Home / The Science Behind NexSpark
Our approach
The science behind how NexSpark works
By Matthew Firth, founder of NexSpark · Jul 20, 2026 · 5 min read
Every dating product claims to be "science-backed." Most of that is marketing. Here's what we actually use, cited plainly, including the parts of the research that are inconvenient for a matchmaking company to admit.
How we read compatibility: attachment theory
Attachment theory, developed by John Bowlby and Mary Ainsworth and extended to adult romantic relationships by Hazan and Shaver in 1987, describes how people vary in how they seek or avoid closeness, reassurance, and interdependence in relationships. It's one of the most replicated frameworks in relationship psychology, and decades of research since have consistently found that securely attached adults report longer, more trusting, and more satisfying relationships. NexSpark's questions read your attachment patterns, and we pair for a bond you can actually feel secure in, not just an interesting conversation.
Source: R. Chris Fraley, "A Brief Overview of Adult Attachment Theory and Research," University of Illinois.
What a secure bond looks like day to day
Attachment theory tells us whether someone can bond securely. Dr. Sue Johnson's Emotionally Focused Therapy work gives a practical read on what that looks like in practice: partners who are Accessible, Responsive, and Engaged (the A.R.E. framework), reachable when it matters, responsive to each other's needs, and consistently present. We look for that pattern specifically, because it's the difference between chemistry on a first date and a bond that holds up six months in.
How you recover after a fight matters more than how you get along
Psychologist John Gottman's decades of couples research identified four communication patterns, criticism, contempt, defensiveness, and stonewalling, that reliably erode relationships over time. The ability to repair after conflict, rather than avoiding conflict altogether, is one of the more consistently supported findings in the field. We factor in how each person tends to come back together after disagreement, not just how compatible you look on a good day.
Where we're careful: Gottman's famous "94% divorce prediction accuracy" is widely repeated but methodologically disputed, the model was fit to outcomes that were already known, which is a different thing than a true prospective prediction. We rely on his findings about conflict patterns, which are well supported, and don't repeat the inflated accuracy claim.
Source: The Gottman Institute, Research FAQ.
Personality, and an honest word on similarity vs. difference
We measure the Big Five personality traits (openness, conscientiousness, extraversion, agreeableness, neuroticism), the most established framework in personality psychology, alongside values, dealbreakers, and life direction. Here's the part most matching products don't say out loud: a large meta-analysis of over 300 studies found that perceived similarity strongly predicts initial attraction, but actual trait similarity is a weak predictor of long-term relationship satisfaction. Some research even suggests personality similarity among married couples is essentially unrelated to how satisfied they are. So we don't assume "more alike" is automatically better. We pair some things for similarity, values, faith, the life you're building toward, and others deliberately for complement, like temperament and social energy, the way real couples actually balance each other.
Why a real matchmaker reviews every introduction
This is the least flattering, most important thing we can tell you: the peer-reviewed evidence that personality-based matching algorithms, on their own, reliably predict who will be compatible is genuinely mixed, and for some popular frameworks (MBTI-style typing especially) it's weak. That's not a reason to throw out structured data, it's a reason not to trust an algorithm alone with something this important. NexSpark uses AI to do the heavy comparison work across a large pool of verified members, but a real matchmaker reviews and approves every introduction before it's ever sent. Nothing is auto-matched. The research is exactly why we built it this way instead of the other way around.
The loneliness problem this is actually solving
None of this matters in the abstract. The U.S. Surgeon General's 2023 Advisory on loneliness found that poor social connection carries a mortality risk comparable to smoking up to 15 cigarettes a day, and is linked to a 29% higher risk of heart disease and a 32% higher risk of stroke. Roughly half of U.S. adults report feeling lonely. That's the actual problem NexSpark exists to help with, not swiping volume, not engagement time.
Source: U.S. Surgeon General's Advisory, "Our Epidemic of Loneliness and Isolation," 2023.
Common questions
Is NexSpark's matchmaking actually based on science?
Parts of it are grounded in well-established research: attachment theory, the Gottman Institute's findings on conflict repair, and the Big Five personality model. Other parts, like whether personality-matching algorithms in general predict long-term relationship success, are genuinely mixed in the research, and we say so rather than overselling it.
Does similarity or difference make a better match?
It depends what you're measuring. Perceived similarity strongly predicts initial attraction, but actual trait similarity is a weak predictor of long-term satisfaction. That's why we pair some traits for similarity and others for complement, rather than assuming more alike is always better.
Can an algorithm really predict who I'll be compatible with?
Not reliably on its own. That's exactly why a real matchmaker reviews every introduction at NexSpark instead of an algorithm auto-sending matches.
What is the Gottman four horsemen research?
Four communication patterns, criticism, contempt, defensiveness, and stonewalling, that reliably show up in relationships heading toward failure. It's well-supported work, though the "94% prediction accuracy" figure often quoted is a retrospective model fit, not a prospective prediction, and we don't repeat that claim.
See who the research says you're compatible with
Answer a few honest questions, no account needed to start.
Reveal your ideal matchFurther reading: how your archetype and match are actually built · attachment styles, explained · what actually predicts a lasting relationship.