AI search and Spanish fintech affiliate: what to watch for as the curve catches up

The fintech affiliate playbook for English-language markets has been visibly under pressure for about two years. The published 2026 evidence has sharpened over the last six months — and the news for affiliate operators is worse than the 2025 data suggested.

Ahrefs republished their AI Overviews study in February 2026 across 300,000 keywords (150,000 with AIOs, 150,000 without), using aggregated Google Search Console data. They found AIOs now correlate with a 58% drop in CTR for the top-ranking page — up sharply from the 34.5% they measured in March 2025. Seer Interactive’s late-2025 research across 42 brands found organic CTR is 35% higher when a brand is cited inside the AIO vs an AIO that does not cite the brand. BrightEdge’s February 2026 industry tracker measured 58% year-over-year growth in AIO coverage across nine industries, including finance educational queries climbing from 16% to 67% AIO penetration in 18 months.

Those numbers settle the direction without settling every magnitude in every market. The Spanish-language fintech market is not the same market as the US market, and the data published about it specifically remains thinner.

This article is the honest read on what to expect, what to instrument now, and where the limits of confident analysis sit.

Spain is not ‘pre-AIO’ any more

Google rolled AI Overviews into Spain in early 2025 as part of its European expansion, and across more than 40 languages including Spanish by mid-2025. By 2026, Spain has had AIO at production scale for over a year. The “Spanish market is pre-AIO” framing some operators still use is no longer accurate.

What is accurate is that the citation graph in Spanish fintech is sparser than the English one, and that means the practical experience inside Spain still looks earlier than the US experience does. Two structural reasons:

Reason one — the citation supply. The engines prefer to cite sources that are structured, authoritative, schema-marked, and cross-linked. The English-language fintech information ecosystem (regulators, comparison sites, financial press, listed-company filings, ratings sites) has been building those properties since well before AI Overviews existed. The Spanish-language equivalent has fewer of those properties at the density the engines have learned to weight heavily. Investopedia wins 42% of finance AIO citations in BrightEdge’s data; the Spanish ecosystem does not yet have a clear equivalent at that concentration of authority.

Reason two — query-class variation matters more than the country average. BrightEdge measured finance overall at only ~5% AIO coverage but finance educational queries at 67%. “What is” finance queries trigger AIOs 91% of the time. The same pattern almost certainly holds in Spanish — informational queries are far ahead of real-time price queries on AIO coverage — but the published per-query-class data for es-ES specifically is not granular enough to quote.

So the right framing is not “Spain is behind.” The right framing is “Spain is roughly where the US was 12–18 months ago in citation-graph density, on top of AIO infrastructure that is already live.” The window to build entity coverage on the Spanish side is shorter than it was twelve months ago and closing.

What the bifurcation in the buyer journey looks like

The framing that has held up across English-language markets is two parallel paths through the buyer journey, with very different economics:

Path A — AI-mediated. The buyer asks an assistant a comparative question (“which is the cheapest broker for US ETFs from Spain”). The assistant responds with a shortlist of two to four named brands plus fee notes and a one-line caveat. The buyer types one of those brand names directly into the URL bar. The visit is recorded as direct traffic. The brand was selected upstream by the assistant.

Path B — direct-click. The buyer still searches Google, still clicks a result, still lands on a comparison page. This path survives but shrinks under pressure from AIOs and the per-result CTR effect that Ahrefs has now measured at 58% for the top-ranking page.

The unit economics of the two paths are different. Path A buyers tend to arrive further down the consideration curve — the assistant has already done some qualification work. Path B is the legacy funnel, and the legacy funnel is sensitive to the AIO-coverage trajectory. Whichever path is dominant for a given market and vertical, the right reporting needs to make both visible. Standard SEO dashboards report Path B and lose Path A in the gap.

The metrics worth instrumenting before the shift fully arrives

Most operators discover they need these metrics only when revenue is already moving. The right time to start tracking them is one or two quarters earlier, when there is no urgency and the baseline can be built calmly.

AIO coverage on tracked queries. Sample your top commercial-intent queries weekly and record whether an AI Overview appears. Semrush Sensor has a public coverage tracker; manual sampling fills the gap. The baseline matters because the rate of change is the interesting signal, not the snapshot.

Citation rate across a fixed prompt set. Fifty prompts, run monthly across ChatGPT, Perplexity, Gemini, and AI Overviews. Record whether your brand is named, whether competitors are named, what domains are cited. The number is meaningful as a delta over time even if the absolute level is fuzzy.

Branded-search lift in Google Search Console. Filter Search Console to queries containing your brand. Watch the absolute and relative trajectory. A rising branded line in periods when you have not run paid brand campaigns is the fingerprint of upstream AI mentions doing the qualification work.

Direct-traffic ratio to your landing pages. A rising direct ratio without a corresponding paid or PR push is the same fingerprint, observed downstream. Both metrics independently catch what last-click attribution loses.

Time-to-first-action on signups. Internal product analytics will show this. If the time from signup to first deposit or first trade compresses without a product change, AI-mediated leads are arriving more pre-qualified.

Chat-widget question specificity. Anecdotal but useful. As AI-mediated leads grow as a share, the questions arriving at the chat widget shift toward specific edge cases (“can you confirm what you charge to wire EUR to USD on weekends”) and away from generic discovery (“what does your platform do”). The basics were pre-answered upstream.

None of these signals require proprietary data, an enterprise SaaS contract, or paid tools to start tracking. They require discipline.

What I am not claiming

Three things worth being explicit about, because honesty on absences is what calibrates a field.

I am not claiming a specific Spanish-language AIO penetration rate today. Published trackers cover US data better than Spanish data. Any specific es-ES penetration number you see in marketing content today is either an internal sampling effort (often with small samples) or extrapolated from US data. Either is reasonable as a working estimate; neither is a defensible point claim.

I am not claiming that paid social or other channels are bailing operators out. There is no strong public evidence either way at the market level. Individual operators have shared their own results in both directions; aggregated, evidence-grade data is thin.

I am not claiming partner-level performance is widening in a measured way. It is a reasonable hypothesis — if AI assistants compress shortlists to two to four brands, partners with strong citation footprints should compound and partners without should atrophy. But that hypothesis needs operators to instrument partner-level performance against citation footprint over multiple quarters before anyone should report it as a settled fact.

What to do this quarter

Three concrete moves, in priority order, that work whether the Spanish-language shift arrives in Q3 2026 or Q1 2027.

Move one — instrument the four leading indicators. Citation rate, AIO coverage on tracked queries, branded-search lift, direct-traffic ratio. Set the baseline now. None of these require committing to a strategic shift; they are the cheapest insurance against being late to recognise one.

Move two — fix the entity layer on the two highest-intent pages. The sign-up landing page and the most-trafficked comparison page. Schema (Organization, FinancialProduct or the appropriate subtype), sameAs links to regulators and external profiles, a definitional opening paragraph, tabular fee facts. These changes have value independent of the AI shift — they help SEO too. They become disproportionately valuable when citation rate is a metric the operator is reporting against.

Move three — separate the GEO budget from the SEO budget on the next planning cycle. Even if the absolute amount is small, the act of separating the line forces the team to articulate what production work is meant to win clicks vs. what production work is meant to win citations. That clarity tends to surface gaps the unified budget hides.

The Spanish fintech market is not in trouble. It is on the same curve the US fintech market has been on for the last two years, with a lag whose exact length is not yet possible to defend with public data. The operators who treat the lag as a building window — instrumenting calmly, fixing the entity layer, separating the GEO line — will arrive at the inflection ready. The operators who wait for the revenue line to move first will discover that the entity work they need is six months longer than they have.