Optimising and Debugging Paid Advertisements

Ways metrics mislead you; how to overcome low reach/low quality score/high CPC etc.

This article is part of my Entreprenerd: Marketing for Programmers book, which is currently available to read for free online.

What Advertising Platforms Want

Online advertising space is sold by a rather strange species of auction. Whenever you bid for something in a typical auction—for example, with a real-estate auctioneer or on eBay—your bid is what you end up paying. With online advertising auctions, things are different: Your bid is the maximum you could be asked to pay, but your actual bill often turns out to be significantly less, at least when your particular mix of ad creatives, targeting, and ad relevancy helps an advertising platform maximise their total income.

Let’s take Google AdWords as an example. Given that Google typically charges only after someone clicks on an advert, it stands to earn the most by showing adverts that have both high bids per click and high click-through rates. If I bid a million pounds per click for my ad, Google will eagerly display it to great swathes of their users, hoping for that million-pound pay-out. But if no one clicks my ad, Google eventually grows weary and thinks, “Screw this guy, we’re not making any money on him because no one is ever going to click on his stupid ad”. Google will gradually cease showing my ad and instead start showing someone else’s, even if this other ad only offers to pay a paltry 70 pence per click. But if people actually click on this other ad (i.e., if it has a decent click-through rate), then a steady stream of 70-pence pay-outs will trickle into Google’s coffers and it will end up earning more money in the end, despite the comparatively tiny bid offered for each click.

Many advertising platforms also care about maintaining a great user experience for their own users. Whenever a searcher thinks that Google’s ads are useful and relevant, this increases the likelihood that this searcher will continue choosing Google as their go-to search engine. For this reason, Google wants to encourage advertisers who create relevant adverts. This is done through its Quality Score metric. (Other platforms have equivalent concepts, such as Facebook’s Relevance Score.) The higher the Quality Score an advertiser achieves, the greater their reach and the deeper the discount they earn in their advertising costs.

Stay Empirical

A word of warning before we get started: You should always check up on your campaigns after “optimising” them. Some time ago I studied all the best practices and integrated them into one of my campaigns. Having thought that I had brilliantly optimised my advertisements with all the latest tricks, I left the campaign alone for three months, feeling satisfied that I had done a great job.

But when I next logged in and looked over the results, I saw a 50% reduction in conversions. I learned an unsettling lesson that day: The best practices are not necessarily the best practices for you. In a field this complex, there aren’t any certainties that you can apply without a second thought—and this is true even if the source of your best practices is the advertising platform itself (e.g., Google AdWords’ official blog).

I should have been more sceptical and scheduled a check-up session soon after making my changes; I should have double-checked things with my own eyes. By not doing so earlier, I lost a lot of business.

How Much Should I Start Off Bidding With?

There is a dreadful uncertainty about how much to bid when beginning a new advertising campaign. With no historical data for you to refer to, you only have the platform’s “recommended bid” to go by.

In business, as in life, only a fool pays the first price suggested to him in a negotiation. True to this truism, the initial bids recommended by online advertising platforms often inflate prices to a degree you would only expect at a bazaar situated in the most notorious of tourist traps. Knowing this, I recommend you set your opening bids to one quarter the amount the platform initially suggests. Because you are lowballing an algorithm rather than a human being, your aggressive negotiation tactics carry no risk of offending. You won’t sabotage the deal, as might happen at the bazaar; the worst that can happen is your advertisements won’t display until you raise your bids. Don’t rush this: If your adverts are not displaying after an hour, that doesn’t mean your bid is too low. Wait a day before making a better offer.

Why is it important to start low and raise your bids from there? Doesn’t this have the same effect, you ask, as starting high then reducing your bids until they are just above the point where your adverts stop displaying? The disruptive factor which stops these two strategies from being equivalent is the presence of competition. Had you started with a high bid, your adverts would have displayed from the get-go, at the expense of your competitors’ campaigns. These competitors would gradually notice that their advertisements had stopped showing, and then, intending to claw back their traffic, they would raise their bids above your initial, greatly inflated bids. Compare this to the situation where you start with a low bid and then raise your bids until they are marginally above your competitor’s. Even after your competitor eventually responds to your stealing of their traffic—which might take weeks or even months—they will only nudge their bids slightly above yours. This slow tit-for-tat bidding war could go on for a long time, meaning that it may take months or even years before the bids are raised to the level they would have started at had you begun by bidding high (i.e., at the platform’s “recommended bid”). By starting with low bids, you slow down competition-induced inflation of advertising costs. Or, to put in another way, whenever you start with a high bid, you prematurely raise the costs of advertising for everyone in that market.

