Transform 2021

Live now: AI/ML Automation Technology Summit

July 12-16

Advertising fraud on mobile applications exploded 50% in the last year, according to a new report from media measurement and data company DoubleVerify. The figures are one of several eye-opening statistics about the current state of media quality and performance on the Web.

Above: The charts show how each device is impacted differently by the various types of fraud/SIVT (sophisticated invalid traffic).

Image Credit: DoubleVerify

Mobile advertising is a hot sector, with Emarketer predicting mobile ad spending will exceed $156 billion globally by 2023. Mobile consumption worldwide has soared, partly due to people staying home during the coronavirus pandemic. Even with a growing vaccinated population and easing of lockdown rules, people are still staying indoors and streaming videos, playing games, and shopping online using mobile apps. As a result, advertisers are pouring money into mobile advertising, and fraudsters are moving in with their fake clicks and traffic to siphon away some of those ad dollars.

Current Time 0:00

Transform 2021 - VentureBeat

While global fraud volume overall remained steady (post-bid fraud/sophisticated invalid traffic [SIVT] rates nudged down from 2% to 1.4% and overall volume didn’t change year-over-year), pockets of fraud continue to plague the industry, DoubleVerify found in its annual Global Insights Report. One example is scams targeting video on mobile apps. Streaming advertising fraud has become a major problem, with DoubleVerify detecting over 500,000 fake connected TV (CTV) devices — gaming devices and streaming consoles — a day.

Fraudsters are also shifting tactics. Last year, bot fraud made up 78% of fraud/SIVT violations on CTV. While bot fraud violations on CTV remain the most common type of CTV fraud, datacenter traffic is a growing problem. Datacenter traffic is a type of SIVT that targets server-side ad insertion (SSAI) inventory, which is especially common on CTV and audio, DoubleVerify said. With datacenter traffic, fraudsters spoof SSAI inventory, such as a mobile impression made to look like a CTV impression, or have SSAI inventory with incomplete information.

DoubleVerify’s global report analyzed over one trillion ad impressions in 80 markets across more than 2,100 brands. Video and display impressions were measured year-over-year from May 2020 to April 2021 across desktop and mobile web, mobile app, and internet-connected TVs.

Read the full 2021 Global Insights Report from DoubleVerify.

Transform 2021

Live now: AI/ML Automation Technology Summit

July 12-16

Presented by Abe.ai

For those of us that are regulars at fintech and artificial intelligence (AI) conferences and follow the development and innovation around AI, virtual financial assistants (VFAs) have gotten a lot of attention at these events for a few years now. But lately it seems that VFAs might be becoming old news — conferences have started moving onto other topics after focusing on conversational AI for several years in a row.

So the questions many people have are: “What’s taking so long? Where are these VFAs? Why doesn’t my bank have one yet?”

Virtual financial assistants take time to build

The hype around VFAs and their benefits the last few years may seem new but it is not, it just got louder. We have seen plenty of primitive iterations of chatbots over the last decade promising similar benefits, but in 2018, Bank of America actually delivered.

They came out with what some called the first “full-featured” VFA, “Erica,” delivering personalized, timely advice and assistance. “Erica” has enjoyed tremendous success (55% adoption as of February 2021), but they were very much on the bleeding edge of this technology when it comes to financial institutions (FIs). They invested multiple years, a large team and a chunk of their technology budget — something not many FIs have been willing to do.

A chatbot in 2014 felt stale and rigid using decision trees to guide you, while the advanced natural language processing/understanding technology of today that we and some others like Bank of America use to build a VFA, is significantly more complex. It requires a team of expert data scientists and several years of development.

While FIs execute their VFA project, it will take time, and how much time depends on the path they choose — build versus buy or a combination of both.

Where the financial institution focuses matters

While many conversational banking projects with FIs often started with every VFA functionality on the table (some combination of customer service, transactions, product offerings, financial insights, etc.), things changed over the past 18 months.

As FI’s budgets were strained, conversations with technology decision-makers quickly focused on cost-cutting features. So while full-featured VFAs may be on the roadmaps of most mid- to large-tier FIs, many are deploying an initial phase with service functionality first, and that is a strategy we have seen work well and recommend.

While a VFA focused solely on assisting the service center is not glamorous, it has been vital in reducing service center costs at many FIs at a time when many were seeing peak demand during the pandemic.

You may not see the VFA pop up in your app with helpful advice on what or when to do it just yet, but many FIs do have the VFAs in the help section of the app, on IVR, or even helping the agent you are talking with get the answer you need.

While FIs execute their VFA projects, branching out into the other functions beyond assisting the service center will take time — how much depends on when technology budgets loosen with a broader focus on innovation again.

The VFA is the mouthpiece, but you need the data

While the perceived delay in next generation VFAs can be chalked up to development time of advanced technologies and the high prioritization of cost-cutting service functionality, there is another factor keeping the hype of full-featured VFAs from realizing full potential.

The VFA may be the perfect delivery method for service, advice, and the nudge towards financial wellness, but without accurate data and timely insights to trigger those conversations, the experience is significantly less powerful. Delivering a personalized conversation to a customer with general information, even if it contains their authenticated data, is not enough.

What makes a personalized conversation with a VFA engaging is the feeling that your FI knows you, your habits, and what you need. Then having the tools to not only engage through a personalized conversation, but also take action directly in the VFA session as if you were speaking to a personal banker.

In order to do this, an FI needs to have their data house in order and work with vendors who can help enrich the data and deliver valuable insights. Once this need for the right data/insights becomes clear to FIs looking to deliver a full-featured VFA, this data work sometimes gets prioritized, extensively delaying the VFA implementation. While FIs execute their VFA projects, they will need to do significant work on their data and insights capability in order to deliver a real full-featured VFA, and that will take time.

So when asking what’s taking so long for you to get a full-featured VFA at your FI, it’s likely that your bank did build or has been building for some time now. We often see FIs come to us after spending two or three years building on their own, looking to partner now to supplement that work or even start over. So, while it still might be some time before you see a full-featured VFA, the good news is more FIs are recognizing the importance of partnering with a conversational AI provider and yep, you guessed it, that will take time.

Download The Next Generation of Conversational Banking to learn how financial institutions can leverage conversational artificial intelligence to go beyond simple reactive use cases and instead generate proactive interactions that engage customers on meaningful money matters and support financial wellness.

Juan Romera is Product Evangelist at Abe.ai (an Envestnet | Yodlee solution).

Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. Content produced by our editorial team is never influenced by advertisers or sponsors in any way. For more information, contact [email protected].