Enquire privately
Shift 4 of 8

Not every problem needs a dispute.

FromAfter the complaintToBefore the dispute

Deliveries slip, specifications get misunderstood and plans change, and most of the time nobody is acting in bad faith. If the network notices when a transaction drifts from what was agreed and invites both sides to confirm or renegotiate, many problems end before anyone has to complain.

Level 1 Concept3 min readLooking for a pilot partner

Two days late, and nothing broken

Kavita’s part is due on 30 September. On the 28th nothing has been dispatched.

The system notices. It doesn’t say “the seller failed.” It says: “This transaction is moving away from what was agreed.” Both sides get a short message: a commitment looks at risk, so please confirm the status or propose a new plan.

Imran replies: “Two days late. The steel was delayed.” Kavita accepts the new date, and the commitment is updated. No complaint. No dispute. The relationship is intact, and the record shows what was agreed and when it changed.

Detect, explain, help correct

Because the transaction is already creating evidence (Shift 3), the system can see when reality starts to drift from the commitment: a missed milestone, a missing document, an unusual delay, conflicting terms, a detail the buyer hasn’t supplied, a payment conflict.

It invites correction first. Both parties get a chance to confirm, explain or fix, before anyone is blamed.

It supports renegotiation. When the original plan is no longer realistic, the path is: original commitment, proposed change, both agree, new commitment. The change becomes part of the record, so nobody argues later about what was agreed.

Software’s job here is detect, explain, help correct. It is never detect, accuse, punish.

Balanced by design

A buyer thinks “the seller hasn’t delivered,” but the record shows the buyer hasn’t yet supplied a required detail. The system surfaces that before any complaint is filed. A seller says “it’s complete,” but the agreed fulfilment evidence is missing. The system flags that too.

Neither side gets to define reality just by speaking first.
  1. Transaction
  2. commitment
  3. monitoring
  4. exception
  1. corrected or renegotiated
  2. the transaction continues
  1. unresolved
  2. formal dispute

The formal process stays focused on real disagreements, not every operational hiccup.

Go deeper

Why help before the complaint

Today the clock starts when someone complains. India’s e-commerce rules require a grievance officer to acknowledge a complaint within 48 hours and redress it within a month[1], and support teams across the country resolve large numbers of complaints well, once they are reported. This shift moves the moment of help earlier, to before anyone has to file.

The volume is real. Between 25 April and 26 December 2025, the National Consumer Helpline handled 67,265 refund-related complaints, nearly 40,000 of them about e-commerce[2]. ONDC’s own consultation paper asked, as an open question, how disputes could be prevented from escalating in the first place[3].

What software alone has done elsewhere

At eBay, a 2010 presentation by its director of online dispute resolution reported more than 60 million disputes a year, with 90% resolved entirely by software[4]. The figure is self-reported, over a decade old and from a different kind of marketplace. It shows what is possible in principle. It is not a forecast for this network.

Getting it right

Neutral language. A notification describes what is happening, never who is to blame.

Easy to respond. Confirming or proposing a change takes seconds, on any phone.

No pressure to accept. A proposed change is an offer. Either side can decline it and use the formal process.

A fair default. If one side doesn’t respond, the process says in advance what happens next, so silence doesn’t become a verdict.

What could go wrong

  • Alert fatigue.Too many alerts and people stop reading them.
  • Pressure dressed as help.A nudge to accept a change could become a way of pushing one side to give ground.
  • False alarms.A wrong early warning erodes confidence in the network, not only in the alert.
  • Silence.If one side doesn’t answer, the process needs a rule that is fair to both.

How we’d test it

Exception alerts for one transaction type, starting with delivery delays, compared with similar orders that have no alerts.

The test
Delay alerts
What we would measure
Share of flagged exceptions resolved without a dispute, how both sides rate the experience, and time saved.
What would end the idea
No better than similar orders left alone, or either side describes the alert as pressure.
In short

Alerts do no better than leaving orders alone, or either side experiences them as pressure.

Sources

  1. [1]Lakshmikumaran & Sridharan, Consumer Protection (E-Commerce) Rules, 2020 notified (2020)Checked via a legal summary; the Rules’ own text not read
  2. [2]Storyboard18, Consumer Affairs dept facilitates Rs 45 cr refunds in 8 months as e-commerce tops complaintsChecked; press report of Department of Consumer Affairs figures
  3. [3]ONDC, Building Trust in the ONDC Network: Consultation Paper (29 September 2022)Checked; proposals for comment, not adopted policy
  4. [4]Colin Rule, Using Technology to Manage High Volume Caseloads: The eBay/PayPal Experience, presentation, Washington DC (1 November 2010)Checked; self-reported and dated

Spotted a source we missed or a rule we misread? Tell us what we got wrong.

Some things can’t be put right in time. For those, the process has to be fair.