Take Your Ads From Failing To Scaling

We’ve all been there. Your ads are running just fine and then seemingly out of nowhere they stop performing.

You’re surprised and disappointed to learn that your ads have stopped producing a profitable return-on-ad-spend (ROAS), they don’t have a steady click-through rate, and/or you don’t have any new quality leads.

When the initial shock begins to wear off, you’re stuck wondering where to go next with your ad performance.

Your brain is buzzing with questions: Perhaps you could have monitored your ads more often? Maybe you could’ve tested more creative? Did you make an unintentional mistake on budgeting or ad placements?

Regardless of how you got there, it’s important to know that it will be okay and there is a path forward. Together, we can take your ads from failing to scaling. 

Ad Performance “Stages of Grief”

When your Facebook ads stop performing, your first instinct is probably to panic. Then you immediately wonder why (and why now) this is happening.

Once you go through those initial stages of grief, you can start looking ahead to rebuild your ads for longer-term stability and scaling. This blog post covers all of these stages and gives you some quick-start ideas for account recovery. 

Don’t Panic

When the initial “my ads are totally failing!” feeling hits you, I realize it’s difficult to stay calm and not panic. Trust me, I’ve been there. Just take a deep breath and ask yourself this question:

Is everything truly failing or is it just a few ad sets and/or audiences?

This is an important question to dive into because your findings will usually show that some audiences or ad sets are actually performing the way they should.

I measure ad performance for most clients in cost-per-acquisition, click-through-rate, relevance score, return-on-ad-spend, cost-per-landing page view and other metrics. After I perform this assessment, I build custom dashboards using the Custom Reporting tool within the Ads Manager.

Then I select “Customize Columns” to pinpoint the data I want to report on.

After I further assess this data, it gives me a much better picture of what is working and what’s not. I then take detailed notes on what is still performing well so I can include it in my rebuild.

Take a Step Back and Think About Ad Competition

Let’s take a quick step back and remember that over the past year, specifically in the first half of 2018, ad competition has continued to significantly increase on Facebook and Instagram, especially in the Facebook News Feed.

Facebook warned advertisers about this shift beginning in 2016 and they kept talking about it in 2017.

In short, there are more people advertising than ever before and the “ad load” (AKA the number of ads being shown in the News Feed) has increased while the overall number of users (specifically, users in North America) hasn’t increased at the same steady rate.

To counter the ever growing ad competition issue, Facebook has offered new ad placements in Messenger, Marketplace and Instagram Stories, so results can definitely be stunted if advertisers don’t experiment into these new placements and just stick to their old, previously successful playbook.

Ask yourself: are you still only targeting the Facebook News Feed? If so, keep reading…

In most cases, ad performance has declined due to one or more of the following reasons:

  1. Audiences are “worn out” and have been overused
  2. Creative hasn’t been updated recently
  3. Placements are too limited 
  4. The bid is too narrow

So let’s unpack each of these issues.

1. Audiences are “worn out”

Say this phrase with me: first time impression ratio. It’s a huge deal, folks. This Facebook statistic tells you, by day and by ad set, how many of the impressions you’re showing are for the first time.

When you start a new ad for a new audience, that number is 100%. As you advertise longer and spend more money, that number declines. It’s important to understand this metric in relation to ad performance because if that number dips below 50%, you probably need to refresh creative or rebuild that audience.

You can begin finding “delivery insights” at the ad set level.

Once you click on that “See Delivery Insights” it takes you into this screen.

From here, you are able to see many data points, but there’s one that’s especially important: First Time Impression Ratio. This number shows you how many of those people are seeing the ad for the first time.

In this case, it’s very low. Normally, we want that to be above 50% for prospecting or brand new potential customers. If you’re updating creative and your audience sizes are right, this number will stay above 50%.

Here, the number is less than 10%, meaning 90% of that audience has seen the ad before. If I looked at the frequency metric, I’d likely see that number be above a 3 or 4 within a 7 day period–which is a major problem.

***Important note: is this post interesting to you? Take our upcoming course on this subject! ***

2. Creative hasn’t been updated lately

Be honest: When was the last time you updated your ad creative? When was the last time you tested new creative ideas in all parts of the funnel? How about testing stand-alone photo posts, or Instagram-specific creative?

One of the most common issues I see when auditing Facebook accounts is a singular focus on one ad type.

For example, somewhere along the way a lot of advertisers heard carousel ads “always work really well.” They can perform well, yes, but for prospecting, they actually don’t deliver sustainable results for most companies. Of course, there are many exceptions — but they tend to drive clicks but not as many purchases.

