Driving external traffic into Amazon can be a great paid search strategy in 2021 because the conversion rate is so high, but getting there can feel a bit like sailing through the Bermuda triangle. The compass goes berserk.

Apple’s is also messing with our digital compasses with the IOS 14.6 update, killing off cookies. Have you seen the latest Apple advertising campaign about privacy? It is a fun illustration of the privacy challenge. It is clear to everyone that Apple rids us of all that annoying tracking. “Why don’t you mind your own business?”. The AppleID used everywhere to personalize services in the Apple cloud, clearly isn’t tracking anything, is it?

In reality, the battle over cookies is just another battle in the “data wars” between the large American data-driven platforms who represent such a large part of our connected lives. I always have a laugh when I hear conspiracy theories like “they know everything about us“. Sure they do, Apple knows where you are, Facebook knows what you like, Google knows what you are looking for and Amazon knows what you buy. But they just don’t sit and have tea together and discuss how they can control you. They don’t communicate or share any data. At all. That’s why the compass is off.

Walled Gardens

One of the best metaphors for describing this data impermeability is the “Walled Gardens” metaphor. Each of the platforms is raising walls around its users to prevent them from leaving and to avoid any of their data to escape. Apple is just raising a cookie wall in the current battle.

When you have a website, you can install Google Analytics for free. This gives Google all the data from your website. When you set up a Facebook page, it belongs to Facebook who will give you some nice inroads into the data. It is pretty hard though to actually figure out who is in your Facebook audience. That is not where any privacy issue lies. If you want to actually communicate with your community, you pay Facebook through advertising. And you certainly can’t track any user behaviour when someone came from outside of Facebook.

“People! Can you just stay in one garden at a time, so we can count, please!?”

Why drive to Amazon?

Unfortunately, these end-users that marketers are trying to understand are unruly. They like to surf, to browse and to zap, much to the dismay of the platforms they were just on. One click, one touch, one slider and they are gone. Away to another channel. Did you notice that when you publish a post with a link to your website on Facebook or LinkedIn, it rarely performs as well as if you publish a similar post without the link?

“Please have your audience stay on our platform so we can monetize.”

Pay for the click via advertising and the barriers is gone.

Let’s imagine for a moment that you are an online merchant, and that you can choose only one channel to send your users to. Which one will it be? Your app, your site, your profile page on a social network?

No, you want them to be on your product detail page on Amazon, with your seller account in the “buy box” ready for a one-click sale. Simply because that is the best selling page on the internet for your product.

Amazon is a strong component of any ecommerce strategy. It is one of the main forces of ecommerce worldwide, and the marketplace is growing at a pace of 40% per year. Amazon Advertising is growing at an even higher pace suggesting the advertising spend is actually driving value to advertisers.

For an equal amount of visitors, an Amazon page typically converts much higher than a website. Perhaps twice as high. The service is fast and reliable, it is providing you with all the information you could possibly ask for, and it is backed by the promise of the lowest price and the fastest delivery. Ever. Amazon is obsessed with customer experience, so things will most likely work out quite well when a user purchases there. Hence a higher propensity to buy. A higher conversion rate.

Are you trying to kill my Shopify ecommerce?

“Do you wanna just kill the ecommerce business I am running on Shopify, or what? Amazon converts really well but I have a much higher margin on my direct ecommerce.”

Let’s do some calculations. Let’s say your ecommerce generates 10k in sales per month, your Amazon account generates 25k. You are spending 3k on paid search for 6k sales. Why would anybody send that profitable traffic into Amazon where margin is lower due to fees? If the Amazon channel converts at a rate which is double that of one’s website, those 3k could generate 12k of sales. Even at a 20% increase in costs – 15% referral fee and then some – your money is better invested in a higher converting channel. And a 100% higher conversion rate is not at all uncommon.

In this scenario, your sales increase by at least 6k and your costs by maximum 2,4k. A net benefit of 3.6k which is likely to increase further due to the organic boost which is likely to happen on Amazon. The higher the sales velocity, the higher you rank in Amazon search results and the more additional sales you generate. The net win could be 5-6k due to the shift to an Drive-to-Amazon strategy.

