed. Of course, this can be industry-specific and dependent on your goals. There are also a few scripts I run to automate some tedious things but that's less important.

Like ericabiz I also tried hiring on a few consultants hourly and it's been hit or miss but there were some insights I missed, so potentially worth it. Do not outsource to an agency, they are not worth the money. You need to groom your campaigns over-time and be dedicated. Use that extra cash on bidding for the ads themselves not the overhead of an agency.

You don't know what you don't know. Always be A/B testing.

Great that you joined the thread. Thanks for sharing your experience. See, the thing is we're getting a bit of quite the opposite opinions. Say, on our IH thread you can find opinions that SEM agencies would actually challenge your models, sets, LPs, etc. https://www.indiehackers.com/forum/6ca52123e7?commentId=-LWH...

Negative keywords are the big thing for me. I manage a six figure campaign for an insurance company and have 3,500 negative keywords. Each ppc click costs around $20 to $40.

I would expect a professional SEM company to spend some time generating a reasonable negative keyword list first but the good news is 50k of spend is probably going to have 30k at least of wasted spend to derive negative keywords from!

Wait, 20 to 40 dollars per click of an ad on google search results? Am I understanding that correctly? If so, that seems insanely expensive even for the guarantee that you're getting a potential customer.

When I was at Google [mesothelioma] was the top-earning keyword and it cost something like $40/click.

It makes a lot of sense when you consider that mesothelioma is a type of lung cancer caused by asbestos, is legally actionable, and that there is a whole industry of lawyers who specialize in litigating asbestos cases. The LTV of a customer who goes to trial in an asbestos case can be in the hundreds of thousands of dollars. Spending $40 for a possible lead is insanely profitable for these lawyers.

Yeah, in some industries, that's the cost. Anywhere where lead generation is a thing, you will see high CPC. Insurance is the #1 highest CPC.

:) a nice one, we should've probably named it "How we spend $50K on Google Search Ads to derive a good list of negative keywords."

Aware of our LTV, I would say a $40 click would still be fine if it brings a customer. As a SaaS, we're A-OK with a CPL of $100 and currently target our search ads hypotheses to $15 or less.

Glad to see you on the thread. Would you mind us updating the article with some of your points? I really like how you laid these things out:

— Tight campaigns going to tight landing pages where keywords match what folks entered. — No one cares about your money more than you do. — It's your job to customize landing pages based on keywords entered, SEM agencies simply can't or won't do it.

Reflecting on that, we didn't have enough resources to craft a bunch of landing pages. However, we'll automate the process.

Wow this advice is spot on. Thanks for sharing it. What are your thoughts on outsourcing this if you don't have time to manage it? Is it worth it? Have you ever worked with any agencies that do a fantastic job or is it just really hard to find one?

+1 to this question, I wonder if you can think of an agency that works as both Product Consultant and SEM

Thanks for sharing! Did you automate the landing page creations to match the campaign keywords? If so, how?

This article's title and intro makes it look like it's mainly Google Ads' fault, but then actually it was more due to the landing page design + user journey (awareness ladder) + working with an agency that scaled the execution and the costs. Nonetheless, interesting insights, worth reading for any B2B advertising manager.

Traffic from Google, Bing, and Facebook is easy to buy, the failure is the company for not understanding how PPC works and letting some agency run wild. A hundred things can go wrong after the ad click yet this is trying to save face by saying we should have started small and scaled and limited keywords. That's all yada yada yada, internally they didn't ask the right questions to the agency after week 1, 2, 3, 4 and analyzing the funnel. Then stop the ads, rebuild LP's, restart ppc. If the agency doesn't want to play along, fire them. Poor planning from to start to end.

Agree on the bad planning issue. We should've planned for the long wait we would need to see the actual funnel changes. The issue is about our purchase cycle. It's a long one: on average, it takes users about 182 days to become paid, while the median is 23 days.

Interesting insight, you're right: it's more like "How a bad landing page design can gobble up $50K of Google Ads spend"

> Uploadcare is an established SaaS product, so we’ve been harvesting the low-hanging fruit from organic and direct traffic for a while now. If you’re in a similar situation, you’ve probably reached the same conclusion: running paid ads is the next logical growth step for customer acquisition.

I'm no expert but this strikes me as strange. Organic and direct traffic is not a low-hanging fruit, it's the result of a product well adjusted to its target market. One would think advertising is the low-hanging fruit (relatively speaking); it's what you do when your offer doesn't show up naturally at the top of the results page.

If you're already winning on SEO it's probably harder to improve traffic with ads than when you have nowhere else to go but up.

SEO is a black box, we definitely try to get predictable results from it and scale it further. Our goal was to find a profitable acquisition channel, so we focused on Google Search Ads testing and failed at target market keywords segmentation. Even, we didn't have the resources needed to craft landing pages rapidly. Will focus on automating the process.

Some key things they didn't mention doing:

-Retargeting/Remarketing - Some of your best ROI campaigns will come from these

-Discount Ladders - for customers to nudge over the edge great to use in combination with the retargeting campaigns.

