If you haven’t run programmatic campaigns yourself before, and don’t know where to start, let us give you a few pointers.


You may have a good understanding of who your target audience is, but if you have not run programmatic before you should try to keep your targeting strategy quite broad to start with and look to refine this down as the campaign runs.  See optimisation 101 for more details on this.


You also don’t want to add too many types of targeting to a single line item as this can be restrictive and cause under pacing. And if you don’t get enough traction you won’t find performance as easily.



Use line items to test different strategies


We would recommend using a similar structure to this for a first campaign:

  • Contextual Targeting Line Item - They type of site the user will be on
    • White Listing can be used if you have a site list or you can create one once you have some performance data - when creating site lists try to make these as broad as you can rather than just a handful of sites (depending on how much budget you want to try to deliver there)
  • 3rd Party Audience Targeting Line Item  - Finding users with a particular demographic or interest or intent wherever they may be browsing
  • Retargeting Line Item - Finding your 1st party site users wherever they may be browsing
  • Open Strategy Line Item - meaning no targeting bar any exclusions of things you really don’t want (this is purposefully broad to allow the machine learning to look for pockets of performance that might exist in places you haven't thought about)


We would also recommend you start with all suppliers (SSPs) and device types to keep your setups broad.  You can then refine these down later.  See optimisation 101 for more details on this.



Assigning budgets and flight times for success


To drive campaign performance it is just as important to give the system time as it is enough budget.  For a first campaign we would recommend a 4 - 6 week flight and at least a £5k budget.  You should also be prepared to allow for the entire campaign flight plus your conversion pixel look-back windows (if you are conversion tracking) before you will see the complete performance of your campaign.


Also try not to split your budget up too much.  As much as it’s good to test multiple strategies, don’t create so many that you end up with a very small amount of budget per line item as this won’t allow the machine learning enough data points to optimise efficiently.



Maximise your reach with creatives 


Use a full range of creative sizes in your set to enable to get the most reach in the market place.  Again you can refine down later once you see what works for you both in terms of delivery and performance.


But don’t use too many creative sets on a single line item. (i.e more than one set of a full range of creative sizes). Again this can negatively affect the machine learning and optimisation as you will end up with your budget being split across too many creative IDs for it to make decisions on performance as easily.  If you want to test different creative messages or designs it can be better to use your line items and different strategies for this depending on your budget.  We can help to advise on this if you are unsure.



Make sure your tracking is in place


This is probably the most important things to ensure you get right, as without it you will not truly understand the impact of your campaign.


Click tracking 


We take care of click tracking for you in most cases, but if you are using HTML5 scripts (rather than zip uploads) or Ad Server Tags you will need to ensure our click macro is added to enable our system to track clicks.  Without this the system will not be able to properly optimise your campaign. 


Our click macro is:

${CLICK_URL}

Or the encoded version is:

${CLICK_URL_ENC}


Please make sure you use the correct version and that it is placed in the correct place in your tag - contact your ad server provider if you are unsure.


It’s also advisable to add a cache buster to your tags (unless it already has one) so that you don’t over count impressions in your ad server.


Ours is:

${CACHE_BUSTER}


And again, should be placed in the correct place in your tag.


Conversion tracking


Wherever possible tracking pixels should be placed on the landing site to enable the system to track conversions against users exposed to the campaign.  You may already have conversion goals set up via Google Analytics but this will only give you a small snapshot of your activity as it relies on an instant post click conversion to track effectively.  In programmatic we see huge traction from users who are exposed to impressions but don’t actually click on them.  We call these post-view conversions and as GA does not provide impression trackers (in the free version) you will not be able to see this from your GA interface.

 

And potentially even more importantly with our the tracking pixels our system will not be able to optimise to conversions for you so you will miss out on the benefits of the machine learning.


The tracking pixels must also be added to your line items under 'Conversion Audiences' in the ‘Audience’ section of the line item set up to ensure these are tracked to the campaign.



Which optimisation and bid type to choose


Optimisation strategies you have available are:

  • CTR - to maximise clicks

  • CVR - to maximise conversions without concern for clicks

  • CVR + CTR - to maximise conversions and hit a CTR goal

  • Margin - to maximise reach/impressions at low cost, but with no performance goals

  • Completed Views - to maximise completed video views (video line items only)

 

And Bid Types are:

  • Max CPM - this is the most you are willing to pay per 1000 impressions

  • eCPC - this is a cost per click goal for your overall budget

  • eCPA - this is a cost per action/acquisition goal for your overall budget


When using a Max CPM bid you will generally see your clearing price eCPM come in at a lower rate than your max bid, as in a second price auction via RTB you will only pay 1 penny more than the next highest bidder.  We apply Automated Bid Adjustments (bid-shaving) to your Max CPM bids, to counteract things like header bidding / first price auctions, ensuring you always pay the best possible price for your inventory. Predictive modelling also ensures you bid for the users with the highest propensity to perform against the goals of your campaign.


For click based campaigns we recommend you start with CTR optimisation and Max CPM bid type.


For conversion based campaigns we recommend you start with CVR + CTR optimisation and Max CPM bid type.


For video campaigns we recommend you start with VTR optimisation and Max CPM bid type.


We generally find these will out-perform the straight ‘eCPC’ or ‘eCPA’ goals, unless you have a long-standing campaign which has been consistently performing (and you are sensible with your settings and optimisations).  You will probably still notice your buying patterns will change a bit when you first switch to and eCPC or eCPA goal, but this should settle as long as your goal is achievable.  If the goals cannot be achieved you will see your campaign buying will slow down (and eventually almost stop) and you will start to under pace so keep an eye on this if you test these bid types.


If you switch your optimisation or bidding strategy and/or your goal is too hard for the machine to achieve, you will see your buying patterns change. The most common visible tactic here will be a drastic reduction in CPM so that the platform can achieve more reach (a higher volume of impressions) while it looks for performance.