Downside of driving traffic to Amazon

Many years ago I remember having a client who only sold on Amazon. Running search and social media campaigns for the client was really frustrating because there was no data feedback. We could only control and optimize the click-through-rate for our campaigns and were dependent on the client to feedback on the effectiveness of a full campaign. As a real-time marketer, having a client sign off budget without proof of conversion was a new experience. We had no control on where we were sending the traffic, but as the client asked for more, the bottom line must have worked out for them.

It is common practice for advertising platforms to rely on conversion pixels. These are composed of a few lines of code which are placed on the pages where your advertising traffic is sent. If you send traffic from Facebook to a Shopify site, you will insert pixels which track key events such as a purchase. This communication feedback, allows the advertising platform to collect data on which ad created an event and at what time. And in turn, it allows the marketer or an algorithm to optimize the advertising campaign towards the desired actions. Amazon, of course, does not allow an advertiser to insert a pixel into it’s site. Neither does Facebook, Instagram or any other online platform for that matter.

On Amazon, a part of the solution to this problem is to use a tool called Amazon Attribution which allows you to tag external traffic when it arrives in the platform. The tool is still in beta and clearly in it’s infancy, difficult to set up and is fairly true to the “last-click” attribution model Amazon mostly uses. If a visitor comes from Google or Facebook into Amazon and click on an ad inside the platform before buying a product, the original source of the traffic will not be counted. If it does attribute it to the external traffic, however, you will be able to know from which ad group or keyword the conversion originated and thereby refine your campaigns.

There are obviously some more reasons for not driving traffic to Amazon: you pay referral fees (although it was actually you just referring that sale), you won’t get your clients email address and thus cannot address them directly for recurring sales. Additionally, you actually risk referring your prospects to competitors inside Amazon if they change their mind after the click. So, perhaps this is not a long-term strategy as you are simply increasing your platform dependency.

How to set up a drive-to-Amazon campaign

If you are used to running Google Ads campaigns, there is nothing new in the setup required for an Amazon campaign, but there is some preparation to do in your Amazon Seller account in order to detect the impact of your campaign. Going through the steps below should put you in a good place.

1. Prepare the set-up on the Amazon side

On the Amazon side, you should open a free Amazon Attribution account for tracking purposes. In order to qualify you must have brand registry. You should also make sure that you have access to the brand analytics report which can provide valuable overall insights for your Amazon account. If you are using a tools such as Helium10 you can track keyword rankings. Any keyword tracker will do, but make sure you have one activated. Also find a sales rank tracker such as Keepa which will allow you to estimate indirect effects of your campaign.

2. Keyword definitions

Now extract keywords from your Amazon Sponsored Product Ads and compare with your existing Google search keywords if you have any. We are assuming you are already using Amazon Sponsored Ads for your account. If not, start there rather than driving external traffic to your products! On the Google Ads side, spend some time charting Amazon’s own activity in your keyword territory by using the Auction Insights reports in Google Ads. Steer clear of keywords Amazon is already bidding on. No reason to bid them up.

3. Specific Google Ads set up

Now set up specific ad groups and keywords for your Drive-to-Amazon campaign. Once they are ready to deliver you will be able to pause the old campaign without ruining its history in case you need to reactivate at some point. The Drive-to-Amazon will be void of any conversion data and can’t be managed with automated bid strategies targeting conversion.

4. Campaign destination within Amazon

Within Amazon, you have various options for sending traffic. For product names and any brand-related terms you should send the traffic directly to product detail pages. For category terms, the brand store can be a better destination. There is also an option to create specific landing pages for your campaign. This could improve your measurability.

5. Measuring and evaluating

Welcome to the Bermuda triangle. Every time you change campaign parameters you need to wait for a while before you can see a consistent pattern of anything. Steer you campaign on the basis of click-through rate at least for a week and check your auction insights for overlap with Amazon every day. After a week (or longer if you have a long sales process), the main metrics to look at are total advertising cost of sales and sales growth. Amazon Attribution won’t show you the full scale of impact you generate. You may see lower attributed sales than expected but higher overall sales. Take the evolution of search rank positions into account to see if you are generating more organic sales.

How about we do that Kessel run in 12 parsecs?

The Millenium Falcon in Star Wars is the fastest ship in the Galaxy, right? But when Han Solo does the run to Kessel in 12 parsecs, it isn’t about speed, it was about taking a short-cut through the deadly fog surrounding the planet. Yeah, love Star Wars.