What’s the Most I Should Ever Bid?

As time goes on, bids for online advertising tend to rise and rise, mostly due to the unrelenting effect of mounting competition. There comes a point when you start asking yourself, “Should I raise my bids again, or have I gone far enough?”

The answer to this question is conceptually simple but technically difficult: Never bid more than your expected profit for that click, taking into account the minimum margin you wish to keep for your business efforts. This calculation requires that you have such excellent analytics and understanding of your business that you know the average profit you earn on each click. Without the ability to estimate this, you’re flying blind.

Be warned that in certain crowded markets the most you can afford to bid will be insufficient to beat the bids of your competitors with deeper pockets. In this case, your only option is to become more efficient (e.g., by improving your advert click-through rates or on-site conversion rates, or by lowering your costs of production such that you can afford to pay more in advertising costs).

When Metrics Mislead You

Online advertising platforms provide you with a plethora of easily available metrics such as CPM (cost per 1,000 impressions), CPC (cost per click), and CPA (cost per acquisition). But when it comes to profitability, these metrics can be a bit loose with the truth. Unless you watch out for their pitfalls, they may lure you down expensive cul-de-sacs.

Without further ado, here they are, starting with the most distorting then ending with the least:

1. Misleading CTR (click-through rate)

The click-through rate measures the percentage of impressions that resulted in a click. It’s a rough proxy for the compellingness of your advert vis-à-vis the audience seeing it.

On its own, a high click-through rate isn’t all that useful a metric. Its weakness is that it is totally divorced from your business’s finances, making it pointless to optimise as an end in itself. If you have a mighty high CTR, say of 50%, this doesn’t necessarily mean that your advert was a commercial success. The question of commercial success depends on what you paid for those clicks. If they cost you £100 a pop, then your advertising was a disaster, in spite of your sky-high CTR. If, on the other hand, the clicks only cost you £0.05 a pop, you’re doing great. But in both cases, the metric that mattered was the CPC (or, if we’re being more thorough, the cost per acquisition, etc.)

The main use of the CTR is in comparing the relative performance of your advert variants. This helps you determine which creatives were more compelling, arming you with the know-how to tweak and improve your other ads. These improvements in CTR should eventually lead to lower CPCs, thanks to mechanisms like Google’s Quality Score and Facebook’s Relevance Score.

2. Misleading CPM (cost per 1,000 impressions)

A cheap CPM is misleading when it’s associated with a dismal click-through rate. Sure, your low CPM proves that you convinced the advertising platform to show lots and lots of people your advert for only a small sum of money. But if no one clicks on these ads, this also proves that these legions didn’t care a damn for your adverts…those thousands of impressions left a dreadfully poor impression.

This combination of excellent CPM and terrible CTR could happen for all sorts of reasons: Perhaps your advert creatives were unconvincing. Perhaps your choice of targeting mechanism was way off and the wrong audience saw your adverts. Or perhaps your adverts were placed so far south in a scrollable page that no one even saw your advert, despite the advertising platform registering an “impression” and charging you for it.1

By the way, I’m not saying that having a cheap CPM combined with a low CTR is always bad. If the goal of your campaign is to establish your brand name in people’s minds, this would be exactly what the doctor ordered.

3. Misleading CPC (cost per click)

First off, congratulations on creating a compelling advertising campaign that drives traffic to your website at a low cost! But before you start celebrating, take a quick look at your on-site conversion rates. If this low-cost CPC traffic isn’t converting into paying customers at a healthy rate, you’ve got serious problems…

We need to separate out two possible pathologies before we go any further with our diagnosis. The first is the benign case where your on-site conversion rates suck in general. All your traffic, no matter what its source, has the same (dreadful) on-site conversion rate. In this case, your low-cost CPC isn’t misleading at all. Although you have your work cut out for you in improving your overall on-site conversion rates, your cheap CPC has helped you make the most of your bad lot. You haven’t been misled at all.

But if you discover that your low CPC traffic converts on-site at a worse rate than other traffic (e.g., traffic from other advertising campaigns or from SEO), then your attractively priced CPC has led you astray. This happens when there is a mismatch between what your advertisement promises (or suggests) and what you actually provide. If your adverts cause people to expect brand new laptops but you only sell used ones, the traffic you paid for will quickly abandon without converting. Likewise, if your audience expected free products but you only offer paid ones, they won’t stick around for long.

