Home Online Advertising Meta Has Lost The Pulse Of Its Customer Base As Automation Replaces Human Services

Meta Has Lost The Pulse Of Its Customer Base As Automation Replaces Human Services

SHARE:

In 1989, when Toyota first launched the Lexus car line in the United States, it saw a trickle of early customers report faults with the cruise control function.

Ruh-roh.

Rather than wait for that trickle to become a flood, Toyota spent heavily to self-recall 8,000 cars and added thoughtful touches upon return. When owners got their cars back, their vehicles had been washed and the gas tanks were full. In the book “The Tipping Point,” Malcolm Gladwell argues that this act of largesse, when Toyota could have waited for a mandated recall and then taken a parsimonious approach, helped shape the premium Lexus brand perception.

Why mention this? Because, on Sunday, a failure of Meta’s ad management system caused advertiser accounts to grossly overspend across Facebook and its Audience Network to the tune of tens – if not hundreds – of millions of dollars.

Meta is still considering whether and how to issue refunds. Eight agency and brand advertisers and two Meta employees told AdExchanger they expect only partial refunds, which will come after weeks – perhaps more than a month – of campaign assessment.

Meta has acknowledged through sales reps that the issue stemmed, at least in part, from the web-to-app optimization product, according to four advertisers who spoke to AdExchanger. While refunds are also being confirmed in sales channels, there has been no official confirmation for advertisers to expect to have their money returned.

Meta’s bug

It should be noted that the money Meta allocated to itself in error was done both in flagrant violation of cost caps and spent during the very early hours of the morning – impressions largely wasted on the Sunday graveyard shift.

Meta didn’t serve 100 times its usual ads in that period – it primarily overcharged for ads. Someone turns on a flashlight app at 4 a.m., and the Meta Audience Network was – no joke – charging more for that impression than it would cost to do a whole homepage takeover on ESPN.

Although AdExchanger has learned one of the root causes of the glitch, related to a malfunction of app-to-web campaign conversion optimization, the company has not published a blog post addressing the issue and has not flagged it within Ad Manager.

Advertisers who missed the drama over the weekend and then logged into their Meta account on Monday, Tuesday or Wednesday would have no indication that the error occurred. Even advertisers with affected accounts still have not been notified directly.

Subscribe

AdExchanger Daily

Get our editors’ roundup delivered to your inbox every weekday.

As of Wednesday, the only official response Meta has so far issued about this error is this terse statement: “A technical issue that has been resolved caused ad delivery issues for some advertisers.”

Is anyone home?

Meta’s account service issues go deeper than the screaming silence of the company’s crisis response this week.

An ecommerce buyer told AdExchanger that although she was able to reach three Meta reps over the past couple of days, two of them “actually seemed to be hearing about the issue from me for the first time.”

And an ecommerce brand CMO, who spent $50 million on Meta ads last year, said he wasn’t able to wrangle a response from a human rep until Tuesday afternoon. The rep assured him that no campaigns had overspent by more than 25% the daily cost cap.

The CMO said the 25% overcharge was more than he would feel comfortable spending on a Sunday. Having hit the daily budget cap at all puts the marketing department in a tough bind, he said, with that money gone and absolutely no returns for it.

The response from Meta’s comms teams to press inquiries has been a similar mix of bewilderment and a sense that this issue is, well, a non-issue.

For example, Meta disputed a line from AdExchanger’s report on Monday, which said it had “yet to weigh in with advertisers” by the start of the business week.

Spokespeople contended that MetaStatus.com, a website showing whether Meta’s business tools are working properly or not, had toggled the “Ads Delivery” product from “Major disruptions” to “Resolved” once Meta thought it was safe for advertisers to start spending again. The company also reached out to some advertisers through sales channels.

But it’s possible, even likely, that thousands of advertisers who lost money still have no idea the issue happened or that they should be pressing for a refund.

The issue blew up on Sunday on Twitter, a hub of ecommerce advertising debate. But a dental practice or the owner of that specialty cheese store on Main Street have Meta accounts and daily cost caps, too, and simply trust Meta to operate effectively in the background.

Thanks for nothing

SMB advertisers are usually happy to take Meta’s advice.

When, for example, Meta suggests that embracing a new best practice, like the app-to-web campaign optimization behind Sunday’s blunder, many do so without asking questions.

