Chief Marketing Officer United States

Decisions that made our advertising smaller, slower or dearer, and why

Total assets of Ledgemoor Financial Corporation: US$19.8bn; June 2026; 128 branches in three states
Annual marketing budget: US$38m; 2025, including US$21m of paid advertising
Metro mortgage applications from majority-Black and majority-Hispanic tracts: 10.7%; Milwaukee, 2025; 4.9% in 2023

Decisions & consequences

Pulling the centenary campaign and publishing deposit facts in March 2023

On Monday 13 March 2023, three days after the first of two large bank failures that month, Ledgemoor's centenary campaign was still running under the line A hundred years of steady. At the end of 2022, 38% of our US$15.8 billion of deposits sat above the insurance limit, much of it held by manufacturers, farm cooperatives and private-banking clients.

Our chief executive, Desmond Ulverstone, and our chief financial officer, Harriet Lowenthal, wanted newspaper advertisements telling customers their money was safe. Our head of consumer banking, Garrett Graumann, proposed a 5.25% seven-month certificate of deposit for new money. I argued against both. We could not support a safety claim for balances above the limit, and the FDIC's rules against misrepresenting deposit insurance made the wording a legal question as well as one of tone, while a rate special would announce that we needed money. I proposed instead that we stop all paid advertising, publish the facts depositors were asking for and put our people on the telephone.

The executive committee agreed that evening, on our general counsel's condition that every figure match what the company had disclosed or would furnish to the market that day. By Tuesday we had published our insured share, the number of accounts above the limit, worked examples of coverage by ownership category and how a reciprocal deposit network spreads large balances across insured banks. By Wednesday 1,400 branch and contact-centre colleagues had a one-page script, and within five business days relationship managers had called 2,600 clients with balances above the limit.

Deposits fell by US$1.3 billion, 8.2%, over ten business days, 71% of it from balances above the limit, and then stopped. Reciprocal network balances rose from US$0.4 billion to US$2.3 billion by June, and total deposits regained their December 2022 level by March 2024. We gave up the new money a rate special might have drawn, which our treasurer put at US$250 million to US$350 million.

The facts page is now refreshed every quarter and its release rehearsed every year. What I would do differently is the order of contact. About 4,100 small-business clients holding between US$250,000 and US$1 million received only a letter, on the eighth day, and several told us afterwards that the silence had felt like an answer.

Rebuilding mortgage advertising that had stopped reaching parts of Milwaukee

In March 2024 our fair-lending officer, Nadia Ellstrom, brought me her annual comparison of our mortgage applications with those of peer lenders, built from public home-mortgage disclosure data. In the Milwaukee metro, 4.9% of our 2023 applications came from majority-Black and majority-Hispanic census tracts, against 12.8% for lenders of similar size. Underwriting and pricing reviews showed no unexplained differences. The gap sat at the top of the funnel, which was my responsibility.

Only 3.1% of our digital mortgage impressions in the metro had reached those tracts. Our agency's programmatic buying set bids by ZIP code according to past cost per application, and a modelled audience bought from a data vendor had been built from our existing borrowers. Nobody had excluded a neighbourhood; the system had learnt where our customers already lived and spent our budget there. Redlining enforcement has treated marketing patterns as evidence, and an absence of intent is not an answer a supervisor would accept.

I proposed removing ZIP-level optimisation from all credit advertising, dropping the vendor audience, setting delivery floors for each tract group in proportion to owner-occupied homes and adding Spanish-language advertising. Our head of mortgage lending, Lyle Renholm, objected that cost per application would climb on a flat budget, and our agency forecast fewer applications. Both were right. Our chief executive kept the US$4.6 million digital mortgage budget flat, so we accepted fewer applications overall, and we funded referral partnerships with three housing counselling agencies.

In the second half of 2024 cost per application rose from US$290 to US$430 and metro applications fell 7%. By the end of 2025 the share of applications from those tracts had reached 10.7%, cost per application had settled at US$360 as the creative improved, and we had added four loan officers in two branches. Approval rates there remain lower, mainly for credit reasons, and we track that with our credit team.

