What is Programmatic?

Programmatic advertising is about using software to intelligently purchase ad inventory online. With ad inventory (the volume of available ad space) increasing on a daily basis, programmatic is designed to remove the human element of buying media space. It can work either working alongside traditional media buying, or it can eliminate the practice entirely. Programmatic allows advertisers to automatically purchase ad inventory based on their particular objectives, the audiences they want to attract, and when they want to attract them.

It means that buying ad space is now much more manageable, with programmatic minimising the level of human intervention necessary. Advertisers now have an unparalleled level of visibility over the performance of their activity.

Programmatic advertising allows brands to adopt a much more intelligent approach to buying ad inventory, making the process highly targeted and extremely cost effective. Brands can use data to target users at an individual level based on their online activity (remarketing) pushing a highly targeted and highly relevant message to that audience based on the user’s previous level of engagement on an advertisers website. Advertisers can now purchase ad inventory based on factors such as a user’s loyalty to their brand, their browsing habits, their purchasing patterns or their device – pushing individual ad creative to each user based on the likely needs or trigger for that user.

What is Real Time Bidding?

Real time bidding, or RTB, is a sub-category of programmatic advertising. Whereas programmatic is largely about the purchase of guaranteed inventory, RTB is about bidding for a particular inventory via an auction with the ad position being awarded to the highest bidder. Essentially, it works like Adwords, but for display. Ads are charged on a CPM bid (cost per thousand impressions) as opposed to a CPC bid (cost per click).

Ad networks, such as Doubleclick, are investing heavily in creating features that allow advertisers to really refine how they bid for ad inventory online. Advertisers can now automatically adjust their campaigns based on macro environment variables, such as regional weather, television schedules, device usage, local events and various other factors; all of which have historically been proven to impact online user behaviour and demand.

It means that advertisers can automatically increase their bids to secure an inventory when they believe that a user is likely to be most valuable, and reduce bids when they feel that a user is least likely to convert. For example, weather or seasonality may influence the purchase of a product, and so an advertiser could use localised weather data to automatically increase bids in certain weather conditions across specific devices such as increasing bids on mobile devices during hot or sunny weekends or evenings.

What RTB fundamentally allows is for advertisers to drive much more value from their ad campaigns, whilst minimising the level of human intervention needed. This is based on the greater level of granularity and control over targeting, with a shift from buying inventory based on the perception of where an audience consumes information or media online (traditional media placement buying) to buying based on a multitude of different targeting methods. This allows you to connect with your target audiences more effectively.

Changing the approach to display advertising

Display campaigns have historically been purchased through traditional methods for buying media inventory, and have relied heavily on placement profiles from media agencies and publishers. The purchase of this inventory was largely based on the product or services that they offer, and whether the audience of that ad position met the advertiser’s target demographic.

This isn’t necessarily a bad way to buy display inventory but, with advertisers now focusing more on acquisition as opposed to brand awareness, this rather rudimentary way of purchasing inventory doesn’t offer as much value. The method generally takes longer to optimise and has resulted in a notably higher level of waste.

Programmatic advertising, when combined with RTB, allows the advertiser to get closer to their audience through more sophisticated targeting methods.

Advertisers are now able to reach their audiences at key “moment of truth” time periods within the buying cycle. Although that audience is not necessarily consuming content or media that is relevant to the product or service, we know that the user is interested in the specific product or service due to their previous engagement. What this means is that you are now buying audiences via RTB as opposed to buying specific ad inventory on a single or multiple placements.

This method enables you to reach your audience on a wider range of inventory that was not previously possible. In addition, advertisers can be much more specific about the audience segments that they are targeting, and customise their activity so that each individual segment only sees advertising messages that are relevant to them.

Historically a retail advertiser may purchase inventory on the fashion section of a national broadsheet, which would display a single ad to all viewers of that specific section of pages. With programmatic and RTB, you can still reach the visitors of that page, but with customised messages that are based on products they are likely to be interested in. You can also reach them on other sections of that website, and continue to reach them on other sites away from that domain.