That type of advertiser – millions of which use Meta’s tools – aren’t scouring the ads dashboard on a daily or even weekly basis. They also don’t have data-driven marketers on staff and probably wouldn’t even be able to tell from looking at the reports that Meta’s system had unjustifiably blown through their budget.

In recent months, as Meta advocated for advertisers to adopt its app-to-web optimization product, the company sent emails to account operators and promoted the tool directly within its platform. Business owners or marketers therefore couldn’t help but see the option to opt in when they logged into the system, according to three ad buyers.

There’s a stark contrast between Meta’s highly visible promotion of a new product and its non-attempt to inform advertisers of its recent error in overspending their daily budget. The outcome here is extra frustrating since the error came as a direct result of advertisers having adopted this very product.

Seriously, is anyone home?

The Meta app-to-web optimization tool that misfired on Sunday is part of a suite of machine learning products that require advertisers to give Meta more direct control over campaign targeting, attribution and bidding decisions.

But app-to-web optimization is a symptom of a problem rather than the disease.

Meta let go of tens of thousands of employees over the past half year. To replace those functions, the company has turned to machine learning-based or AI solutions that, as the overspending episode demonstrates, are simply not ready for primetime.

During a previous Meta platform bug outbreak in November, an agency CEO told AdExchanger that his Meta account rep had changed three times in three weeks following layoffs. This week, his agency’s SMB accounts lost thousands of dollars they cannot afford to lose, and now there’s nobody left to reach out to for help – other than the automated response system.

He’s had to get creative in his attempts to escalate the issue.

“You try people you know [who work at Meta] or somebody who someone else knows and gave you their email, and you hope they can help,” he said.

Even if advertisers are eventually made whole – don’t hold your breath – their losses in the meantime are like an interest-free loan to Meta for weeks or even months.

Two agency advertisers and one brand operator, who have each spent hundreds of millions of dollars on Meta ads over the past few years, asked reps what to expect in terms of refunds. All were directed to visit Meta’s generic “About refunds” support page.

Per the support page, refunds come in the form of ad credits “where possible.” It would no doubt be galling for advertisers to first have their budgets wildly misspent and then, rather than getting their money back, be issued with ad credits to spend more money on Meta.

If Meta’s response so far is any indication, the company may not be aware that dishing out ad credits instead of cash refunds to deal with this issue would come across as a slap in the face rather than as a mediation concession.

Not to mention that there’s no one to talk to about it.

“I could deal with the losses and refunds as they come and the fact that the AI optimization tech will go haywire again someday,” said one DTC brand marketer. “But the service chatbot responses, I cannot bear.”

Must Read

LG Electronics

Alphonso Shareholders Win Their Suit Against LG Electronics Over Corporate Board Drama

After being summarily booted from the board of LG Ads in late 2022, Alphonso’s founding team has won its lawsuit against LG Electronics.

Bye-Bye Sizmek! Amazon Advances Flashtalking And Smartly As Alternatives In Advance Of The Shutdown

According to emails seen by AdExchanger that were sent to Amazon customers this week, Amazon is officially naming integration partners to offload clients of the Sizmek ad suite, now the Amazon Ad Server.

2024 Promises More Premium Inventory – And Bigger Budgets – For In-Game Ads

Given the deprecation of third-party cookies and the reemergence of contextual targeting, 2024 could be a big year for in-game ads – so long as game publishers position themselves as a source of premium inventory.

Privacy! Commerce! Connected TV! Read all about it. Subscribe to AdExchanger Newsletters

AdExchanger’s Top 3 Connected TV Newsletter Issues Of 2023

This was such a busy year in CTV land that we had to launch a dedicated newsletter just to keep up with all the trends, from measurement, currency, targeting and attribution to streaming data, identity, supply-path optimization and new ad formats – just to name a few.

M&A 2023: Ad Tech Deals Were Muted, But That Could Be A Mark Of Maturity

Who got bought in 2023, and who did the buying? Here’s a non-exhaustive list of some of the most notable ad tech M&A activity from this past year (with a few media and agency deals tossed in for good measure).

Comic: The Great Data Lakes

Snowflake Acquires Data Clean Room Startup Samooha

Snowflake has acquired Samooha, a startup that develops software to make clean room technology accessible to marketers who aren’t necessarily SQL wizards or data scientists.