So instead of carousels, why not try focusing on content that looks or feels user-generated, or in fact is user-generated? In today’s competitive environment, it helps to mix in some creative that looks like something a friend took and also explains what the product does, and why it might benefit that person’s life. 

If you can successfully update ad creative and test new ideas every two weeks, you would be in much better shape. 

3. Placements are too limited

I can’t tell you the number of times I’ve seen accounts that only target the Facebook News Feed for placement. Don’t get me wrong, I love the good ol’ News Feed, but at this moment in time, it’s just too competitive to have that be the only place you’re targeting.

Here are some other placements where I’ve recently seen stable performance:

  • Facebook Marketplace
  • Facebook Messenger
  • Instagram News Feed
  • Instagram Stories

Perhaps most noteworthy, I’ve personally seen Instagram stories delivering positive results on prospecting audiences similar to that of the Facebook News Feed.

Have you experimented with new placements?

By simply expanding your options, you can help lower prices and reach more customers in a less competitive environment.

4. Your bid is too narrow

Last week on a coaching call, a client asked me to explain the pros and cons of bidding certain ways under a conversion objective. I started talking and didn’t stop for over an hour. Clearly, there’s a lot to talk about in reference to this subject!

One of the most common scenarios I see is when advertisers bid in a 1-day click window.

I myself have used 1-day click bidding and in many instances, it can work, but let’s think about the signal you’re giving Facebook. You’re saying to Facebook, “Find me people who are likely to buy something within one day of clicking on my ad.” 

This tactic can be incredibly limiting. 

Think about how many times you click on an ad and then buy right away. Most users don’t! We click, look around, think about it and consider that purchase. Then maybe I’ll go back on the weekend and buy the product.

This is the core of the lesson: bid within the window you think your users will actually convert. In many cases, starting with a 7-day click or 1-day view bid can be more helpful to widen that target audience a bit.

There are countless other discussions we could have about bidding, but a little experimentation on conversion windows can go a long way.

Rebuilding for Stability

By now you’ve done the ad autopsy and you’re ready to begin rebuilding for more stability. Rebuilding is a data-rich process based on utilizing what’s worked before, testing out new strategies, and trying to reach people in different parts of their buying journey. When I rebuild, I use the guide below.

Notice that each audience segment has a different offer and different ad copy. Fans are different than previous site visitors and previous customers differ from potential customers that have engaged with you.

Every grouping needs its own tailored creative and message. These offers aren’t specific to what I do every time but it gives you a beginning idea of how I think about it.

Take the Wheel

This post is just the tip of the iceberg of how to turn ads around from failing to scaling — there are plenty of other options and solutions. What has worked for you when things have started to go south? Let me know below in the comments!

Training Course

If this topic is interesting to you, I’m teaching a course called “From Failing to Scaling: Trusted Strategies for Account Recovery and Growth” with Jon. We’d love to have you join us.

You’re not alone if you are experiencing issues with ad performance right now. You want to scale, but you might not know how exactly to turn things around. This course helps you with the following:

  • Steps to take right away to stabilize an underperforming account
  • Understanding how to properly and sustainably scale
  • Bringing new ideas into the mix to save performance
  • Notice the warning signs moving forward
  • How to save money and take control

Join us!

The post Take Your Ads From Failing To Scaling appeared first on Jon Loomer Digital.

Advertisements

Facebook Page Reach: Page-Level Reporting

Facebook Reach is one of the metrics you hear about a lot when it comes to Facebook reporting. When we report on Facebook Reach, there are two main classifications to remember – Facebook Page Reach and Facebook Post Reach.

Facebook Page Reach tends to be the least troublesome of the two, so let’s start here first.

Properly understanding reach tends to be a source of confusion for many marketers, particularly when it comes to reporting on Organic Facebook Reach. I often see it arise in conversations with other members of the Power Hitters Club who are working on performance reports.

Jon has written about this metric in the past, particularly after Facebook made changes to the way reach is defined.

A general disclaimer: For organic Facebook Page Reach, I do not recommend over-emphasizing this metric. It can be a vanity metric that is not connected to actual results.

However, you may have a reason for more holistic performance reporting, such as when you have a particularly high performing advertised post or campaign, and you wish to include organic results in broader performance analysis.

How to Get Facebook Page Reach Data

The simplest way to access page-level Facebook metrics is to click the Insights tab at the top of your page. Note: if your page is in Business Manager, you’ll need to log into Business Manager first to view this option.