Driving traffic to Amazon is a short-cut to conversion, and thinking about the strategy holistically, there are many cases in which this is better for the end-user although you generate less margin. If it is right for the user, it must be good strategy.

For merchants building their e-commerce site, it must be terribly frustrating to adopt, and for marketers it can be a challenging path. You will need to prepare your campaign in detail, use manual optimization and find the right direction by monitoring a variety of indicators all the while not trusting everything the data tells you.

And finally, the Drive-to-Amazon is contrary to the overall objective of many marketers in 2021, namely that of building more and higher quality first party data. Nevertheless, the Drive-to-Amazon strategy can be a real kickstarter giving momentum to both you e-commerce and your Amazon activity. It is deeply rooted in an approach of enhancing your best performing channel and your best performing assets, it has the capability of multiplying your sales and at the same time giving users a better experience and provide learnings in the process. We are down to the essence of what Amazon provides: a shaky ride for marketers capable of generating outstanding short term results! You should try it!

Advertising is on a roller-coaster ride in the 2020’es. As one of the first budgets to be affected in times of uncertainty, advertising took a dip at the start of the pandemic. It also picked up for new types of advertisers. Then the entire underlying data set for targeting based on user behaviour changed, as people’s financial situations changed, as their relation to touching products and going to shops degraded, and as ecommerce picked up. And then ecommerce boomed and digital advertising rose to new heights again.

And under the hood, things are changing too. Artificial intelligence is making inroads into advertising management, new privacy legislation changes the use of data. And the disappearance of third-party cookies is promising to send shockwaves through the adtech industry.

All the while, the major platforms, also known as the triopoly of digital advertising, Google, Facebook and Amazon continue to experience double-digit growth at 20-40-60%. Search engine advertising, at the core of the Google offering is still on a positive and growing trend. This is despite saturation of click prices in mature markets. And despite a halted growth in the basic market of search, as the majority of the worldwide population now has access to the internet.

But if the the number of searches performed by users – the search market – is not growing, and if click prices, the basic monetization of searches, can’t rise any higher, where is the search advertising journey going? For Amazon, search is still growing, for Bing, the CPCs may not yet have hit the ceiling, but Google might be close to maxing out on those two growth dimensions.  And so it is time to look inwards.

What Search Engine Advertising is today

Most search marketers who have been brought up with Google and Bing ads today agree that Amazon Sponsored Products should be considered as “paid search” too. But they can also get caught throwing display ads from the Google Display Network into the mix. Perhaps because those are still “pay per click”, or PPC, one of the other names search marketers often give to search engine advertising. Furthermore, the overlay of audience targeting which has arrived in search advertising engines in recent years is taking the focus away from the keyword.

Digital advertising is similar to financial investments today. You choose among a wide range of ad streams for which you try to predict future returns on the basis of past performance. Some are short term bets, others are long term. No outcome is ever guaranteed.

In the range of possible options, Search Engine Advertising represents a wide number of streams. Brand keywords, category keywords, automatic targeting, automated optimization and bidding, maximization of clicks or revenue or return on investment. Reliable and unfied tracking is key to succeeding.

But there are also some huge differences between financial investment and advertising. In finance, there is no hidden outcome, no organic growth, no retargeting pools, no brand effect and no synergy effects either. This is where advertising stops being scientific, it has a vast number of moving pieces which interact with each other.

What are the drivers of change in paid search?

For many years, search advertising was always the same: find the right keywords, write compelling ad copy, adjust bids for optimum visibility, send users to highly converting landing pages. It was spreadsheetable because the number of variables remained low.

Over time, however, more variables entered the equation: users morphed from desktop to smartphones, tablets and became multi-device users. The ads evolved with extensions and contextual elements to become multivariate entities with no fixed form or shape. And the audience targeting overlay made the number of variables explode. We explored all of these variables in our 2018 Search Trends report to conclude that something fundamental had changed.

Advertising engines started building optimization tools for managing this complexity. In Google, these tools are known as Smart Bidding. They take a wide range of variables into account, apply machine learning, and target one specific and measurable outcome. In the 2020ies, Smart bidding has become the norm rather than the exception. Search marketers are no longer testing between keywords, they are testing between algorithms.