This method helps to minimise wastage and cost by not having a blanket targeting approach for all viewers of that particular domain or section. Instead, advertisers are targeting those users who are either previous visitors, or who closely match the profiling you have conducted of your current target consumers for prospecting activity.

In this more efficient model, a very small level of the campaign resource (namely, time, budget and initial audience profiling) has been invested in understanding the audience at the very beginning of the campaign. Gathering the learnings from this insight ensures that you can target a more relevant audience to drive stronger engagement. This provides a solid foundation in order to scale up your activity

Building your audiences for Programmatic Advertising

In the early stages of a campaign, advertisers should always deploy both conversion based and remarketing pixels on-site. These pixels should be deployed a significant time before a campaign is launched in order to build a comprehensive understanding of typical on-site behaviour. There is no set timeframe for this as it will differ significantly based on the level of traffic each site generates on a daily basis.

The remarketing tag is vital. This allows advertisers to gather data on their current customers and it will provide a single view on how users behave across a wider digital landscape, away from your website. Overlaying this insight with conversion tag-based data will allow you to segment between converting and non-converting website viewers, and also where the user sits in the online conversion journey (based on the level of on-site engagement).

Always start with your remarketing activity first, before going live with any form of prospecting activity. The results and learnings taken from these campaigns will help instruct the strategy for the pure play prospecting activity. This will also make it easier for you to reach new audiences whilst not forgetting about your current customers.

Audience building & segmentation

If you want to reach your audiences through programmatic and real time bidding, you need to know where that audience resides online (in other words, where they are digitally active) and how they engage with brands. Without this insight, your paid campaigns can become an increasingly costly exercise. Your remarketing and CRM data provides you with a huge amount of data on your existing customers and visitors, including your frequent customers, your high basket value spenders and your dormant account holders.

By segmenting these audiences in a greater level or granularity utilising some of the key considerations below, you can create a clear picture of your core customer base (remarketing) and your target audience for prospecting.

First-party based segmentation & targeting

Geographical

Analyse performance of your remarketing tag by region, or analyse your current customer base to spot location trends. Is your product or service popular in a particular city or region? Identify these areas and try to map specific products or services to particular locations.

Behavioural

By understanding the on-site behaviour of your existing customers and visitors, you can get a greater understanding of what your prospective new target audience could also behave like. You can then extrapolate data that can be used away from your website.

For example, you could identify which devices your users are accessing your website on, what operating systems they are using and which times of day they are visiting your website. All these can be used to build a much clearer picture of your prospective audiences.

Social profiling based segmentation & targeting overlay considerations

Social profiling tools are a great way to get a much clearer picture about your existing customers, which supports your overall prospecting strategy for attracting new audiences that have not been previously exposed to your brand. One way to use these tools is to take advantage of your first party email data from your existing customer base.

Social networks allow organisations to upload this data as a custom audience. From there, advertisers can utilise these insights to expand on the data and identify behaviour characteristics and trends that were not previously available. For example:

Interests

Utilising social insights tools will help give you a greater awareness into what your current customers are interested in. This provides strategic insight into the types of content, and the content themes, that they are likely to engage with across programmatic advertising platforms.

You may find that these trends have a direct relation to the product or service that you offer (for example, an airline will naturally want to appeal to people who demonstrate an interest in foreign travel), but it is likely that their interests are not intrinsically linked to your brand. This gives you valuable insight that you can utilise to target users based on their inherent interests, as opposed to just the types, categories or topics of content they are consuming. These interests, when mapped back to placement tools, will also help you to decipher which creative formats and content you should look to deploy across activity.

Demographics

Additional demographic data can be extrapolated from social media profiling tools, enabling you to go deeper than just the gender and specific age range of a target audience. This data looks at additional factors that could affect how you communicate to your current customers and prospective new customers. Typically social media platform tools will enable you to gather insights such as parental status, age ranges of children, expectant parents, relationship statuses. This could open up different products or services to target to users and create potential upsell opportunities

Household & Employment

Household data can be gathered from social media profiling tools to give an indication on the residential status of your current customers, such as whether they are homeowners and the types of property that they occupy, as well as additional employment based information such as employment function, industry and level of seniority. This can be a key factor in regards to deciding the value of your bids, and the products that you promote to these audiences.