After you click the Insights tab, you can click Reach on the far left. At the top right, you can adjust the date range. As you’ll no doubt notice, this is very high-level information. The options on the top right of the chart can split the data into Organic or Paid.

Facebook Page Reach Insights Panel

If you are a bit more daring with your data and want freedom to manipulate it more (don’t be afraid to jump in!), you can export Page Level insights info directly from the page.

You can export by clicking the Insights tab, then clicking “Export Data” from the top right.

Facebook Page Reach - Page-Level Data Export

You’ll see options to export Page Level data from this dashboard by clicking the Page data option. You can adjust the time range, and specify the data to be exported.

Facebook offers several options here to change the layout of the data you export. I normally like to download all the data in bulk, instead of running into a situation where I wished I had captured something and have to re-export.

If you want to make a change to your selection, you can do so from here. You can also save a favorite layout so that you don’t have to rebuild it in the future.

Facebook Page-Level Export Data Selection

Once you have selected the data points you need, click Export, and it should generate a .xls file for downloading.

If you’ve left All Page Data in the export, the first thing you’ll notice is just how much information is captured. There are many tabs offering deeper, fragmented detail. For now, we’ll just focus on the Reach Metrics.

Getting Facebook Page Reach Information out of Page-Level Data

At the time of this writing, on the main tab, you should see the following columns related to Reach: Daily Total Reach, Weekly Total Reach, 28 Days Total Reach.

These are also broken out into Organic and Paid for each time range, as well as something Facebook refers to as “Viral Reach” – which essentially means that a Facebook user saw the post along with some form of social context. That is, they saw the post along with a message saying their friend has interacted with it.

A helpful tip in case you get confused: There is a definition under the header of every column of the export to tell you what it means.

Now comes the fun part!!! (Also where most people tend to make a few mistakes.)

The golden rule on this: you cannot add together lines to get a total number of reach. Said another way – you cannot simply add together reach numbers, line by line, for a total.

This is because each data row is a measure of unique people for that date, and you may have people reached on multiple days who could be counted twice if you simply sum the rows.

For example – let’s say you wanted to know the total number of people who were reached by your page from July 1 – July 5. It seems logical that you could simply add together the numbers from the rows associated with those days. However, Reach is a measure of Unique Users, which is a very important distinction. When you add these numbers together, they are not de-duplicated… so your sum would be incorrect.

We can illustrate this with a hypothetical example…

Let’s say we had the following results, for three different days in July:

  • 100 People on July 1
  • 200 People on July 2
  • 100 People on July 3

If you simply added these together, you would assume you have reached a total of 400 FB users over these three days. However, you (almost) always will have some level of user overlap of reach from day to day. Therefore simply adding these numbers together will give you an inaccurate count.

It’s important to note that de-duplicating is primarily an issue with metrics associated with unique people (such as reach, Daily Page Engaged Users, or any “user” metrics). If you are reporting on a metric that is not unique per user – such as Impressions – you can add across rows without any of these issues associated with duplications in your data.

De-Duplicating Facebook Page Reach Results

While page-level reach reporting is somewhat limited based on how we can de-duplicate user-specific data, there are some ways that Facebook’s system does this for you automatically. An example is the inclusion of the Weekly and 28 Day numbers.

Based on the way these numbers are presented, they should be providing rolling counts of select metrics against the time frame indicated. What this means: for a Weekly Total Reach count, the number indicates the user reach Facebook estimates for your page on a weekly basis, for the dates associated with that specific row.

If you’re more on the nerdy side (or just curious!), you can do your own de-duplication exercise by adding the daily data for a 7-day period and comparing that with the Weekly data reported for the same date.

This will give you an idea of just how badly you can overestimate results if you were to simply add the raw numbers together. This also provides a better understanding of the number of people who heard from your page more than once in a specific period.

Here’s an example with some sample data from a page to illustrate. We’ll walk through the numbers to make sure this is clear.

Facebook Page Reach 7 Day Calculation Example
*In this example, the Page-Level export includes columns H and J. I’ve added I and K as calculated columns.

If we use the weekly reported numbers from June 30th as our specific example, we would have simply added the rolling day count from the 7-days prior period had we been doing this manually. This would give us 4,274 total people reached (15 people on June 24th + 24 people on June 25th… etc). However, Facebook reports that the total Weekly Unique Reach was 3660 people. That’s a difference of 614 people.

What does this tell us? For the 7 day period, we had 614 people who were reached at least two times over the period.