These problems can be abated by clarifying your offer in your advert creatives. For example, if you’re worried about too many freebie hunters clicking your adverts, you could include your product’s price inside your advert body. This newly modified advert will almost certainly depress your CPC because now the people chasing freebies will be deterred by your price tag. But the clicks you do receive will be from people comfortable with paying for your products, and they are sure to convert at a higher rate. As your CPC drops, your CPA should soar.

In some cases, you’ll need to change your targeting to repair mismatch problems. I lost more money that I’d like to admit by advertising my products in India and Pakistan. The problem was that my product wasn’t suitable for these markets—they would have needed a local adaptation. I should have excluded these countries through targeting.

4. Misleading CPA (cost per acquisition)

Cost per acquisition represents the price you pay the advertising provider for each conversion (most typically, the price for each sale). The weakness of this metric is that it states nothing about the magnitude of the conversion. Maybe all those cheaply acquired conversions were for abnormally small basket sizes. You’ll know this is happening if the basket sizes for this particular campaign compare unfavourably to your averages.

A friend of mine who owns an ecommerce store advertised a “loss leader” product. (This is when you sell a product at such a low price that you lose money; the goal is to convince people to come to your store and buy other, higher-margin products during the same visit). The advertising campaign for the loss leader had an excellent CPA, but after analysing the figures in detail he discovered that the customers won through this campaign purchased nothing other than the loss leader. The seductively low CPA figure obscured the commercial truth that his campaign only succeeded in reaching an unprofitable, bargain-hunter audience.

It must be emphasised that although a low CPA can mislead, often it won’t. For example, businesses that have only a single possible basket size (e.g., ebook sellers, retailers of paid apps) always have the same conversion magnitude, so therefore the CPA is a realistic and accurate measure of their advertising effectiveness.

5. Misleading ROAS (return on advertising spend)

ROAS is the amount of revenue generated by every pound spent in online advertising. This metric accounts for variations in basket size which are missed by the CPA, making ROAS a superior choice whenever you want to take the pulse of your advertising campaign’s success.

But it is not perfect. One way an attractive (i.e., high) ROAS can mislead is when different orders have uneven profitability levels (e.g., because of the effect of variable tax rates, shipping costs, inventory costs, or some other factor). Suppose that Product A has a profit margin of 60% and Product B only 30%. An order for £100 of Product A would now be worth twice as much to your bottom line as an equivalently large order for Product B. But this difference would be completely missed by the ROAS, since this only sees the £100 in revenue earned in both cases.

There is a way to somewhat mitigate this shortcoming: Instead of basing the calculations on revenue, you base them on profit, or some approximation thereof. For example, you could subtract sales tax, shipping, and inventory costs from your gross revenue, then see how much of this reduced figure was generated by your advertising spend. This should give you a more accurate picture than basing the ROAS metric on simple revenue.

If you want to be finicky—which is generally a good thing when working with business metrics—you should know that even this improved ROAS has a blind spot: It fails to account for the total lifetime value of your visitors. Customers tend to become repeat customers, and any ROAS measurement linked to individual orders will fail to capture the total value a customer will bring to your business over the long term. Perhaps customers found through adverts on a website for a trade journal buy ten times as many widgets in the coming years compared to customers found through other ads. An ROAS measurement won’t be able to tell you this, because its entire understanding is limited to the information contained within the initial order.

There are possibilities for integrating elements of lifetime value into your ROAS metrics, but these can be rather challenging to implement. One is configuring your website and Analytics to stitch together distinct user sessions belonging to the same person. Another is leveraging technologies like the Analytics Measurement Protocol to tie together future transactions and account for sales that happen outside the normal website channel. For more on these two, refer to the chapter on Google Analytics.

Overcoming Low Reach

Perhaps the most frustrating problem in the realm of online advertising is when your ads are unable to get enough airtime to spend your budget. Even the most brilliant advert ever concocted will fail to deliver any conversions if no one ever sees it. This failure to receive as many impressions as desired is known in advertiser jargon as having “low reach”.

i. Targeting woes

When debugging low reach, the first thing to determine is where the fault lies. One possibility is that you targeted too narrow an audience and maxed out on all its advertising potential. The easiest solution here is to loosen your targeting—are you sure there aren’t any unnecessary targeting restrictions in place? Once you’ve ruled this out, the next thing to do is seek out new targeting mechanisms in a bid to reach your intended audience from other, as of yet untried angles. You might target new keywords in Google Search, different subreddits over at Reddit, or just released real-world behaviours on Twitter. You might even try a new platform altogether, such as Snapchat Ads.