This data collected should then be cross-referenced across each of the individual segments previously identified in order to understand where there may be areas of crossover from one consumer segment to another. If, for example, you find a correlation between interest groups and the time of day audiences visit your site, you can use this information to specifically target that segment at a that time and/or their preferred device. This method can be applied for both existing and intrinsically linked prospective new customers.

This process allows you to spot clear correlations between audience segments and, most importantly, allows you to understand what is likely to trigger them into performing a positive consumer action. These key incentivisation points will ultimately dictate which messaging is targeted to each audience group, and at which point in the buying process.

Optimisation of programmatic advertising is a fine art and at the heart of all optimisation activity should be your business objectives and KPI’s. The optimisation process should always lead back to your overall business objectives, structured around these core metrics and through the granular segmentation of campaigns and line items within the accounts.

Looking at how the account is structured, and analysing this against your historical first party data, helps you to identify the correct bid method for each of the individual segments.

Clustering each audience based on tangible behavioural activity is crucial, as this allows you to segment your audience to an accurate level of granularity. This gives you greater control in deploying campaigns that deliver a message to only those customers that will deliver an ROI. Analysis of each bid method and determining what the likely outcome would be from the deployment of that bid method ensures that your ad spend is being deployed correctly.

Structuring your activity

Line items within campaigns will then add a greater level of granularity to targeting activity, and it is at this level that you can break out additional targeting considerations such as demographics and interests allowing you to optimise these clusters separately for maximum impact of optimisation, and for clear, granular and structured reporting

Line items within any display advertising campaign should naturally evolve based on performance. Don’t be afraid to break out your campaigns with further line items based on historical performance that the campaign has delivered, as this extra level of granularity is where you can make huge improvements to your overall performance.

Activity Deployment

In its simplistic form there are two ways to categorise any programmatic activity; remarketing activity and prospecting activity. These two elements (like in any form of biddable media) should work in parallel with one another but should be segmented out against each other within any programmatic structure. The below sections highlight best practice approaches to both remarketing activity (site interactors and dormant users) and also pure play prospecting activity.

Dormant user remarketing

There are two core forms of remarketing that should be at the forefront of an advertisers mind. One looks at remarketing back to dormant users who have strayed from the advertiser’s brand, while the latter focusses on more in-market based users, segmented by an on-site engagement model.

You need to consider both the length of time that you have set your cookie window duration for users to still be eligible, as well as the typical frequency of purchase or a benchmark that reflects how often a user should be coming back through to site. The diagram below highlights the way a typical FMCG retailer with a high return ratio may look to categorise their dormant consumer base, and target users based on the level of time away from the brand.

Whilst it may sound counter-intuitive to market to these audiences, this data actually provides you with a wealth of information that can be useful in not only re-engaging those audiences, but also support your consumer demographic profiling.

The less amount of time that a user has been dormant, the less incentivisation they are likely to need and the fewer touchpoints they are likely to use. This means that you should be spending less in attracting this audience overall, but applying significantly higher bids for these users along with some level of incentivisation.

However, those audiences that have not engaged with your brand are likely to need a significantly greater level of incentivisation. They have likely strayed from the brand due to a single factor or multitude of factors, and in many cases it is about rebuilding brand perception and brand loyalty with these users, just as much as it is about driving conversion volumes.

The cost of acquisition of these users is generally higher than that of a user who has been latent from the brand for a shorter period of time. With this is mind, advertisers should look at a lower bid strategy in order to manage this outlay.

In regards to structuring creative, you need to look at intitially deploying engaging creative ads to draw the attention of these users, alongside heavily incentivised messaging (such as discounting or promotions) to draw them back to the brand.

Site visitor remarketing

Site visitor-based remarketing is the most frequently deployed type of remarketing utilisation and the most effective form of remarketing via programmatic advertising. This form of remarketing activity typically focuses on your site users within a 30 day cookie window, and uses a segmentation of audiences based on their previous level of engagement with the advertiser. This allows you to customise creative, ad formats and bids based on where the user is in the online customer conversion journey, and based on their previous purchase history.