While some bit of caution is advised for taking these numbers too deeply to heart (due to the fact that the reach metric is estimated and therefore sampled), this is one method that allows you to get a general idea of your overlap.

Reporting on Impressions

Given the challenges of correctly assessing and interpreting sampled data and manual de-deduplication, it can be easier to stick with the simpler, non-unique metrics such as Impressions.

We can validate that Impressions can be added together with no issues. We can do this by using a similar approach as the above Facebook Reach de-dupe method. This can be seen using this sample page data below:

Facebook Impressions Manual Calculation - 7 Days
*In this example, the Page-Level export includes columns W and AA. I’ve added Z and AB as calculated columns.

The 28-day metric can use a similar approach.

Reporting on Frequency

You can also combine these data points of Reach and Impressions to get a general idea of Daily, Weekly, and Monthly frequency levels. However, remember that frequency is an average of all impressions. Some people might be reached many times per person, and some are only reached once. You can use this method to report the overall average.

Here’s an example of the method in practice. Remember: Impressions divided by Reach equals Frequency.

Facebook Frequency Calculation Example
*Estimated Daily Frequency and Estimated Weekly Frequency are columns that I’ve added to the spreadsheet. The Page-Level export included Daily and Weekly Reach and Impressions numbers.

Bonus (nerdy) tip: If you want to get an idea of the frequency distribution (how many people were reached one time, vs. two times, vs. three times, etc) for the Daily, Weekly, or 28 day metrics, you can find those in the additional tabs along the bottom of the page level export. You should find that the cumulative averages of these distributions are equal to the estimated method we used above.

It will look something like this:

Facebook Reach Frequency Distribution Example

These numbers are telling you the overall frequency distribution that your page delivered on a specific date. On June 3, this page reached 109 people 1 time, 13 people 2 times, 2 people 3 times, etc.

Monitoring Frequency

How many times your page content reaches a similar audience can be a good thing to monitor. I normally recommend paying closer attention to frequency at the ad-level (or post-level), instead of page-level.

Bigger problems can arise here when high frequency occurs and users tire of a single ad. By running many different ads (or posts) for a single page, higher levels of frequency may not be as problematic.

If it seems these frequency numbers are higher than you’d like, you could consider doing different targeting (primarily using your paid activity) to evaluate different target audiences for your content. Another option is lowering your budgets for particular ads that have higher frequency.

People often ask about the ideal frequency to mitigate these issues, and there is no simple answer to this. Facebook has published an interesting framework on thinking about frequency, which may be useful for considering your own results.

What Does this All Mean?

  1. Reporting on your Facebook Page Reach has some challenges. We should fully understand what’s behind the numbers before simply adding a bunch of rows together in a spreadsheet.
  2. Impressions are a safe metric you can add together with reckless abandon (mostly).
  3. You can estimate Average Page Frequency. You can also investigate details on frequency distribution in Page-Level exported data.

I’ll write more in the future on Facebook Reach Post-Level reporting, which carries its own strengths and weaknesses.

Your Turn

Do you use Page Reporting? Do you have any particular challenges with comparing performance data from Paid Ads vs your Organic results?

Let me know in the comments below!

The post Facebook Page Reach: Page-Level Reporting appeared first on Jon Loomer Digital.

Highlights from Google Marketing Innovations keynote

Tuesday saw Google’s Senior Vice President of Ads, Sridhar Ramaswamy, and his team, take to the stage to talk us through the latest innovations coming to Googles Marketing Platform in 2018. The Keynote covered everything from the recently announced rebrand of Google AdWords, to the highly anticipated Cross Device reporting on Google Analytics.

In case you missed it, here are our key takeaways from the keynote:

Goodbye Google AdWords…hello, Google Ads

Initially announced at the end of June, the rebrand of Google AdWords came as no surprise during the keynote. Effective from the 24th July, Google AdWords is becoming Google Ads – which will encompass all paid search, display and video products.

There’ll be no immediate impact for anyone who’s adopted the new AdWords (or should I say Ads?) interface. Although, if you’ve been putting it off, I’d suggest making the leap now as, come the 24th July, it’s likely the old interface will be inaccessible – especially as Google have warned us the switch date is coming this month.

DoubleClick is also seeing a rebrand to be part of the new Google Marketing Platform, alongside the Analytics 360 Suite. This becomes a single port of call to plan, buy, measure, and optimise your digital marketing activities. DoubleClick Search is becoming Search Ads 360, while Display & Video 360 will bring together the features from DoubleClick Bid Manager, Campaign Manager, Studio, and Audience Centre.