Roughly speaking, your search for new targeting angles will lead you to consider more general targeting options, more specific ones, and ones that are laterally related to your original idea. More general” would mean that instead of targeting “christmas cards”, you target “gift cards” or even just “cards”. “More specific” would mean the inverse; now you target “reindeer cards”, “santa cards”, and “christmas cards for kids”. The goal is to accumulate a larger audience by targeting many specific niches. “Laterally related”, then, just refers to any other relation that somehow connects back to your original targeting intention, even if that relation is tentative and tangential. Here you might target “christmas wrapping paper” in the speculative hope that there’s enough overlap between wrapping paper and cards for this advertising to return a profit.

Because the number of things in the world is increasing—be they brand names, fields of study, Facebook pages, cultural memes, or newly minted words—there’s always the opportunity to be the first mover into a targetable semantic space that other advertisers have yet to reach. By keeping up with the popular culture of your domain, you might be able to stay ahead of your competition.

ii. Overaggressive filters

Another targeting-related cause of low reach is the effect of inappropriately eager filters that act to categorically block your adverts from showing. I have in mind campaign-wide negative keywords, bans on adverts appearing on mobile devices, filters that prevent adverts from showing below the fold, and “exact match” keywords which are used when a broader match would have been more appropriate.

iii. Tough competition

Low reach can also occur when someone else is offering the advertising platform more money than you are to reach the same audience. You’ll know this is the culprit when your metrics indicate that your targeted audience is plenty big, but you’re only reaching a tiny proportion of it.

One such metric is Google AdWords’ “impression share”. This measures the proportion of times your adverts were shown relative to those of others competing for the same advertising slot. When your impression share is low, this tells you that your adverts aren’t getting as much airtime as they could.

Over on Facebook you can infer that competitors are outbidding you by comparing the estimates of your reach (shown when keying in your targeting data) with the number of people you end up reaching. Large discrepancies between these two figures indicate that it’s competitors and not your choice of targeting that is the cause of your problem. A word of warning though: You shouldn’t be too eager to attribute your low reach to competitors. Because of the prevalence of ad blockers, a certain percentage of your audience—perhaps even the majority—may never see your ads, even after a decade of you trying to reach them.

Anyway, assuming that you’re reasonably confident that your case of low reach is caused by competition, the solution is to increase your bids and optimise your advert creatives so as to improve click-through rates/Quality Score/ad relevancy.

iv. Low Quality Score

Google uses the Quality Score metric to decide whether your adverts are even eligible to appear for a given search query. As such, a low Quality Score can directly lead to a low reach. What’s more, even if your Quality Score is high enough for your adverts to appear, it may not be sufficiently high for your ads to appear in any respectable position above the fold. This is because Quality Score also determines your ad’s position. And because many web visitors never scroll down to the lower results in a Google search, this means that low-ranking advertisements will rarely be seen, thereby further reducing their reach. We’ll discuss ways to improve Quality Score in a separate section below.

Other platforms have concepts similar to Quality Score (e.g., Facebook’s Relevance Score), but it seems like these alternative metrics only affect your cost of advertising. There is no clear indicator that low quality scores on these platforms would prevent adverts from displaying in the same way that a low Quality Score on Google does.

v. Tightly rationed audience

This is the most benign cause of low reach. Sometimes your choice of targeting means that you’ll only catch tiny glances of your underlying audience every day; it may be unrealistic to expect even half of them to appear online within the reach of your adverts after even a few months. The cure to this problem is straightforward: Simply leave your advertisements running for a longer time. This approach can be fine-tuned by taking steps to push your advertising platform into showing a greater proportion of your adverts to new people (e.g., by using built-in per-person frequency capping or creating negative remarketing lists populated with people who’ve already visited your website).

Those who study problem solving in the abstract have observed something that’s relevant to everything we’ve said so far about increasing reach: You can never really solve a problem; you can only replace it with another one—one that you (hopefully) consider more tolerable. Whenever you take measures to increase your reach, you often decrease your advertisement’s cost-effectiveness with that same stroke. You should anticipate needing to make this trade-off.

