The penalties for being wrong or lazy get steeper every day
I can’t tell you how many advisors have written the following: “Studies show
60% of Americans don’t have an estate plan.” That’s a terrible state of
affairs in the U.S. But where did you get that number? What study? Did you do your
own survey? Did you read it somewhere? Did someone mention it at a conference
you attended? If you can’t back up a fact or stat, then find the original
source or don’t mention it.
When you fudge it, you’re not only damaging your credibility,
but you could be facing serious legal, regulatory, and professional
consequences. For SEC-registered investment advisers, public writings, blog
posts, and newsletters count as marketing materials. Rule
206(4)-1) strictly bans untrue statements, unsubstantiated
material facts, and misleading inferences. Even if you’re not registered with
the SEC, you owe it to your clients, prospects, strategic partners and
followers to substantiate the facts and claims you make in your public-facing
content.
Whether your material is being reviewed by your compliance
department or an outside editor where you’re trying to get published, these
eagle-eyed reviewers are increasingly under pressure to verify facts and
matters of attribution before approving the material. Increasingly they’re using
AI detection software to review all submissions that come across their desks. If
you get flagged, you might not get another chance with them.
Real-world examples
Here are some of the most common errors we see when it comes
to misattribution:
Citing a prestigious outlet (that wasn’t the source). When it comes to
national divorce rates (67% of second marriages and 73% of third marriages), we’ve
seen a number or advisors and estate planners cite this Psychology Today
article (5
Risks for People Marrying a Second or Third Time) as the source. Those
alarming stats are right in the first paragraph of the aforementioned article, but
Psychology Today didn’t say where it got the data. The article just
happens to appear near the top of Google search results for “divorce rates 2nd
and 3rd marriages.” Don’t be lazy. Be accurate.
This USA
Today article is also widely cited as the source of those sobering divorce
statistics on second and third marriages. But USA Today didn’t do the
research, either. To its credit, USA Today linked out to an external
source, but it was to the general divorce.com site which has dozens upon
dozens or random statistics on divorce. If you look closely, divorce.com tells you:
“A prevalent myth about divorce rates in the
U.S. suggests that 50% of first marriages, 67% of second marriages, and 73% of
third marriages end in divorce. However, these statistics are not based on any
verifiable data, and their origins are untraceable, as highlighted in Shaunti
Feldhahn's book The Good News About Marriage.”
Divorce.com continues: “This
misleading information has been widely quoted, including in reputable sources
like Psychology Today and Forbes. It is important to acknowledge that these
figures are inaccurate and should not be perpetuated.”
Ouch!
Fishing expedition. Another way we’ve seen advisors
play fast and loose with sources is just taking the reader to the home page of a
massive general government website, hoping the reader gets too discouraged to
keep digging. For instance, one client wrote: “the birth rate for U.S. women
aged 40–44 rose to 12.8 births per 1,000 women (up 1% from 12.7 in 2024).”
That data is from Centers for Disease Control and Prevention (CDC) and the U.S.
Census Bureau. But just taking readers to cdc.gov or census.gov
could take hours to verify. Instead, we instructed our client to include the correct
link to: https://www.cdc.gov/nchs/products/databriefs/db571.htm.
I know you’re laughing, but we had another wealth advisor
client who would answer every question we asked about sources by having his
assistance send us half a dozen PDF reports, each 100-plus pages in length and
say: “It’s from one of these reports.” Yeah, thanks for nothing!
End run. As our client Randy Fox wrote in Wealth
Management Magazine recently, “Roughly 6 million small and medium-sized
businesses are expected to change hands over the next decade.”
Randy’s article and the 6 million statistic got picked up by a number of blogs
published by financial advisors and estate planning firms. Many did not
attribute that source to anyone, or they incorrectly cited Randy or Wealth
Management magazine as the source of the 6 million stats. Instead, as correctly
stated in Randy’s article, The McKinsey Institute for Economic Mobility Study was
the correct source of the 6 million number.
End run #2: Here’s a paragraph from a book manuscript
a prospect sent to us for review:
“Only 16% of people write down their goals. Those who
write down the goals are 42% more likely to achieve them.”
The stats he cited sounded very reasonable. But when we
asked where he got them, he hesitated before admitted: “I forget. I read it
somewhere.” That’s not good enough. Then he admitted he used AI to help him
draft the manuscript “and it always gives me sources,” he assured us. So, when
we asked to see the sources AI came up with, he gave us a link to this blog
post on LinkedIn. Yes, the post mentioned the 42% statistic, but it
didn’t cite where it came from.
Two wrongs don’t make a right.
A week later, our client sheepishly came back to us and said
the missing source was CNBC because he heard it on a news report they did. To their
credit, CNBC correctly
cited Dr. Gail Matthews, a psychology professor at Dominican University of
California, as the source of the 42% statistic. But alas the CNBC link was
broken, and it pointed to a general bio page for Dr. Matthews, not to the actual
research containing the 42% statistic. As mentioned above, a news report in
well-respected mainstream media is typically not the source; it’s just a
conduit.
Don’t cite sources from AI without checking
Academic research conducted by Middle
Tennessee State University, Arizona
State University and Duke
University found that AI tools sometimes cite articles by an author who usually
writes about your topic. AI might even identify a journal that published on
your topic, but the title, pages numbers, and dates are completely fictional,
according to the aforementioned academic studies. AI’s web search capabilities
have improved since those studies were done, but they’re still not reliable for
identifying actual sources, as the Duke study concluded. “Depending on the
topic and availability of data it has on that topic, it may summarize the wrong
source or provide inaccurate summaries of specific articles, sometimes making
up details and conclusions,” Duke researchers added.
Conclusion
When it comes to sourcing matters of fact and attribution,
take the time to get it right or ping me any
time if you need help. Your reputation is too important to play fast and loose
with the facts.
What are you and your colleagues doing to improve your fact
checking and research capabilities? I’d love to
hear from you and why.
#businesscommunication, #practicemanagement,
#factcheck, #AIhallucination
Posts