Smart bidding is the arrival of artificial intelligence (AI) in ad optimization. But it is not the only thing happening. In our 2020 Digital Marketing Report, set in the context of COVID19, we identified radical changes in user behaviour, and disruption in the data as immediate effects of the sanitary crisis. But overlaying these trends, the penetration of AI, and the data challenges became apparent. Privacy legislation is changeing both user behaviour and tracking methods, and the large platforms are increasingly accumulating data and restricting marketers’ access to it. On top of AI, user behaviour and privacy legislation, we are also entering a user tracking paradigm shift as third-party cookies are being phased out and the concept of “cohorts” taking over.

6 inflexion points for PPC automation

Automation is what drove the industrial revolution, and it is also a major driver of the digital revolution we found ourselves amidst. So whenever a search engine suggests automation tools, does that mean you should use them? Let’s look at some good reasons for automating or not.

• SKAGS: “Single Keyword Ad Groups” were a thing. Some agencies even automated processes to build SKAGs in their client accounts. Where search advertising is going today, there are many cases in which SKAGs risk performing worse. Optimization algorithms pull in the opposite direction of the SKAG, as they take more variables into consideration.
• Smart campaigns: Google Ads has a highly automated campaign type known as the Smart campaign. It simplifies campaign creation down to the basics and requires no keyword research. For small operations with a clear goal and reliable tracking, this type of automation can create great results. Just don’t ask how it did it, as you will get very limited insights.
• Dynamic Search Ads (DSA): Enter a URL and let Google Ads find the keywords to target. Then hope to find some of them in search terms reports.
• Responsive Search Ads (RSA): Enter variations for titles and descptions and let Google Ads find the winning combination.
• Smart bidding: Various types of smart bidding exist with the common feature of using a wide range of variables to optimize a search campaign towards once precise goal. This includes bidding on branded keywords to generate more conversions.
• Google Data Studio, PowerBi: Automate reporting of your digital advertising campaigns into dashboards updated in real time.

For most of the items above, there can be a bit of a trade-off. Only the very last point on the list is a no-brainer: absolutely automate your reporting as much as humanly possible.

The challenge with automation is that it applies fast algorithms to sets of data. You need to be careful to provide the right data via reliable tracking and well-considered limits. If your tracking is not 100% reliable, automating campaigns will not work efficiently. If you leave the advertising engine define what data is included, you may see cannibalisation of other campaigns, of other channels.

Keep your eyes on the data and your foot on the accelerator

Search advertising is looking inward for growth. Its engines are aiming to monetize more of the search by using the data to its advantage.

Search advertising for small campaigns are easier to create than ever before. Search advertising for large campaigns is potentially more complex than before, as data access is becoming increasingly blurred, more algorithms are applied and targeting mechanism change. A good approach is to mix and match the various algorithms, and work actively on delimiting and providing the best data for your campaigns to work with.

Keep your eyes on the data and try to interpret it holistically across all your communication channels. Set your strategic goals and pin them down with the right KPIs, then use as much automation as you can to accelerate outcomes. Not quite the autopilot yet.

If you are interested in search trends and outlooks for digital marketing, you will find more in-depth analysis in our recent reports: “Major Trends in Paid Search”, “Paid Search Strategies”, “Marketing on Amazon” and “Digital Marketing in Uncertain Times”.

I was not involved in that project and suddenly felt very happy not to be. Maybe it was just a question of terminology, maybe something else was behind those words. You can’t remarket to something that hasn’t been marketed, can you?

Remarketing had become very popular indeed. Why was that again? Well because it generated “proven” conversions. It generated performance. Certainly not because of its (non-existent) capability to build brand awareness. This was of course before remarketing became really unpopular with users who felt stalked by cookie-fuelled remarketing ads for things they didn’t want or already bought.

Today marketers build advertising campaigns with a nuanced understanding of their audiences. They assume that each individual member of an audience will go through various stages on his/her journey towards the product or service we advocate. They can do detailed analysis of the end users’ characteristics in the shape of Personas and then try to understand in which way they can be touched and influenced in each of the phases they are going through.

The original user-journey mapping concept, the purchase funnel, originates from E. St. Elmo Lewis, who in 1898 described his AIDA model, which has four stages: awareness, interest, desire and action. This model was later adopted in the fields of marketing, sales and, more recently, ecommerce. The logic of the funnel is that you qualify and thus reduce your marketing scope when you pass from one stage to the next. This fits well with the measure-everything philosophy of digital marketing, where goal, transformation and tracking points can be established for each passage.