Overcoming Low CTR

Your click-through rate (CTR), calculated by dividing the number of times your advert was clicked by the number of times it was shown, is both a measure of the appropriateness of your targeting and of the compellingness of your advert creative in relation to that particular audience.

What counts as a good CTR depends entirely upon your business: 0.5% could be amazing for a mass-market ad for toothpaste, whereas 5% could be laughably bad in a niche market for yacht repair.

A better way to gauge success is to see how one particular ad’s CTR compares to that of your other adverts. Suppose that you have eight adverts running in the one ad group, and that one of these ads has a CTR twice as high as any of the others. This indicates that there’s something special about this one high-performing ad; it’s your job to figure out what this special something is. Maybe you discover that the hyper-effective ad used a fear tactic in its headline, whereas all your other ads went for a “we’re better value” tactic, which happened to be nowhere near as effective.

Another useful source of feedback is to measure your CTR against your competitors’. Google Display Network has a “Relative CTR” metric which tells you whether you perform better or worse than the other people competing for the same ad placements. A Relative CTR rate of x10 would mean that your CTR is ten times better than the average.2

The main commercial justification for improving CTRs is that it lowers the cost of advertising. Because of how ad auctions work, adverts with high CTRs pay less per click than those with lower rates.

i. Fine-tuning targeting

Perhaps the simplest way to improve your CTR is to tighten your targeting. If a proportion of the people you are advertising to have no interest in your product, then of course your click-through rate will suffer. The way out of this wastefulness is to study the results of your current advertising campaigns and identify any pockets of irrelevant targeting. For example, suppose you run a LASIK clinic (laser vision correction) and advertise with the broad match keyword “laser surgery”. By studying your Google AdWords report of the actual search queries which triggered your ads, you would learn that your advertisements were showing on searches like “laser tattoo removal surgery” and “laser wart removal surgery”. All the people searching for wart and tattoo removal have no chance of clicking on ads for laser eye surgery, so this would obviously drag down your advertising campaign’s overall CTR. You could repair this problem by adding negative keywords for “tattoo” and “wart”, thereby surgically fine-tuning your targeting to exclude people who’ll never click. A similar sort of optimisation is available on Google Display Network. Here, you would fine-tune your targeting by placing negative placements (i.e., telling Google not to show your adverts on websites or apps that previously gave you poor CTRs). Another option here is site/category exclusions (i.e., telling Google not to put your adverts on obituary websites or on parked domains, etc.) Yet another optimisation along these lines (and available in every platform) is geographic exclusion, whereby you stop advertising to people in regions with low CTRs.

ii. Completely different targeting

So far we’ve talked about fine-tuning your targeting with micro-adjustments. Sometimes this works, but other times it isn’t enough to move the meter. In this case, it’s possible that your current targeting mechanism is way off the mark and the only way to rescue your CTR is to try drastically different mechanisms (or to move to another advertising platform altogether). Instead of targeting by keywords, you might try targeting by workplace. Instead of targeting by workplace, you might try targeting by real-world behaviour. Instead of targeting on Twitter, you might try targeting on LinkedIn, etc.

iii. Form of advertisements

Once you believe that your targeting is in order, the next place to look for a CTR boost is within the form of your adverts. Are you using all the latest bells and whistles (e.g., AdWords Ad Extensions or Twitter Cards)? Are your ads viable to appear in the Facebook newsfeed with the maximum image size? Are your ads present on the latest platforms and devices?

Closely related to modifications of form are optimisations as to how your copy appears within adverts. For example, many advertisers Write Their Ads In Titlecase So As To Attract Attention. Or they put \^odd\^ ***symbols inside their headlines and ad bodies. Or they stamp logos on all their images, thereby establishing brand trustworthiness.

iv. Content of advert creatives

As soon as you’ve exhausted the possibilities for improving your CTR by modifying the form of your adverts, it’s time to look at their content. This involves applying the techniques from the copywriting chapter so as to enhance your creative. A better image might help your ad stand out when it appears on a page crowded with competing advertisers; a catchier headline might convince more people to read on; and an alternative angle for your ad body might lure more people into clicking and visiting your website. Experimenting with advert creatives isn’t particularly labour intensive, so there’s no harm in trying out a range of ideas.