Google Marketing Suite will see a new integration centre to better connect the range of products within the suite.

New targeting comes to YouTube ads

Nicky Rettke walked us through the latest products coming to YouTubes advertising range – focusing on three new targeting methods:

  • TrueView for Reach: bringing the simplicity of impression-based buying to YouTube – this is ideal for driving awareness to a broad customer set.
  • TrueView for Action: optimised for driving website conversions, TrueView for Action will see video advertising paired with a prominent call to action linking direct to your website. We will also see Form Ads coming later in the year for lead generation goals.
  • Maximise Lift Bidding: will leverage machine learning to help reach people who are more likely to consider your brand after seeing an ad.

Supercharge your ad copy

Machine learning was the main talking point of the search ads section. Google’s push towards machine learning has been apparent over the last year, with increased smart bidding and optimisation already available. Google are now extending this to ad copy, with the help of responsive search ads.

These have been rolling out since beta testing ended in June and are already available to many advertisers. Responsive search ads bringing a level of multivariant testing to your ad copy optimisation.

Not only are Google giving us a helping hand with machine learning, they’re also rewarding our adoption with prime real estate – responsive search ads display up to 33-character headlines and two 90-character description lines per ad.

By inputting up to 15 headlines and four description lines in to the ads, the machine learning algorithm uses multiple variations to find the optimal configuration based on the user’s search.

Cross Device reporting and remarketing

Clearly a product many have been waiting for, Anthony Chavez’s introduction of the new cross device reporting saw a cheer ripple through the audience.

The growth in mobile device usage has been rapid in recent years, making it harder to decide where to focus your marketing efforts. New Cross Device reports available within Google Analytics seamlessly combines data from people who visit your site across multiple devices, giving a concise view of how users are interacting with your site and brand.

This new way of reporting will allow for cross device remarketing audiences to be built and used across Google Ads. With a brief touch on privacy, only users opted in will be shown within the report and no first party data will be passed over.

The new Cross Device reporting will give more insights into users’ habits, enabling us to link moments together to gain a better understanding of our customers and where to focus our marketing efforts. For me, this was the highlight of the keynote and I look forward to unlocking its potential in the future.

Automated product feeds & smart shopping

Maintaining a product feed for shopping ads can be a manual and time-consuming process, so Google are bringing their automation to shopping campaigns. Automated feeds will launch later this year, crawling your website to create its own feed. This will open the door to advertisers who have been put off previously by the daunting prospect of setting up and maintaining their first shopping feed.

We saw smart shopping campaigns launch in May, these will take automated feeds one step further, by optimising your shopping campaigns to the goal you select, taking the manual optimisation out of shopping.

Third party integration with eCommerce platforms will also be coming to shopping campaigns later in the year, alongside new business goals to drive local store conversions and new customer acquisition.

New local campaigns coming to Google Ads

Aimed to help offline performance, the new local campaigns help drive in-store visits by linking directly to your Google My Business account to help build effortless campaigns. With minimal input, machine learning will do the rest of the work and optimise your campaigns to drive footfall and offline conversions.

An interesting keynote was delivered by Google this year, machine learning was again a common theme running through all of the new products announced, making adoption of machine learning and automation almost inevitable. If you haven’t already, its time to start taking advantage of machine learning and watch your campaigns grow!

 

7 pitfalls when using Google Data Studio

Ok the title is click baity – Google Data Studio is an awesome tool and many of you are probably already using it on some level. For those of you that don’t know Google Data Studio, it’s time to get on board. After all, it’s free and allows you to simplify the visualisation and accessibility of your data.

There are a multitude of systems it can connect to by default – don’t forget it is a Google product, so it’s connections are biased towards the Google ecosystem.

Being able to visualise the data from multiple systems in one place simplifies and streamlines reporting. For example, surfacing organic Analytics data alongside Google Search console clicks and impressions or connecting it with Google sheets to help pull in aggregated search data that’s updated live.

Google Data Studio

There are some gripes I have it with though, and it’s a good idea to be mindful of the shortcomings with the platform before you dive straight in.

So let’s start with seven pitfalls you need to be mindful of when using Google Data Studio.