Taking into account the previous level of engagement of a user is important for three key reasons; areas, relevancy and optimisation. Ensuring that you are as relevant as possible to each of the customer segments through accurate and engaging ad content, combined with the most appropriate landing page, can help accelerate the user conversion journey and reduce the number of “potential” touch points from other online channels. This aids delivery of last click, post click and post view conversions through remarketing and lowers the lifetime cost per acquisition (CPA) required to acquire or re-engage that customer.

Taking the example of a consumer who views a product page, an advertiser would typically look to deploy dynamic based remarketing across programmatic activity highlighting the products the user has viewed online in order to try to bring the user back through to the site to make a purchase. Given that you know these customers have not made a purchase previously, or added a product to the cart, you could incentivise this with a “free delivery” on your first order message, or 10% off for new customers.

For a previous customer however, we would still recommend using dynamic remarketing based on the products that the customer has bought in the past. Ensure from the feed that the product they purchased is not displayed to them (unless buying cycle is frequent) and focus on similar products, upsell products or products that “complete the look”.

Given that they would not qualify for a new customer discount, incentivisation and messaging needs to be more focussed on “chosen for you” or “tailored to you” with an emphasis on pushing exclusive offers or discounted products before opening these up to non-converting website visitors.

Prospecting

Prospecting is the process of identifying and targeting new customers or leads, who have not had any previous exposure to the brand. Utilising information you know about you current customers, and applying these assumptions to your prospecting audience segmentation and targeting is paramount to the success of the activity.

It is important to stress that the best programmatic campaigns contain multiple targeting elements and to deliver the best performance and cost effectiveness, advertisers should look at deploying multiple elements of targeting outlined below in order to stand the greatest chance of delivering your overall business objectives and KPI’s.

Contextual targeting

Contextual targeting can be used to reach out to prospective users based on the content of the information that they are consuming at the time across a number of related websites or pages.

This can be done on both a generic and an incredibly granular level of targeting, with ad creative that targets much more specific keyword sets that will typically carry higher click through rates. As long as you, as a brand, ensure that your value proposition is replicated throughout the customer journey, to each of the specific audience segments identified this should lead to greater conversion rates on-site.

This form of targeting essentially targets the context of the search, whether that relates to a particular geographic location, a particular size or model, or anything else that can indicate that the user is looking for something more specific than a general keyword search.

Demographic

Layering demographic segmentation to your campaigns and line items ensure that you can bid more effectively for your most valuable segments within your audience as a starting point. This element will also give a greater detail of control when it comes to deciding which creative types, products, messages each gender or age segment receives. This will help to stimulate a higher CTR% by matching the correct products or services to your audience segment who are most likely to engage with this activity and go on to purchase.

Location

Overlaying first party customer data, with learnings taken from your search campaigns in regards to performance at a location level, again provides a higher degree of granularity and control over your campaigns. Geo-Targeting also works much more effectively when promoting products or services that have a higher perceived degree of significance to your target consumer in those locations. A recent example would be online sportsbook companies up weighting CPM bids in locations in and around Liverpool for users to sign up utilising Liverpool enhanced odds to win their upcoming fixture or a competition.

Third Party Data / DMP’s

Third party data or the use of demand side platforms (DMPs) are also recommended across programmatic activity in order to expand the reach of your activity. Utilising a DMP on the publishers side can also be linked to the SSP (supply side platform), and can allow you to tap into a further greater level of granularity about a particular networks cookie pool. This allows you to add extra layers of sophistication to your targeting using pre-defined criteria or buckets in order for you to select and potentially drive a greater response for your target audience.

You can add these additional targeting layers on for an additional cost on your CPM bid, and you will only be charged for the most expensive segment which you will select. Therefore, if users appear in multiple audience buckets, you will not be charged multiple additional costs for each audience.