The ads of your established competitors form a particularly good source of inspiration. To take advantage of this treasure trove of ideas, load up Google’s AdPreview tool, search for your keywords, then take notes on how competitors in other cities, countries, or niches are advertising.

v. Match between targeting and advert content

There is an interdependency between the targeting of your adverts and their content. It is only by matching the one with the other that you can achieve the highest possible click-through rates. For example, if your advert creatives emphasise the ability of your pocketknife to cut through fishing line, this might achieve an excellent CTR when targeting fishermen, but not so much when targeting wannabe tough guys.

The way to avoid these problems is to meticulously separate your adverts into small, thematic, tightly targeted ad groups. The idea behind this practice is that the advertising copy of each ad group should almost exactly match the sentiment and personality of the person likely to view the adverts contained within it.

These ad groups are smaller than you might think. Professional Google AdWords advertisers, such as Hanapin Marketing (PPC Hero), recommend having between 3–15 keywords in each group.3 Generally speaking, high value keywords (i.e., keywords that get a lot of impressions, clicks, or conversions) should appear in the smallest ad groups (i.e., of only 3–5 keywords). This helps you focus on optimising the heaviest hitters.

Overcoming Low Quality Score (Google-Specific)

Google issues your keywords a Quality Score on a 1–10 scale, with 1 being abysmal and 10 being God-like. It’s worth improving your Quality Score because a good score leads to increased reach and decreased advertising costs.

According to Google, the big three components of your Quality Score are your expected CTR, your ad relevance, and your landing page experience. However, these factors are irrelevant when you launch new keywords which have only accrued relatively low numbers of impressions (i.e., less than a few thousand). Because Google has insufficient data from your account to give you a Quality Score, it averages the performance of other advertisers targeting those same keywords. This means that until you pass what’s known as the “impression threshold”, don’t expect your Quality Score to respond to your efforts at improving it.

Assuming you’ve passed the impression threshold, what’s the fastest way to improve your Quality Score? Since your expected CTR is the largest contributor to your Quality Score, the CTR is a good place to start. Google bases your expected CTR on your previous CTR performance for a given keyword (and for keywords related to it).4 Therefore, your Quality Score will rise as a side-effect of applying the techniques we’ve already seen in the section about tackling a low CTR.

The next best way to improve your Quality Score is to work on your ad relevance. This is all about how closely your ad matches the intent behind a user’s search. You’ll get a lot of bang for your buck by studying the report about what actual search queries people make and ensuring that these same queries appear somewhere in your advert creative (be that in its title or its copy). It’s also important to ensure that you don’t have too many disparate keywords per ad group. A wide range of keywords cannot possibly address any one user’s query directly, and will be sure to drag down your ad relevance. Again, this advice is very similar to what we suggested for improving low CTRs above.

The last way to improve your Quality Score is by polishing your landing page experience. The important factors here are page load times, transparency and trustworthiness (e.g., giving your contact details on your website and explaining what you do with a user’s data), navigability/usability (e.g., not having obnoxious pop-ups, and being mobile-friendly), and the presence of relevant content (i.e., your landing page should match the content of your advert and keywords).5 An effective way to ensure relevancy is to create landing pages specifically for each ad group.

If you’re unsure which of the big three components of Quality Score is most problematic in your specific case, Google gives guidance through a feedback tool integrated into the AdWords platform.6

Overcoming Low Relevance Score (Facebook-Specific)

Facebook calculates your advert’s Relevance Score by comparing your advert’s expected positive feedback against its expected negative feedback. Positive feedback includes how much Facebook expects people to like your advert, share it, view the contained video, or click on it. Negative feedback is how often Facebook expects people to hide your advert from their timeline or report your video.

A high Relevance Score significantly decreases your cost of advertising, so it’s obviously worth improving it. One way is by increasing your CTR, a topic we’ve repeatedly covered earlier. Another way is to follow Facebook’s recommendations to 1) use laser-narrow targeting, 2) use their special call-to-action buttons, and 3) give details about your business within the advert creative.7

Yet another way to improve your Relevance Score is to refresh your advert creatives often; Facebook members will get annoyed if they see the same advert a hundred times. I know I’d consider hiding an ad at that point.

Perhaps the most interesting point of leverage you have in improving your Relevance Score is to take account of the social aspects of Facebook. Funny or quirky photos are going to get liked, while offensive or inappropriate photos are going to get hidden.

Overcoming High CPC

(I’ll use the term “CPC” here to refer to either “cost per click” or “cost per 1,000” impressions [CPM], since the ideas are the same.)