  1. Aggregating metrics from different sources

This is probably my number one gripe. You want to add metrics for AdWords, Bing, Facebook etc. to create a total aggregated metrics to calculate ROAS, but there is currently no direct way to aggregate metrics from multiple data sources in Google Data Studio. The messy solution to aggregate the metrics, as in the example below is to use Google Sheets to aggregate the sources, and then connect it back up to Google Data Studio to display the results. This isn’t ideal (in fact, it’s awful). For agencies and in-house analysts who need aggregated metrics, this is probably the number one reason you’ll be looking at other providers.Google Data Studio display results

  1. Limited data connectors

It’s a Google product so connections to systems are limited. There are third party connectors available to download and use but beware not all are maintained. We use the Supermetrics suite of connectors extensively and the support that these guys offer deserves a special mention.

Limited data connectors in Google Data Studio

  1. Metrics labelling

Let’s say for example you display the Site CTR metric from Google Search Console. If you were to relabel the heading to CTR, for example, it also relabels the metric – this is not cool. Debugging this involves going into the metric and removing the name to see what actual metric is being used to pull in the data. They really should include a display name field.

Limited metrics labelling in Google Data Studio

  1. Calculated metrics

One of the cool features of Data Studio is the ability to use calculated metrics.

For example, you have 3 goals that you need to sum up in analytics to give you an aggregated view of engagement, you can do something like this in Data Studio. Don’t forget though, you can’t aggregate this over different sources.

Calculated metrics in Google Data Studio

You create a calculated metric by hitting the ADD A FIELD when editing the source connection.

How to create a calculated metric in Google Data Studio

Calculated metrics appear in the source mapping appended with ‘fx’. Below is an example of an ROI calculated metric.

ROI calculated metric in Google Data Studio

ROI calculated metric in Google Data Studio

The problem is, custom metrics definitions are not saved when you change to a different source e.g. if you change the source from Joe Blogs AdWords to Mr Smiths Adwords, you have to create every single custom metric again for each source – there is no way to save this into a template.

  1. Exports

The only way to export a Data Studio dashboard in the interface is to export to PDF.

How to export a Data Studio dashboard

This can mean trying to work it into another deck is a bit problematic – I’d really like to see some form of integration with Google Slides in future.

  1. Comparison arrows in tables

For actionable data, you’ll need comparison metrics. The problem is, you can’t select how you want to display the up and down arrows per table column in Data Studio – changing down to be green would change the whole table. In the below example, an increasing Avg. CPC is probably not a reason to celebrate.

Comparison metrics in Google Data Studio

  1. Comparison metrics precision

Similiar to the above, you can’t change the precision of the comparison metrics in tables. They are always to a single decimal point.

I’m sure there are other gripes you have with Data Studio – feel free to tweet me @_AlanNg and share your frustrations!

On the flip side, it’s not all doom and gloom. Keep your eyes out for a new post on all the reasons we do like Data Studio.

Amazon Search Optimisation: The Usurper of Search

Day one of SEO training and you come away learning one thing, Google own search… But what if it doesn’t? Well at least when it comes to eCommerce. I’m here to show you how Amazon rose to overtake Google and share with you how to use Amazon search optimisation to rank higher on Amazon.

Who the hell is this guy trying to challenge the all-powerful Google?

Hey! I was just as surprised as you, but when you see figure 1 and figure 2, this brings us into a strange new world, where Google is second best.Graph showing where shoppers start a search 2015

Figure 1 where shoppers start their search 2015

Graph showing where shoppers start a search 2016

Figure 2 where shoppers start their search 2016

 

9/10 retail customers go to Amazon after they’ve found the product on your site. 45% of the UK search on Amazon before they use Google and in the US it’s 52%. This coupled with the rise of voice search, where Amazon are again killing it, means we as SEOs need to be prepared.

How is Amazon killing it in voice?

  • They have sold more than 5.1 million smart speakers in the US since launch in 2014.
  • Amazon account for circa 90% of all voice shopping spend.
  • Over 70% of people who did a voice search last year used Alexa.

To give you some perspective, Will Reynolds recently announced on LinkedIn that less than 1% of his client’s enquiries starts with “Ok Google”.

We need to get on board with Amazon search optimisation now, because by 2020 there will be 24.1 million smart speakers in the US alone and by 2019 the voice recognition market will be worth $601 million.

Here’s what you need to know about Amazon SEO.

The top 20% of retailers make 80% of the revenue and Amazon’s A9 and A10’s algorithm uses conversion rate, relevance and customer satisfaction to rank products.…Sound familiar?

