Prospecting: How to Use Programmatic to Attract New Customers

Programmatic display can be used to fulfil a number of strategies for advertisers, but primarily it’s executed in two ways: retargeting customers who have been to your site before, or prospecting new users who are yet to experience your brand online.

Let’s look at the latter and how it works in practice.

Why prospecting?

So, you have a website. It’s getting traffic and driving sales. Fantastic!

The next step for any brand is getting more traffic. No matter how big you are, every client I have ever worked with – from large high street names whose brands are instantly recognised, to smaller SME companies who need to increase brand awareness – a marketing manager’s objectives are always the same: year on year growth. Whether this is sales or visitors, it’s always about getting more. If you’ve been a marketer for some time, you no doubt understand this well!

What is it?

Put simply, ‘prospecting’ is finding new customers. With programmatic, this is achieved by using technology to search out and target undiscovered prospects on the web.

Prospecting starts with a conversion pixel placed on your sale or signup confirmation page – the page you definitely want to increase visitors to! This pixel collects non-identifiable data from a user’s cookie every time they convert, and sends it to your Demand Side Platform (DSP), where it’s stored securely.

Over time, as more people convert, the DSP builds an image (model) of what your ideal customer looks like: their interests, propensity to buy online, geo-location and various other behavioural indicators. On an individual basis, this information can’t do much, but with scale (more data collected) a series of accurate lookalike models can be created.

How does it work?

The following process can have many different parties involved, but let’s look at it in simple terms.

When a user lands on a publisher’s site, an auction begins, where advertisers compete to serve their ad. This publisher sends the user’s cookie data through its Supply Side Platform (SSP), to be received on the buy side by the DSP. As with the advertiser’s own customer data, these cookies infer the user’s behavioural traits.

The DSP then refers to the advertiser’s lookalike models to identify whether or not this user is a relevant viewer for their ad. If the user’s data shows many of the same attributes as the models, the match is made. The DSP then decides how much to bid to serve an ad to the user. Depending on the likeness between the model and the actual user, the bidding price will vary: the more alike the two, the higher the value of the impression, therefore the higher the maximum bid, and vice-versa.

If the user’s data does not reflect that of the models, the DSP will opt out, or place a much lower bid.

If the DSP’s bid wins, the advertiser’s ad is served – the prospect has now entered their funnel.

If this prospect goes on to convert on the advertiser’s site, new insights from their data become available for the advertiser to use in future lookalike modelling activity.

Prospecting_image

What else do I need to know?

From the offset of the campaign, data is essential – the more data you have, the more accurately your DSP’s algorithm will be able to predict a user’s behaviour. It’s worth noting, however, that there are other tools at our disposal, which are used as we collect this data; it isn’t all based on algorithms.

For many years, 3rd party data companies have been collecting consumer information. This is invaluable for the performance of campaigns, and with this 3rd party insight, you can use the following targeting in addition to algorithmic learning:

Contextual Targeting: based on users’ interests, according which websites they use and how relevant these are to your brand or product.

Demographic targeting: based on age, gender, profession etc.

Behavioural targeting: based on people’s propensity to purchase online, their spending habits and other purchases.

Geo-targeting: based on people’s physical locations.

These additional targeting methods allow us to manually teach our algorithm at the same time as we collect data. It also allows us to offset any ‘false learnings’, for example, events or dates in the annual calendar that have the potential to misrepresent people’s behaviour.

There’s much more to prospecting than meets the eye, but if you’re already harvesting and managing data on new and repeat customers, you’ve already made the first step.

Ready to move forward with prospecting, or find out more?

David Ayre, Performance Director, MDNA
E: david.ayre@admedo.com
M: +44 (0)74 2712 6395

Dave_circle_crop