A high CPC burns through your advertising budget faster, yet gives you less for it. You reach less people and trigger less sales than would have been possible if you had a lower CPC. Indeed, the ongoing viability of your paid advertising campaigns depends on your CPC staying lower than the average profit you earn per click.

The simplest way to reduce your CPC is to bid less. As we discussed in a previous chapter, there is a progressively rising cost of advertising to each additional person. By reducing your bids, you skip the most expensive-to-reach people in any given audience, and, as such, the high cost of advertising to these few won’t spill over and inflate your overall cost of advertising.

The trade-off when bidding less is that you’ll reach fewer people. For this exact reason, reducing your bids is sometimes a dreadful idea. This is so in cases where your revenue is limited by the number of customers, and where the prices you pay per click are still less than the profit you expect to earn for it. Here, you want to do everything in your power to gain as many customers as you can—including bidding more.

The other ways to reduce CPC are nothing new; they involve techniques we’ve already seen, such as experimenting with different targeting mechanisms, improving your CTR, and increasing your Quality Score.

Overcoming High CPA

Your cost per acquisition (CPA) depends on the entirety of your conversion funnel, which includes not just your advertisements themselves, but also your website’s ability to convert this paid-for traffic into paying customers.

There’s not much new to add at this point in the book. We’ve already looked at how to optimise adverts in this chapter, and we looked at techniques for improving on-site conversion rates in a previous one. The only remaining factor to discuss is the need for a harmonious connection between your adverts and your website.

Imagine you own a site that generally has a great on-site conversion rate. You decide to send it extra traffic through paid adverts which themselves have impressive CTRs and CPCs. You expect that this combination of effective adverts and an effective on-site conversion rate should culminate in an especially effective CPA. But this effectiveness fails to materialise; for some reason you end up with a dismal CPA. What went wrong?

After some analysis, you learn that your on-site conversion rates, while normally excellent, were appalling in this one case. This discrepancy tells you something important: Your adverts somehow didn’t match your website. Perhaps your ads promised something that your website couldn’t deliver. Perhaps they sent you traffic that speaks another language or cannot view your pages from their browsing devices. Or perhaps your advert suggested that your product was free when it wasn’t. The possibilities for mismatch are endless, and the only way to improve your CPA in this case is to take out your detective hat and start searching for clues.

The End

I’m flat-out flattered that you’ve read my book all the way to the end.
I take the highest of compliments from your taking the time out of your
busy life to listen to my thoughts for so long. I sincerely hope that
you took something valuable away from this book that will help you
thrive in your own projects, whatever they might be.

In a weird way, I feel that we know each other, asymmetric as that may

Review Request

Reviews and word of mouth are the lifeblood of any book’s commercial
success, so you’d be working a small miracle for me by rating this book
on Amazon or GoodReads (click here),
tweeting about it to your followers, or mentioning it whenever you see
someone asking questions about online marketing.

  1. Some advertising platforms have special features whereby they only charge you for an impression when an advert was actually viewed. For example, Google’s Active View technology, released around 2014, only registers an impression when the advert was at least 50% visible on a user’s screen for at least one second. Google also has a Display Network-only option (under “exclusions”) which can exclude your adverts from displaying “below the fold”. There’s a theoretical argument that you should always use this exclusion, even when you are bidding on a CPC (cost per click) basis: Although there’s no charge for unclicked CPC advertisements, having impressions appear below the fold decreases your advert’s click-through rate, which in turn causes your Quality Score to suffer and your cost of advertising to rise. 

  2. To view relative CTR in your AdWords account, go to the “Ad groups” tab, select “Customize columns” from the “Columns” drop-down menu, then select “Relative CTR.” 

  3. http://www.ppchero.com/16-commonly-asked-ppc-questions-and-answers/ 

  4. http://services.google.com/fh/files/misc/settling-the-quality-score-whitepaper-final.pdf 

  5. https://support.google.com/adwords/answer/2404197?hl=en 

  6. Log in to AdWords. Go to the Keywords tab of your campaign, then hover your mouse pointer over the “status” speech bubble for any keyword. A pop-up will appear detailing how you perform in each Quality Score dimension. 

  7. https://www.facebook.com/business/a/ads-relevance-score 

More Articles: Click here for full archive

SEO and Internationalisation

Separate language from territory; clarify your international site structure to Google

SEO with Social Networks

Why Twitter Cards and Facebook Open Graph aren't total wastes of time

Ad Creative Writing Basics

Tips for image choice, headline, and the body