Facebook claims its advertisers could potentially reach hundreds of thousands of New Zealand millennials not reflected in local population numbers.
Data comparing official population numbers to metrics accessed through Facebook's ads manager tool show the social media company has heavily over-inflated estimates in certain age categories.
The most staggering divide was evident in the 20-24 age category, for which Facebook's numbers come in at more than 150,000 above official population estimates.
While the Stats NZ forecast for 2021 shows only 347,800 people in this age group in New Zealand, Facebook claims users can reach 500,000 on its platform – equating to an over-inflation of around 44 per cent.
The data was also off significantly in the 25-29 age cohort, where Facebook claims a reach figure of 470,000 on the platform as opposed to the 390,050 humans there actually are. This totals an extra 79,950 people or an increase of 20.5 per cent.
The disparities do become smaller as the age groups become older, but Facebook was still off by 43,420 in the 30-34 age cohort.
A member of the Facebook communications team defended the large differences and referred the Herald to a policy document explaining that reach estimates aren't designed to match census data and may be influenced by the number of accounts used per person, the number of temporary visitors in a particular geographic location at a given time and Facebook user-reported demographics.
The argument here is that reach shouldn't be understood in the same context as people in an official national census.
This is not the first time Facebook has been criticised due to its potential reach figures. As far back as 2017, respected trade publication AdNews reported on major differences in Facebook's potential reach metrics across 12 countries. The total accumulated difference in that study amounted to 42 million 20- to 29-year-olds.
The New Zealand data revealed today was pulled and cross-referenced by Tim Dorrian, the managing director of Wellington-based digital marketing company Aro Digital, who says the framing of reach creates the impression that Facebook is referring to people.
"There's a big difference between accounts and people," Dorrian tells the Herald, adding that the disparities could be even larger given that not every person in those age groups would have a Facebook account.
"The inflation is coming from a number of places," he says.
"It's potentially bot accounts, people with secondary accounts or maybe just accounts that have been set to a different age. So, someone might be 65, but they've set their birthday to 1992 and they could then show up in the 25-29-year-old range."
Dorrian says it's imperative that the numbers that appear in Facebook's tools are correct, given how many advertisers in the country are influenced by them.
Facebook does not charge users based on its reach metrics (its charges are based on impressions and clicks on ads), but Dorrian says these figures influence advertisers' views on where they should be spending their money.
"People are making media buying decisions based on those numbers. They're making decisions on what platforms they actually want to advertise on."
Data always carries a margin of error, but Dorrian says the extent of the differences in this instance is noteworthy.
"If it were a couple of thousand above it probably would not be the end of the world, but an inflation of 44 per cent in the 20- to 24-year-old age range is a huge discrepancy. And that probably needs to be addressed."
In previous years, the differences could have been chalked off due to the presence of travellers in New Zealand, but the global pandemic has seen international tourism cut off from New Zealand.
Dorrian says that anyone using online platforms needs to be cognisant that their advertising could sometimes be served to individuals or devices that either aren't real or aren't in your ideal target audience.
This has long been considered an international problem, but the increasing use of global platforms has opened local businesses to the risk of spending money on ads that aren't seen by the intended audience.
For this reason, Dorrian advises marketers to avoid vanity metrics – such as likes or followers – and focus instead on the business impact of an online campaign.
"You should be trying to achieve some type of commercial objective, whether that's a purchase on an e-commerce website or getting someone to call you. If you do a campaign that's focused on these types of objectives, you can then calculate how much it costs you for each acquisition and work out if it's worth the investment."
He argues that this approach can better protect the business from metrics difficult to verify independently, while also ensuring the best value is extracted from online platforms.
Revelations about Facebook's metrics in New Zealand come amid the backdrop of intense international media scrutiny as the tech giant fights a 2018 class-action lawsuit initiated by small business owners alleging the social media giant has for years inflated its estimates of the number of people an ad could reach.
Internal emails revealed during the litigation show that a Facebook product manager's efforts to have the metrics definition altered were rebuffed by Facebook executives, according to a report in the Financial Times last week.
The lawsuit, filed in northern California, alleges that Facebook knowingly included fake and duplicate accounts in its potential reach metric, misleading unwitting advertisers.
Facebook has argued that the metrics are only estimates and advertisers do not pay the company on potential reach but rather for actual impressions and clicks on ads.
The trial is ongoing in the United States.