We used a simplified version of this model in our research for the Search Trends Report and it was clear that the most advanced players in the space we looked at, had a very broad scope of analysis for this. They considered it essential to understand this user-journey and most of them mapped their campaigns to the user journey in one way or another.

For a deeper look into how the most advanced paid search marketers consider understanding the user journey essential and look well beyond their main marketing channel in order to target users during their journey, you should read this article:  Where does paid search fit into the user journey? 

Early digital marketing was linear

When digital marketing first took off – as is the case for many marketing approaches – it was driven by strong converters. The most successful paid search campaigns 15 years ago were all trying to trigger an impulse buy and move along a straight line from which the user was not meant to deviate.

The set up would look like this:

1. find a “money keyword” (a keyword which indicates that the user who made the search query has a strong intent – a buying intent)
2. optimize a landing page with a view to immediate conversion
3. overbid everybody on the money keyword so you would harvest a maximum amount of visitors
4. split test the landing page to continuously improve its conversion rate

In the wild west of early paid search marketing, this approach made people rich. It also drove up competition and therefore auction prices and therefore made the search engines rich too. Those campaigns would be considered “performance” campaigns as everything was related to conversion: investment levels, remuneration of those involved.

And on the opposite side of the spectrum you would have the Brand campaigns. They would simply aim to position the brand in relation with the right queries or on the right websites. They would measure the number of impressions that were generated by their ads and would tie a lot more importance to being at the top of the page than to how many click-throughs that position generated, as there was no direct conversion on the destination site. A landing page could be a homepage.

Those two worlds – performance and brand – were very far from each other.

Digital marketing has come a long way since those early days. In performance campaigns, competition and cost increased, but so did targeting options and data availability, making the playing field more complex. Brand campaigns began to look at longer term value due to improvements in tracking and would start to synchronize with Performance-related activities.

State of the Art in 2019

When looking at hundreds of case studies we would have expected to see quite sophisticated set-ups for campaigns, but one of the things that struck us was that they rarely described more than 2 campaign phases. Two is of course a fundamental advancement compared to the former linear campaigns but with teams insisting on having in-depth understanding of the full user journey, it seemed simplistic still. The first phase would be essentially focused on Awareness (and/or Interest) and the second phase dedicated to Conversion. So where did that Consideration stage go?

In fact, it is hiding in there although it doesn’t have a dedicated campaign phase. All the communication that is being redirected at a Consideration-stage audience is usually tied together either with the preceding phase or the following phase. As an example, for Awareness campaigns, remarketing audiences will typically be addressed for consideration rather than directly for conversion. And in a conversion campaign, to broaden and qualify audiences, « inmarket » audiences will be used. These are covering our Consideration stage, so let’s have a look at a number of ways in which they appear in no particular order:

• Media partnerships
  Many large advertisers establish medium/long-term media partnerships. This allows them to have a continuous visibility and sometimes interactivity with a part of their consideration stage audience.
• Brand searches
  It is well known that searches with brand terms are among the most likely to convert. Brand searches are on the brink of conversion – sometimes the user will be an existing client (what’s their address, how do I log-in), sometimes they will be conversion audiences (let me find the website so I can buy the product) and then sometimes they will be in the consideration stage (on my list of alternatives I visit every brand to help me choose). It is useful to segment brand searchers by the use of customer match and remarketing audiences to identify those who are really in a consideration stage.
• Remarketing to derived audiences
  From an awareness campaign, it is sometimes possible to create derived audiences of those who have seen a video ad or those who have otherwise engaged or simply been exposed. These will be consideration stage audiences not yet ready to convert and in need of positive reminders of a brands existence.
• Inmarket audiences
  Due to increasing amounts of data and machine-learning algorithms applied to user data, it has become possible to define audiences which have a higher probability of Intent. This is very close to the definition of the Consideration stage. These Intent audiences are often labelled “inmarket” referring to a user being “in the market for …”. They are of course sector- and sub-sector specific.
• Prospect nurturing
  B2B digital marketing had a slow start compared to B2C. One of the reasons was that it becomes proportionally more difficult to track efficiently, the longer your sales cycle is and the more stakeholders are involved. And in B2B sales cycles are long and there are typically multiple stakeholders. One of the successful approaches used by B2B marketers is what is referred to as Inbound Marketing. It is a content-driven approach which aims to establish a relationship on the basis of an initial positive interaction and typically register an email address which enables for a process called “nurturing”. This process is typically mapped to a typical sales cycle and addresses the needs of the persona. This entire concept is clearly the materialization of a well-defined consideration stage.

