Optimizing for Success — Why Focusing on Impressions Is Losing You Money.

It is estimated that $1.2Bn is spent on digital marketing throughout the MENA region each year, the question is how much of this is wasted due to ineffective ad setup and optimization decisions?

“Half the money I spend on advertising is wasted; the trouble is I don’t know which half”

The advertising industry has moved on quite significantly since John Wanamaker’s time, and present day digital marketers are afforded access to significant volumes of data that facilitates the measurement and optimization of campaigns.

At inphota, we provide our partners with real-time conversion data to enable their media buying agencies to optimize campaigns towards commercial outcomes. However, despite making such data available we often see agencies optimizing their campaigns towards a single metric, impressions.

Whether it’s as a result of data paralysis or a culture of reporting grounded in traditional media measurement techniques, focusing solely on impressions has a negative impact on commercial outcomes.

To demonstrate, I’ve worked through a real world example to highlight the importance of defining campaign KPIs that impact business performance. In this example, a sole focus on impressions distracted from the ultimate commercial goal, selling tickets to an event.

It’s important to note that a single piece of video creative was promoted to three separate audiences on Facebook:

• Audience A: Lookalike audience derived from followers of the brands Facebook Page.
• Audience B: Audience of prospective customers derived from first party data collected through the inphota platform.
• Audience C: Interests based audience

The below results represent the reporting metrics provided by the agency managing the campaign.

Taking a look at this chart for a moment and ask yourself how would you define the most successful audience?

Did you choose Audience C?

The agency running this campaign did. Their justification, Audience C delivered a greater volume of impressions.

Having started with a budget split across the three audiences the agency made a single optimization during the live campaign. Identifying that audience C was delivering a greater number of impressions and allocating more budget to it.

At this point it is important to note that when setting up an engagement campaign on Facebook you’re charged a CPM rate, Cost Per Mille (Cost per Thousand Impressions). In layman’s terms, the more you spend, the more impressions you will receive.

Despite not having any information related to campaign set-up or optimization choices, knowing that the agency optimized towards Audience C because it generated more impressions tells me that the budget splits were not equal during set-up. This inherent bias in the ad set-up strategy led to the team allocating more money to the Interests based audience when establishing the initial campaign budget splits. As a result, Audience C was always going to ‘perform well’ as it was initially allocated a higher proportion of the budget. A small initial mistake that compounded throughout the entire campaign.

The agency had a real-time feed of completed purchases (Registrations) yet chose to prioritize impressions as the defining optimization metric. There are times when building brand awareness is valuable, however, in the context of a one off event, with a limited purchase window, 2hat use is brand awareness if the event has passed and there isn’t a product for a prospective customer to purchase?

Further analysis of the results provided make the commercial impact of such decisions clear.

The Cost Per Acquisition (CPA), calculated from the data provided, shows that Audience C delivered a CPA 33% higher than Audience B. By putting more budget towards audience C the agency was essentially paying an extra dh62 per completed registration.

Taking this one step further we calculated the Return On Ad Spend, ROAS. Each event registration generated the client dh199 meaning that both Audience A and Audience C generated a negative ROAS, essentially losing the client money. In the context of audience C, a ROAS of 0.8 equates to 80 fils in revenue for every 1 dirham spent.

Audience B was the only one to actually make the client money yet the agency did not deem it to be successful as they could not see past impressions as a defining metric. Context is important here, the agency had access to the registration figures, but did not identify the value that Audience B provided as they did not look beyond the default reporting data provided by the advertising platform.

At a glance, Audience C was the most successful Audience in terms of total number of impressions and registrations. The problem is that this audience generated those figures with a significantly greater budget outlay.

This places emphasis on the importance of results relative to spend. CPA and ROAS are simple to calculate and add a new dimension to the standard awareness based metrics.

• CPA = Total Amount Spent / The Total Number of Sales
• ROAS = Total Value of Sales Generated / Total Amount Spent

In real terms the campaign lost the client money, but we should also consider the opportunity costs. Consider a scenario where the agency chose to optimize towards purchase, how many registrations could the client have achieved? Taking the CTR and Conversion Rate for Audience B I’ve calculated the likely conversions in the event this audience received the same budget allocated to Audience C.

Calculating in this way accounts for economies of scale in media buying. By placing a larger budget behind Audience C the agency received a volume preferential rate on CPM for this ad set which equated to a difference of dh13 per one thousand impressions compared with Audience B. Using this data, rather than a division of CPA to create a more accurate budget adjusted estimate of how many registrations Audience B could have achieved had it been allocated the same budget.

Using this method of calculation we can say that Audience B would have likely generated 2–2.5x greater sales than Audience C if given the same budget.

There is a general consensus that a ROAS of 4:1 or higher indicates a successful campaign, this suggests scope for further optimization of the campaign to incrementally increase value to the client.

Beyond testing different creative, there are two simple changes that I’d have recommended making to this campaign to further increase ROAS.

1. Frequency Capping

Dividing the total number of impressions by the reach of the campaign gives you frequency, the average number of times each person saw the advert.

The Frequency for each Audience is as follows:

Having previously analysed our customer acquisition funnel we calculated that usually takes between 3 to 4 interactions with a customer to convert them to a sale. In this context it’s fair to say that someone who has seen an advert for an event 9 times and hasn’t converted is unlikely to become a customer.

Setting a frequency cap during campaign setup can help ensure that you show the advert to a greater number of potential customers as opposed to showing the advert to the same people multiple times. The benefits of doing this will vary depending on the product, take an FMCG brand for example, for a product with little differentiation in a competitive marketplace a higher frequency could keep a brand front of mind and trigger impulse purchases in store. However, in this context there was a need to generate immediate sales as there was limited value in brand recall once the event had passed.

2. Establish Exclusions

When setting up a campaign I believe it is as important to define who you do not wish to target as it is to define those you do.

It is often overlooked, but setting up a custom audience that enables you to exclude customers who have already purchased will help you deliver greater returns. At inphota we’ve established a data connection that passes this information straight through to the ad platforms. However, lack of technical support need not hold you back as you could create and update an audience manually, it’s more labour intensive but the value it generates is significant and worth the effort if more technical solutions are not viable.

Conclusion

This campaign generated dh66,000 from an advertising spend of dh76,000, a dh10,000 loss before any agency fees are accounted for. A stark example of the dangers of focusing on impressions as a defining metric.

This is a particularly prevalent example, however not the only one I’ve encountered when auditing campaigns. In a region constantly striving to be the biggest and best, there’s a temptation to optimize towards the largest possible number. However, as I’ve shown in this example, impressions do not always correlate with sales. It’s time that we toss the vanity metrics aside and instead focus on generating the largest commercial returns for our clients.