Do Amazon care that their algorithms are massively behind Google’s? Not in the slightest. Why? Because Amazon’s main priority is targeting visitors with products that will most likely result in a purchase. They push the best performing products to the top to increase their exposure, enabling more sales. This leads me on to a little known fact about Amazon…. Amazon holds no loyalty to its sellers. How do I know? Try putting Asics into Amazon. They’ve been a long-term customer, they supply directly to Amazon but what’s the first thing that appears at the top of the search? A New Balance running shoe. Looking for batteries? You’d expect to see Duracell or Energiser up there, nope, just Amazon’s own brand, Amazon basics. I imagine if you’re in retail and not on Amazon, you maybe sweating just a little bit.

This is where I come in to help.

How to rank higher on Amazon

Now if you came late to SEO, you’ll have no idea what Google was like before Penguin or Panda. I’m here to show you that Amazon optimisation breaks down into four key factors and those factors will help you rank higher on Amazon:

  • Organic
  • Paid
  • CRO
  • …but what about links?

In terms of organic, content is king again and what you really need to focus on is your product – the devil is in the detail when it comes to your product title and your products description.

How to structure your title

  • Don’t stack keywords, use 2-3 and place them early in the title.
  • Use characters to turn your title into phrases.
  • Try to optimise your title readability by varying your title lengths for mobile and desktop (See below).
  • See Figure 3.

Amazon search result

Figure 3 The Organic top results for iPhone chargers 23/05/18

Title Text Lengths

  • For desktop organic results use around 115-144 characters.
  • For paid ads stick to around 30-33 characters.
  • For mobile titles keep them between 55-63 characters.

Mobile, like everything else in SEO is very important – Amazon users switch between searching on mobile and desktop all the time. Earlier this year, Amazon shoppers used mostly desktop, but now mobile searches are about 100,000 in front.

Back end search terms are by far the most important tool for organic ranking on Amazon. This is a great place to enter all the buyer keywords you may have not been able to fit in your product listing. It’s not visible to customers but is indexed by Amazon, so now is the time to stack keywords!

  • There are five fields to fill in – remember only 250 characters are indexed – try not to exceed this. If you do, make sure you start each field with the best keywords to ensure they are counted.
  • Make sure you include common misspellings but there’s no need to duplicate.
  • You don’t need to worry about punctuation, repetition, singular or plural words as Amazon already has got your back on this.

Amazon Keyword Research

For keyword research use the Amazon Keyword Tool. It’s free to use and gives you long tail keywords based on Amazon suggestions and can be used for organic and paid keywords as well.

Amazon PPC

This is the easiest way to drive your sales up on Amazon (Surprise, surprise Amazon is just like Google, you give it money and it helps you out). Not sure where to start? Start by creating manual PPC campaigns to target your keywords, then pump up the bids to drive exposure. Again, just like Google, running your paid activity for a short period at cost or even at a loss will be worth while to increase your organic ranking.The paid top results for iPhone chargers on Amazon

Figure 4 The Paid top results for iPhone chargers 23/05/18

See how much clearer the ads look on paid than they do in the organic top results? Nice short to the point ads. Creating a campaign is really simple as well.

  • You type in the campaign name, so you can monitor the results.
  • Add in how much your daily budget is (Min is £1).
  • Pick the dates you want your campaign to run.
  • Finally, you choose whether you want Amazon to target your ads on their data or you can configure this manually. I’d recommend the latter because nobody knows your customer as well as you do.

Now it’s time to set the key words you want to target – again, you have the choice to do this manually or automatically through Amazon. If you’ve used the keyword research tool from earlier, it’s time to use it again but now, like on AdWords, you can decide if you want the keywords to match exactly, broadly or as a phrase. A mixture of all three is the best way forward.

Now it’s time to see what the CPC for your keywords are and decide how much you’re willing to spend. What’s the best way to get conversion on your ads? Offer a deal.

You should always let your campaigns run for at least four weeks to get the best data you can from Amazon. After that four weeks, review what’s working and what’s not and amend your campaigns accordingly.

Now for CRO

  • Customer reviews: The higher your reviews, the higher Amazon is going to rank you. Like Google, you need to get a few reviews before Amazon starts to rank you against the bigger competitors. No one really can agree on the magic number but from what I’ve seen you need a minimum of 50.
  • Answered questions: These are essentially what blog posts are to your standard SEO content strategy. The best way to get these are to ask your customers what questions they have after buying your product (Also it’s a good way to try and get a review).
  • Image size and quality: I can’t stress this enough, these must be 1000 x1000 pixels so your buyer can zoom in and you need photos from all the angles of your product.
  • Price: Because Amazon values buyers over its sellers, it will always push the lowest price to the top, so you need to be auditing your competitors regularly to see what they are charging.
  • Exit rates: If someone looks at your product then immediately exits the Amazon site, Amazon is not going to be happy and will knock you down. Make sure you are in no way misleading buyers to avoid this.
  • Bounce rate: Same as the above, if someone lands on your product then leaves quickly to look at someone else’s, Amazon is going to see you as a hindrance to its buyers and penalise you.

