Fintech

The signup form is the only conversation you get.

A phone number is all a funnel collects before someone abandons it. Income band, credit health and employment come back on that identifier in the flow — so onboarding, limits and offers adapt while the person is on the screen.

1B+Persons
150+Attributes each
25+Consumer data sources
2M+Interactions monitored
In production

Businesses already running on the graph.

Digital products and platforms deciding on their own users with ZipLabs data.

  • SeekOut
  • Gem
  • 6sense
  • AeroLeads
  • PeopleBox
  • Weekday
  • Isprava
  • Square Yards
  • Scripbox
  • Keya Homes
  • L&T Realty
  • Scaler
  • Sell.do
  • Babblebots
What breaks

Every question you ask costs a user.

A funnel has to choose between knowing who somebody is and keeping them in it. That trade is the whole problem.

Drop-off

The form is already as short as it goes

Every field added to onboarding is paid for in abandonment. So the flow stays short, the record stays thin, and the product decides on almost nothing.

Thin file

No history is read as no good

Someone with a short credit history gets the same treatment as someone with a bad one. A whole segment of earning, employed, perfectly good customers is declined by default.

Flat

One offer for everybody who signs up

The same limit, the same product, the same onboarding for a graduate and a director. One of them is underserved and the other is a risk you priced as if they were the same.

What changes

Decisions made inside the funnel.

Each of these runs on the identifier the user already gave you, while they are still in the flow.

  1. asking for it 9 fields 4,100 in
    looking it up 3 fields 6,800 in
    a shorter form that knows more

    Fewer fields, the same knowledge

    Everything you were going to ask for is already attached to the number they typed. Take the questions out of the form and the profile still arrives — just without the people who quit halfway.

  2. + +91 ••••• 6032 step 1 of 3 A. Rao · 29 · salaried anonymous known
    enriching in the flow… applying the policy… profile returned before step 2

    A number becomes a customer, mid-flow

    The enrichment call happens between one screen and the next. By the time the user sees the second step, your policy engine already knows who it is talking to.

  3. JS J. Shah no history salaried · 5 yrs approvable
    stable employer income band

    A thin file is not a bad customer

    Employment, tenure and income band say something a short credit history cannot. That is the difference between declining a segment and being the product that finally said yes to it.

  4. signups: same form, same minute
    sizing the offer…
    1 thin file, salaried starter
    2 ₹35–50L, 800+ premium
    one funnel, three products

    One funnel, three products

    The same signup routes to the starter card, the salaried offer or the premium tier, because the flow can tell them apart. Nobody is shown a product they will be declined for.

  5. DN D. Nairuser, 2 yrs change observed…
    new employer income band up
    DNPVHK 3 due a limit review

    A raise is a limit increase

    Your users do not file a change of circumstances. Watching the base for a new employer or a higher band is how an upgrade offer goes out in the month it became true.

  6. salaried₹12L+ metro · 24–40
    building the audience…
    likely to be approved 64% of spend
    ✓ fewer installs you have to decline

    Stop paying to acquire people you decline

    An install that fails underwriting cost the same as one that passed. Targeting on the attributes your own policy uses moves that spend to people the product can actually serve.

What powers it

Call it, or let it fill in behind you.

The same profile, either from your own service in the flow or written onto the record in the CRM your team uses.

Security and compliance

Regulated buyers, audited answers.

Matching runs on hashed identifiers, purpose and retention are agreed before the first record moves, and every attribute traces to where it came from — which is what a model risk review and a data protection officer each need, for different reasons.

Security & Trust Centre
GDPR EU & UK
CCPA California
India
SOC 2 Audited
FAQs

What an engineering
team asks first

Is this fast enough to sit inside a signup flow?

It is built for it — a single identifier in, a structured profile out, called synchronously between screens. Latency and throughput targets are agreed with you against your actual volume rather than quoted as a headline, because a number that holds at ten requests a second and not at a thousand is not worth printing.

Does this replace a bureau pull?

No. Credit health here is a profile attribute for routing, sizing and prioritising — deciding what to show someone and what to ask them next. The regulated pull happens where your policy says it happens, on the smaller set of people who get that far.

Which attributes are India-only?

Income band, credit health and existing obligations are India coverage. Demographics and professional data run globally on the same 1B+ person graph, so a product operating outside India still gets the professional layer. Consumer Intelligence lists what each package contains.

Can we get told when a user changes rather than polling?

Yes — webhook delivery is part of the Grow side. Register the base you care about and the change arrives when it is observed, which is what makes limit reviews and upgrade offers something you run continuously instead of quarterly.

What happens when you have no match?

You get an explicit no-match rather than a filled-in guess, and you are not charged for a hit that did not happen. A flow that has to distinguish between "we know nothing" and "we know this is thin" cannot work on a vendor that quietly returns an average.

Test it on real traffic

Run it against last week's signups.

Send a sample of identifiers from a real cohort. We will return the profiles and you can check them against what those users actually did next.