Every credit campaign now carries a delivery plan co-signed by fair lending, and the board risk committee sees tract-level delivery twice a year. I would do two things differently. I should have asked for delivery data in 2021, before disclosure data forced the question. And in the first months some applicants waited nine days for a loan officer's call, which undid part of what the advertising had achieved.

Ending per-account payments to finance creators

By mid-2020 Quillmark Investing paid 54 finance creators US$40 for every funded account opened through their referral links, and they supplied 29% of our new accounts. I had built the programme in 2019 and approved most of the creators. Compliance reviewed a sample of their posts after publication. In June 2020 one creator, with nearly a million followers, described selling options as a way to collect free money every week. By the time our compliance team saw the video it had 1.4 million views, and over the following month we received 61 complaints, several from customers who had lost money trying the strategy.

The stakes were regulatory and commercial. Content a broker-dealer pays for can become its own retail communication, which FINRA's rules expect a registered principal to approve and the firm to retain. Retail investing was growing quickly, and creators were our cheapest channel.

I proposed ending per-account payments entirely: flat monthly fees of US$2,000 to US$12,000 according to output, pre-approval and archiving of every paid post, a ban on discussing trading strategies, options or past returns, and disclosure of the paid relationship in each post's first line. Our chief executive, Anders Lindahl, and our head of growth opposed it, and an investor director argued that we were abandoning our best channel in our best year. Our chief compliance officer, Marisol Treadgold, and I presented the proposal to the board together in August, and it was accepted.

We gave up volume. Thirty-eight creators declined the terms. Creator-sourced accounts fell 62% in the fourth quarter, blended acquisition cost rose from US$88 to US$131 per funded account, and we missed the 2020 account plan by about 140,000 accounts. Twelve-month retention of creator-sourced accounts rose from 48% to 71%, and complaints mentioning social content fell from about forty a month to six. When FINRA asked firms about their influencer arrangements in 2021, we produced every paid post, its approval and its payment within days.

Paid partner content at Quillmark has been reviewed as advertising ever since. What I would change is the transition. We gave creators three weeks' notice and no transitional fee, and several careful educators with small audiences left along with the rest. I also left our customer referral bonus, which raised the same question, until 2021.

Advertising the mortgage rate our median borrower received

In 2013 Strathmere Home Lending's advertising headlined a 30-year fixed rate as low as 3.99%. Regulation Z required the annual percentage rate beside it, and we showed it. But the rate assumed a credit score above 760, a 25% down payment and 1.5 discount points, and only 11% of our 2013 borrowers obtained it or better. We logged 340 complaints that year about advertised rates, and loan officers spent the opening minutes of most first calls explaining why the caller's rate would be higher.

The stakes were reputational and commercial. Rate-comparison sites ranked lenders by headline rate, and our purchased leads came largely from those rankings. Raising our advertised rate meant falling down the page.

In February 2014 I proposed headlining the rate our median borrower had received the previous month without discount points, recalculated monthly and signed by our secondary-marketing desk, with the annual percentage rate equally prominent and a table showing rates by credit-score band and down payment. Our head of retail sales, Stellan Odegaard, opposed it, arguing that leads would collapse and loan officers' pay with them. Our chief executive, Lucinda Varrick, agreed to a one-quarter test in Minnesota and Colorado against our other states.

Leads in the test states fell 23%. Lead-to-application conversion rose from 13% to 20%, so applications fell only 4%, and funded loans were flat. We adopted the approach nationally in July 2014. Complaints about advertised rates fell to 80 in 2015, and purchased leads fell from 46% of applications to 21% by 2016 as referrals grew. We gave up our prominence on the comparison sites, and some loan officers left for lenders whose advertising produced more calls.

Pricing and marketing have shared one advertised-rate rule since, and the desk that sets prices signs the figure we publish. What I would do differently is consistency and consultation. Our feeds to comparison sites carried the old lowest rate for eight months after the test, so we were telling two stories at once. And I designed the rate table without loan officers, who found it hard to explain; their version, introduced in 2015, was clearer.

Discuss a leadership challenge