-Abandoned cart email follow ups - No mention of these here. Again, these backend flows will really help profitability.

Ideally you'll break even on those leading awareness/acquisition campaigns using the things others here have said and the retargeting campaigns will be nothing but gravy.

Abandoned cart email follow ups - No mention of these here. Again, these backend flows will really help profitability.

I know this helps you, but it doesn't help me, as a customer.

If I abandoned your cart, there's a reason for it. Hounding me after I've already decided to take my business elsewhere is only going to make me think even more negatively about your company.

What I would like to see if one of these abandoned cart e-mails genuinely ask me why I abandoned the cart. Did I take my business to a specific competitor? Were the prices not competitive? Did I just get too busy and I'll finish later?

Show a little concern for your customers. Pestering them isn't good long-term business.

We have a drip campaign running where we ask users why they didn't activate, etc. The reply rates are pretty low on those. Any idea of a higher-reply-rate channel to ask folks why they didn't sign up/activate/purchase?

You're using the wrong metric. Think quality of data, not quantity.

People who don't reply are already done with your brand. People who do reply are willing to give you a second chance, and are 100x more valuable than any reply rate would indicate.

I mean, we do process those requests individually. I just thought you might have some hints to get higher quantities.

We don't have an "abandoned cart" case (when users bounce, retargeting to every user seems a bad idea)

We ran retargeting campaigns for those who signed up but didn't make a purchase. However, the problem was rooted in activating within a platform (uploading at least one file).

May want to rename to the article headline of 'We wasted $50K on Google Ads so you don’t have to'. As it is, the lack of a comma implies that the insights were also wasted.

If the title of the original webpage wasn't what you suggested another alternative would be to use the oxford comma:

"We wasted $50K on Google Search Ads, and the insights we got on the way"

That's not the Oxford comma, and since there isn't a series here it wouldn't really be appropriate.

Some key lessons:

- Measure ad effectiveness before continuing to pour money into it (denial is a helluva drug).

- Market to your customers, and know who they are. Make sure keywords are targeting the right audience before optimizing click-through.

- All steps in the funnel are affected by the first. If you're bringing in the wrong audience, it won't matter how much traffic you drive and how many ads you place.

Definitely, what you say is correct. However, our point was "even if it seems you're doing everything right, there always are mine traps to avoid." For instance, measuring the effectiveness for us funnel-based would mean waiting for 182 on average and 23 days median.

Still better than the high-schooler guy from my school in Spain (it was news around 2 years ago) who managed to blow up 100k€ or his parents money on youtube ads, and the only lesson there was, was to not give any bank account details to a kid.

fascinating that an ad campaign worked for them. They appear to be selling a dev product, I always assumed all dev's first install was ublock. In facts it was natively installed on all the computers at my university.

Not all developers are averse to advertising. Some understand that the web isn't a free lunch, and without advertising or subscriptions there would be little web content.

Commercial content

I seem to remember the internet being a pretty fascinating place in the 90s before advertising became mainstream (or Google Adwords was a thing). But then it was user generated content and communities.

Interesting comment. We tested different campaigns targeted at developer and business personas. Initially, we started as a product for developers, and those folks can definitely find us themselves judging by Direct and Organic traffic volumes (word of mouth, etc.) I'd also point that it was hard segmenting keywords between those personas, because for us a "business persona" is usually "code-aware": CTO, etc.

Maybe they should have considered this:

Most people who do searches are probably tightly focussed on the answers. Ads are pretty much unwanted distractions in such an environment.

yeah, that's also the case. another insight that we got is we should craft more content answering specific questions rather than advertising our landing page at every step of the Awareness Ladder

Google ads are tough. Very steep learning curve. They work for big brands with deep pockets and want to raise awareness, such as movie studios.

tl;dr: before spending $50k, spend a few hours learning how to manage campaigns[1] and follow the recommendations

[1] https://support.google.com/google-ads/answer/7539883?hl=en

We hired an SEM agency to set up and manage campaigns. Judging by what I read in the thread, they did a good job on that.

Facebook ads with strong sales follow up would probably work better for them.

Did you look at what their service is selling. I've found people are more interested in personal ads when browsing Facebook and don't regularly follow the funnel to make major purchasing decisions for the business they work at while browsing Facebook unless its a small business/entrepreneur where they are ALWAYS in business mode.

My guess is this is because people go to browse Facebook for a short break from work. Or they browse FB in off hours and making a purchase of a product like this would definitely feel like I was in work mode so I would never do it on nights/weekend.

I specialize in B2B FB ads in my consulting career, they work really well. You collect the lead on FB and nurture it by email and phone during work hours.

Agreed, that's another branch of our testing. We'll use FB for lead generation and nurturing those further.

When you collect leads on FB, do you use a landing page as your lead magnet or a piece of content?

+1 to the question, would you use an Awareness Ladder to match purchase stages to content? Like, what's the flow?

fb lead ads

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact