Skip to content

Case study · AmpliFi™ GEO

From #40 to #2 in AI answers.

How BrightStar Credit Union went from nearly invisible to second in AI answers about banking near its new Miami Beach branch, among 93 competing institutions in its market, in six months.

Rank in AI answers about banking near the new branch

Among 93 competing institutions in the market, including Chase, Bank of America, Truist, and Wells Fargo.

1 · StrategyFind the questions it was missing
2 · IntelligenceMeasure every release
3 · MarketingContent and website fixes
Client
BrightStar Credit Union
Solution
AmpliFi™ GEO
Market
Its new Miami Beach branch
Timeframe
April to October 2026

At a glance

Six months, 38 places.

BrightStar now trails only the market's longtime local leader, another credit union, and the gap between them is now 3.7 points.

Rank among 93 competing institutionsMarch: #40#2

October 2026

Share of AI answers naming BrightStarMarch: 2.38%31.28%

October 2026

Gap to the #1 institutionMarch: 59.5 points3.7 points

October 2026

National banks ranked below itMarch: noneChase, Bank of America, Truist, and Wells Fargo

October 2026

The challenge

A new branch AI barely knew existed.

BrightStar Credit Union had just opened its new branch in Miami Beach. In March 2026, when the baseline was established, ChatGPT named it in 2.38% of answers to local banking questions, good for #40 out of 93 institutions.

A crowded market
  • The longtime local leader, another credit union, appeared in 61.9% of answers
  • Bank of America and Chase each appeared in more than 40%
  • The leaders were entrenched, and the new branch had no track record
Strength that didn't travel

BrightStar already ranked #1 in AI answers across its established South Florida markets. But AI models answer location by location. Without content tying its products to the new market, the models had nothing to cite.

The priority

BrightStar's marketing leadership made the new branch the top priority: build visibility in the market and close the gap on the leader.

Our approach

Every piece of content, a measured experiment.

Not a hope. Each release had a locked baseline before it went live and its own measurement window after.

  1. Track the questions people actually ask

    Local banking questions ran through ChatGPT on a recurring schedule: the best auto loan, credit union or bank, where to keep savings. The set grew to 29 questions covering products, membership, and branch intent.

  2. Find where the credit union is missing

    It did well on general credit union and branch questions, but was nearly invisible on savings, CDs, auto loans, mortgages, and home equity lines of credit (HELOCs).

  3. Write from the live website

    Every article was grounded in facts on BrightStar's own website. Rates link to its live rates page or appear as plain text confirmed current before publishing.

  4. Measure each release

    Each content drop became its own experiment with a baseline, a live date, and a 30-day window. A comparison market served as a control for the branch page test.

What we did

Four waves of work.

A first round of product articles, a fix to the new branch's web page, then two monthly article releases aimed at the product gaps.

Wave 1 · April

Four product articles

Local guides for credit cards, auto loans, HELOCs, and CDs. Three of four lifted visibility on their target questions, and rank jumped from #40 in March to #19 in May.

The CD article was the exception: visibility fell from 4.55% to 0%. The credit union's CD rate pages load by script, so crawlers and AI models can't read them. That also explained why it was missing from third-party rate comparison sites.

Wave 2 · Late August

Branch page FAQs and a code fix

We added FAQs to the new branch's web page. Our go-live audit found the FAQ's structured data was broken: a line of code displayed as raw text, so search engines and AI models couldn't read it as an FAQ.

Once the web team fixed it, visibility on the 21 tracked questions rose from 25.17% to 28.39%. We also caught a sitewide template bug that cut article summaries off at a fixed length.

Wave 3 · September

Savings, home equity, and credit union vs. bank

Three articles answered the questions where competitors were winning. Visibility on 29 questions rose from 20.20% to 22.23%, and rank climbed from #15 in August to #6 in September.

Wave 4 · October

Five articles in one release

Membership eligibility, auto loans vs. dealer financing, condo and jumbo mortgages, debt consolidation, and the branch visit experience. The mortgage piece focused on condos and jumbo loans instead of first-time buyers, given local home prices.

Visibility across the 19 target questions rose from 15.50% to 22.11%.

Visibility is the share of AI answers to each article's target questions that named the credit union, in the 30 days before and after release.
ArticleReleaseBeforeAfterChange
Debt consolidationOct13.51%33.33%+19.8 pts
Credit cardsApr4.55%20.04%+15.5 pts
Condo and jumbo mortgagesOct0.00%12.00%+12.0 pts
HELOCs and home financingApr0.00%6.24%+6.2 pts
Auto loansApr0.00%5.07%+5.1 pts
Auto loan vs. dealer financingOct3.38%5.00%+1.6 pts
Membership eligibilityOct25.00%25.00%No change
Branch visit experienceOct46.85%46.67%No change
CDsApr4.55%0.00%-4.6 pts

The results

Every big jump followed a release.

And the one stretch without new content is the one stretch where the credit union lost ground. Over the same period, its share of AI answers rose from 2.38% to 31.28%. It now outranks Chase, Bank of America, and City National Bank of Florida, all of which appeared in more than 40% of answers when we started.

Rank in AI answers: baseline, then April to October 2026Rank among 93 competing institutions: baseline #40, April #34, May #19, June #11, July #12, August #15, September #6, October #2. No new content from late April until FAQs were added in late August. #1 #10 #20 #30 #40 BaselineAprMayJunJulAugSepOct No new content until FAQswere added in late August #40 #34 #19 #11 #12 #15 #6 #2
Rank among 93 competing institutions: the March baseline, then each month from April to October 2026. Source: AmpliFi™ GEO.

What we learned

Four lessons for any market.

Consistency beats bursts

After the April wave, no new content shipped until late August, and rank slid from #11 in June to #15 in August while competitors kept growing. The climb resumed only when monthly releases began.

Product content moves the needle; brand content holds it

The biggest gains came from articles answering a specific money question in a local context: debt consolidation, credit cards, jumbo and condo mortgages. Membership and branch articles held steady at already solid levels.

The website can quietly cancel out the content

A broken FAQ, truncated summaries, and script-loaded CD rates each kept AI models from reading what the credit union had published. Technical audits paid off as much as new articles did.

Measure every release on its own

Per-release experiments showed which pieces worked and where to change course. When the CD article lost ground, the next savings article stated a certificate rate in plain text that AI models can read.

What's next

3.7 points from #1.

The next waves target the product questions where BrightStar is still rarely named, with every release measured the same way:

  • Savings
  • Home equity
  • Mortgages
  • Auto loans

How we measured

Rankings come from AmpliFi™ GEO visibility data covering 93 competing institutions, using full calendar months from March through October 2026. Questions ran through ChatGPT, and visibility is the share of answers that name an institution. Because the question set grew over the engagement, rank is the cleaner measure of progress than visibility alone.

When members ask AI, are you in the answer?

A 30-minute conversation about where you stand in AI answers today and which questions are worth winning.