Utilising these areas can typically allow you to refine your audience with overlays such as “users more likely to buy online”, “users with a credit card”, for example. There are hundreds if not thousands of additional targeting criteria you can define through third party and DMP data.

Personality Profiling

Personality profiling tools are also widely available in many DSP’s. Personality profiling allows you to utilise the science of psychology in order to understand further behavioural traits and a users propensity to purchase your desired goods or services.

These additional targeting overlays incur additional costs on a CPM basis, but give you a greater detail of insight based on how they have been categorised by providers such as VisualDNA. Such tools allow you to understand an audience’s openness or receptiveness to particular products or services and buying methods. Linking these back to your first party data and objectives ensures that your prospecting activity is given the greatest chance of succeeding to hit your objectives and KPI’s.

Viewability

Viewability is a crucial element of additional targeting layering across your programmatic activity, and one that is often overlooked by advertisers. Naturally, as you would expect, adding in additional layering to targeting to ensure you are focusing on the placements which are most viewed does carry additional costs on top of your base CPM. However overall cost effectiveness and efficiency is much improved. This can help to eliminate and will certainly reduce the volume of impressions that you are charged for that are not in view, or fully visible, and have a major impact on overall performance.

Ensure that elements of viewability play a major part in your overall targeting and optimisation, but also tread on the side of caution as going too aggressive and granular across this activity may significantly reduce the reach and visibility of activity.

Measuring the success of your programmatic campaign.

To get the true value of programmatic and RTB activity, advertisers need to be realistic on the returns they expect to get from activity. Similar to traditional media display buys, the true value of display activity is the role it plays in the user conversion journey, and its attributable value to other digital channels in the mix. Those advertisers who are expecting big results from programmatic and RTB activity, but are still utilising a last click attribution model must first re-evaluate this before commencing any activity in order to get an accurate understanding of the impact their activity is delivering. There are two main elements to this:

Post Click & Post Impression Tracking

Approximately 90-95% of all display conversions that are reported in DSP’s will be a combination of both post click and post view conversions, with the remaining 5% being made up of last click conversions direct from the ads served themselves.

Advertisers must ensure that they have a clear picture on what impact display is having on other channels and, most importantly, which channels RTB activity is delivering assisted conversions for. Most free analytics packages will go as far as tracking post click using conversion paths, but if you are not using an analytics tool capable of tracking post impression data such as Sizmek or Atlas (to name a few of many), then you will never see the full touch points on your customers conversion journey. This will hold you back in understanding the impact a display impression has had on the course a user decides to take to engage with your brand or convert.

Attribution

Moving away from a last click attribution model and analysing different attribution models whilst deploying display activity is a necessity. There are many attribution models that work well for display advertising and digital as a whole such as, time decay attribution models or position based attribution models.

When deciding on the right attribution model for your brand it is vital that you ensure this method is in fitting with your current digital channels and future activity. Be sure to establish a holistic view of how each channel is interacting with one another, and what impact your channels are having across the overall digital performance.

Key Takeaways

To get the most out of programmatic activity, you need to start with a strong understanding of what you are trying to achieve from deploying this activity. Whether you are trying to increase brand awareness, attract new customers, stimulate online conversions or any other business objective, you need to have this key goal in mind before you commence with any form of activity.

You also need to have a fundamental understanding of who your current customers are, how they like to engage with you, and how they behave not only on your website, but online generally. The stronger this insight is, the more targeted and more personalised your RTB activity can become.

Segmentation of audiences is crucial to ensure you are serving the right creative, in the right places, at the right time to prospective and returning customers, and that you have a solid foundation in which to scale your activity in a way that is cost effectively.

It is also important to consider multiple targeting options, rather than hanging your hat on one or two targeting strategies. This will really reap dividends in the long-run and allow you to dig deeper into your audience segments, gaining granular insight that helps you to identify your most valuable audience groups.

As the vast majority of display activity is not likely to result in last-click conversions, ensure you have the right attribution methods and accurate cookie windows in place. This allows you to gain a greater understanding of what incremental volume programmatic and RTB activity is driving for your business, and what other channels this activity is having the greatest impact on.