But if Amazon is so much like Google then where do links play a role? In short, they don’t, they’ve been replaced with something that Amazon hold more dearly than any other factor. The quickest and easiest way to win at Amazon SEO is a sale.

Sales are Amazon’s equivalent to links. The more sales you have, the higher you’ll rank, and higher rankings lead to more sales – just like downloads for App store optimisaton. Outsell the competitors that outrank you for your keywords and you will shoot to the top. DO NOT be tempted to put through fake sales and fake reviews. Just like Google, Amazon hits you with penalties and these are a lot harder to get passed. You can’t even contact your buyers to offer them discounts directly without Amazon picking this up.

The scope of search in 2018

Figure 5 The scope of search in 2018

Now some of you may have seen the above chart and think this undercuts everything I’ve just said, but this chart only looks at search as a whole and doesn’t just focus on shoppers looking to buy products. When you take this into consideration, you get charts like figure 1 and 2.

Simpsons meme

In all honesty when it comes to eCommerce, Google is a lot like Abe Simpson here; a ton of knowledge and lots to shout about but no teeth, because what’s the point of Google if it’s not to make money?

Final thoughts

What to expect from Amazon’s A11:

  • Amazon will probably link with PayPal to cut down on black hat sales and reviews.
  • Content will probably be more precise, so stick to shorter and clearer titles.
  • We’ll probably see improved ranking for stores that have their own eCommerce sites but still use Amazon Pay.
  • CPC is expected to be controlled by relevance score and much more.

Keep a weather eye on the horizon because Amazon is making big moves and I guarantee it won’t be long before you can do your monthly grocery shop, your banking, buy a car and book your holidays on Amazon.

Keyword Rankings and Forecasting in SEO using Shiny

Previously, we’ve seen how R can be used to retrieve data from APIs such as Google Analytics. Often with data, you’ll conduct the same type of analysis repeatedly, using the same kind of code, for various projects. To make life easier, wouldn’t you want to build a front-end, so you can just plug in your numbers and get your analysis out? With something called Shiny, you can!

What is Shiny?

Shiny is an R package built by Rstudio. With a Shiny application there are two parts, the backend and the front-end. The backend is the Server part of the code. This is the backbone of your work and where your normal repetitive analysis would take place. The front-end section of the code is the UI file of your code. This is where you can build the visuals to be able to run those repetitive analyses. However, it is possible to combine these parts and build an application in one single file, or just build the UI into your backend.

Why should you use Shiny?

Consider this scenario, you’ve built some code that lets you regularly interact with APIs. But you always change one aspect of the code so it’s specific to the project you are working on, such as changing the Google Analytics View ID to get the correct view. With Shiny, you could build an application that calls the API automatically and get the View ID that you need, and then retrieve the data required with the press of a button. A more explicit example is using AWR. You can access their API to get your list of projects and ranking dates, then display them in your UI. This way, you can load up your application, choose the project and date that you want, and get the keyword rankings in the format you need.AWR rankings viewerIf that doesn’t convince you, how about SEO forecasting? You could build a front-end onto your forecasting code so that when it is required, you can just plug in your data, press Go, and speed up your analysis. Or even better, building the application so one of your colleagues can do the forecasting for themselves using the same methodology that you use. To go a step further, put your application on a server so it can easily be accessed without them having to use R.IRMA forecasting toolThe first advantage of this is that it makes the whole process more efficient. You don’t have to open your code up and edit it. You’ve done it before, why do it again? A second advantage is that it brings reproducible and robust analyses to people who don’t need to build or even understand R code to be able to conduct an analysis in R. Third, you now have a process and methodology in place that can easily be referred to.

There’s so many possibilities with R that make it so powerful. We’ve previously seen that we can integrate R into our data extraction process using APIs, so that we can combine data extraction with data analysis. What Shiny brings to the table is that we can bring the data extraction and analysis process to people who don’t know statistics and data. This also allows the process of making an actionable decision, whether it is by someone who knows R or not, much more efficient.

Summary:

  • Use Shiny for extracting keyword rankings
  • Use Shiny for SEO forecasting
  • Use Shiny to performing data analysis
  • Use Shiny to do anything!