In our research, it is still rare to see campaigns entirely mapped to a user journey. We believe the best possible strategic planning takes the route of audience research, persona building and user journey mapping and that doing this will become increasingly important and also increasingly mainstream as the number of variables in the digital marketing equation continues to increase.

We have, however, come a very long way since the initial duality of Performance on one hand and Brand on the other. Most advertisers are trying to build their path down from Awareness or up from Conversion to try to guide the user at every stage towards their offering. And excellent marketers also think beyond conversion as new opportunities in loyalty and advocacy are arising.

It used to be quite simple. When taking over a Paid Search campaign you would automatically consider that whoever had managed it before was a jerk so you would rebuild the campaign from scratch: new keywords – or rather more keywords, new account structure, fresh ads and then optimize, optimize, optimize. You would add in a “Bid Strategy” which was never really something well-defined but it let you use the magical word “strategy”.

So many things have changed in the way paid search is structured, in what is possible to do and in the integration with other channels. And with the arrival of retargeting, of audience targeting and of a host of dynamic variables that can be used to model a campaign, we can really talk about search strategy in 2019.

In our research for the Search Trends report, we identified a number of these strategies and in the following we will look at 2 families of strategies that have arisen. Today’s more ambitious strategies are either measurement-driven or insights-driven and require a deep understanding of the user journey, which leading teams consider essential and for which 52% of them now use “personas”, a segmentation technique whereby you can provide actionable input for the more complex types of campaigns.

Measurement-driven strategies

Measurement-driven strategies are based on what represented the original reason for paid-search success: reliable tracking of sales originating from digital advertising. Amazon are now entering the digital ads scene and already represent 10% of Google Ads, expected to double to 20%. Within Google, meanwhile, there has been a gradual budget shift from search ad

s to shopping ads. This gives birth to a winning search and shopping strategy. The other metrics-driven strategy we observe is the drive-to-store strategy, which addresses the complex issue of using digital advertising to drive users to a physical point of sales. The main enabler for this strategy is the store-visits metric, allowing for increasingly reliable measurement of visitor flows in a physical location. Leading paid-search teams selectively embrace offline activity: 32% of our sample systematically use offline tracking, and for 14% offline is part of their offering. We predict that this will rise in the future.

Insights-driven strategies

The other group of successful strategies is insights-driven. They require in-depth user research to understand the user journey, and make use of personas, which they map against it. They then use several campaign steps, often involving different channels. They are heavily reliant on measurement as they target a large span of the user journey and collect more measurement points; they need a model to attribute value correctly to the various channels and providers involved – in other words, a reliable attribution model. The leading teams use advanced attribution models in their projects and have moved away from the much-maligned last-click attribution model.

The audience-led strategy has several phases and uses audience building for later conversion to fit with end-user decision cycles. Retargeting, audience sharing and negative audiences are the mechanisms used to keep the target on the user, improve budget and maintain brand-image consistency. We observed full adoption of simpler audience-targeting mechanisms, and we noted that adoption of the more recent machine-learning-based “in-market audiences” by leading paid search teams, at 77%, is as high as traditional geographic targeting as shown in the illustration.

The data-driven strategy differs from the audience-led in that it focuses more on user context and external factors than on the user state. As such, it injects external data into campaigns in order to improve campaign effectiveness. Data-driven campaigns are technically complex and use a variety of data sources: finance, weather, news, events, etc. The only limit to what data can be actioned is the insight and creativity of those actioning it, and these strategies are almost impossible for competitors to detect and counter.

The common characteristic of the four paid search strategies mentioned above is their potential to drive double-digit performance improvement over existing strategies. They are also emerging strategies that will allow for continued success in 2019 and beyond.

This post is based on an extract from the #searchstrategyreport found on https://www.innovell.com/search-strategies-report/