Retargeting has always been used as an efficient tactic for e-commerce campaigns, but can brands use programmatic to replicate retargeting success for prospecting and user-acquisition? Writing exclusively for ExchangeWire, George Levin, CEO and co-founder, GetIntent, explains the advantages of using programmatic for e-commerce and the importance of bringing these tools in-house for the best success.
Marketing for e-commerce has been always been performance-oriented, while success has been measured and driven by very strict metrics that historically have included: ‘cost per order’, ‘cost per user acquisition’, or even ‘customer lifetime value’. At the same time, historically, programmatic (everyone defines programmatic differently, here I define it as buying through DSPs/trading desks using OpenRTB or a similar protocol) was always more focused on brand awareness with less aggressive metrics: brand lift, reach, click-through rate, and/or viewability. For a long time, programmatic for e-commerce was limited mostly to dynamic retargeting – in which the market was very consolidated, being dominated by a few companies like Criteo and AdRoll.
The whole dynamic retargeting concept is quite simple: programmatic platforms (e.g. Criteo, AdRoll, etc.) capture clients' first-party data with their own pixels. It then produces ads containing browsed products, cross-sold goods, or just simply products left in shopping carts. This technology is well-known and, by definition, called 'dynamic creative optimisation' (DCO). The question there lies: why did e-commerce platforms used to spend significant budgets only on retargeting within programmatic buying, and why didn’t DCO work well for prospecting? The answer is simple, performance was poor and couldn’t be compared to paid search efficiency due to three reasons: lack of accurate reliable data, and the ways to activate it, lack of flexibility in buying (using custom algorithms), and inaccuracy of attribution models.
Dynamic retargeting is simple because it uses high-quality first-party data owned by the brand. Simply put, the more time a consumer spends on a website, the higher the probability of purchase. Additionally – based on the consumer’s previous history, it may also be easier to understand/predict what kind of products they find the most intriguing. On top of all that, final execution is made with one partner, who collects data, builds creatives, owns AI-optimisation and, finally, owns a bidder.
It’s not easy to replicate retargeting success with prospecting because it tends to need more involvement from the in-house e-commerce team. More tests should be done, and more hypotheses should be tested, to find good and efficient ways for prospecting using programmatic buying. In other words, retargeting in general is very standard and doesn’t depend on vertical/niche (can be 99% automated), while prospecting is always different and should be based on very specific knowledge of brand and its customers.
Thus, I believe efficient prospecting and user acquisition can actually be done only in-house. Unfortunately, not every e-commerce platform has budgets to keep dedicated in-house programmatic teams. In my opinion, today’s e-commerce brands can run successful programmatic campaigns with a small team of only one to two people. Ad tech has now become a commodity, rather than a perceived 'rocket science' that it used to be five years ago. Now brands don’t need to hire expensive programmatic gurus to set up sophisticated programmatic campaigns/logics. Everyone can afford a white-label DMP, which may be used to activate their own first-party data along with high-quality data from reliable partners. This DMP could be easily connected to a white-label DSP or bidder, which is available at a much cheaper cost than using existing DSPs, and will support custom bidding algorithms.
At the end of the day, a very accurate, smart, and custom campaign can be executed in-house using several vendors. Here are a few examples of campaigns that can be executed by a small team:
– DCO creatives that change the bid, the offering and message is based on location, gender, and income. If the customer is already in the user base, the bid will be skipped to avoid wasting prospecting budgets on existing clients.
– Selling merchandise based on an event that was recently attended. For example, after buying tickets online for a concert, and attending the concert, merchandise and gear based on that artists, or similar artists, will display programmatically.
– Car dealerships offering a car that is available in a specific user’s zip code. It also may skip the bid if a user doesn’t travel far from his home (probably he/she doesn’t need a car, so there is no need to waste money).
The last, but certainly not the least, important problem is attribution accuracy. Unfortunately, most e-commerce brands still use a last-click attribution model. In layman’s terms, if a customer journey consists of several touchpoints: a click from a social media ad, programmatic display/video view and, finally, paid-search click, most buyers attribute this order to paid search only. Another bad scenario is when an e-commerce platform uses post-click attribution. This is when an order is attributed to all sources that generated a click within a post-click window (usually 30 days). This may simply end up with attributing the same order to five different sources. I don’t recommend trying to solve this problem in-house. There are a few very strong vendors that can handle this problem. They use AI to calculate the actual impact of every touchpoint and how it has affected the final purchase.
The conclusion is that programmatic is finally becoming a strong tool for e-commerce brands and is not just limited to retargeting. However, in order to make it work, e-commerce brands need to grow their in-house expertise on how to appropriately manage programmatic campaigns. The combination of knowledge with customisation is necessary for programmatic success.