Report — 2026-09-14

Do

• Run a client-visible milestone register: every milestone carries a named Devya owner, written acceptance criteria, a baseline date, a current forecast date, and a RAG status — updated weekly in the same artifact the client already reads, not in a separate status deck.

  The single worst review in today's pool did not fault the code or the people; it faulted the absence of a structure that tells a client who owes what by when. Accountability that lives only in Devya's internal tooling does not exist from the client's side.

  Hyperlink InfoSystem — verified 1.0/5 review (Quality 1.5, Schedule 0.5, Cost 0.5, Willing to Refer 0.5), Founder of a Dubai sports platform, May 2026, $50,000-199,999: 'The project lacked clear accountability, structured milestone tracking, and proactive escalation of delays.'

• Make delay escalation automatic and time-boxed rather than discretionary: any milestone whose forecast slips more than five business days past baseline triggers a written slip notice plus a recovery plan inside 48 hours — sent by Devya, unprompted, whether or not the client noticed.

  Clients in this pool tolerate a late milestone far better than a late disclosure. The reputational damage lands on the concealment window, not on the slip itself.

  Hyperlink InfoSystem's 1.0 reviewer recorded that communication was consistent throughout yet milestones were still missed with incomplete deliverables — steady contact that carries no bad news reads as evasion in the review afterwards.

• Put a calendar stop-loss on every engagement over three months: at 25% and 50% of elapsed contracted calendar time, publish a completion audit — percentage of contracted scope formally accepted versus percentage of calendar and budget consumed — and hand the client a written continue / renegotiate / stop recommendation from Devya's side.

  A drift of a few percent per sprint is invisible sprint-to-sprint and fatal at two years. The vendor that names the divergence first keeps the relationship; the one that lets the client discover it gets a one-star review with dates in it.

  Hyperlink InfoSystem, Dubai sports platform: engagement ran Sep 2023 - Sep 2025 and was still only 60% complete with the app 'minimally usable' — two full years of elapsed calendar with no recorded checkpoint that forced the conversation.

• Publish a dependency and risk map before each phase, at integration granularity: every third-party API, payment gateway, client-supplied asset and internal-team dependency listed with its owner, its lead time, and its timeline impact if it slips — issued before the phase starts, not when it breaks.

  Integration-heavy features are where estimates in this segment actually fail, and clients are asking for the forecast explicitly rather than for more frequent standups.

  Apptunix — client (life insurance company) on Clutch: the one improvement area was 'providing even more detailed project forecasting and timeline visibility for complex features that involve multiple integrations and dependencies... more granular visibility into potential risks, dependencies, and timeline impacts earlier in the process.'

• Lead the commercial pitch with a schedule commitment, not a rate: publish Devya's on-time-milestone percentage across the last twelve months and offer a contractual schedule remedy (fee credit against the next invoice when a baselined milestone slips without an approved change).

  In the $25-49/hr band the cost dimension is already saturated — buyers are not looking for cheaper, they are looking for a vendor whose dates hold. Schedule is the one sub-score every firm in today's pool underperforms on, so it is the only differentiator still unclaimed.

  Across the pool, cost meets or beats quality while schedule trails: Hyperlink 4.8 overall (Quality 4.8 / Cost 4.8 / Schedule 4.7), Techgropse 4.8 (Quality 4.8 / Cost 4.7 / Schedule 4.7), Apptunix 4.5 (Quality 4.5 / Cost 4.5 / Schedule 4.4 — lowest sub-score in the pool).

• Instrument the client's baseline metrics in week one — cycle times, error rates, manual hours, ticket volume — and re-measure at handover, so the closing review can be written in deltas rather than adjectives.

  The reviews that actually sell the next deal are the ones carrying numbers. A client cannot quote an improvement they never measured before Devya arrived, and by the time they want to write the review the baseline is gone.

  Apptunix — automotive parts retailer review quantifies the outcome throughout: order processing cut from 10-12 minutes to 3-4 minutes per order (65-70% reduction), order error rate from 8-10% down to below 2%, manual work reduced over 60%, real-time tracking across 100,000+ products. That review does more commercial work than any five-star sentiment entry on the same profile.

• Optimise the proposal and directory profile against the three filters GCC buyers actually name — directory rating, geographic proximity, and budget fit — and treat Cairo's one-hour flight and shared working day with Riyadh and Dubai as a headline commitment (named on-site days per quarter, Gulf-hours coverage), not a line in the About section.

  Buyers in this segment state their shortlisting criteria verbatim in the review form. Proximity is a stated, checkable criterion that Devya wins on against the Indian and Eastern European shops in the same rate band — but only if it is advertised as a delivery commitment.

  Techgropse FZC (Dubai) reviewers, 'Why did you select Techgropse over others?': 'High ratings / Close to my geographic location / Pricing fit our budget' (e-learning platform) and 'High ratings / Pricing fit our budget / Good value for cost' (transcription services), with discovery routes recorded as 'Clutch Site' and 'Online Search'.

• Treat third-party proof as perishable and rented: date every review Devya cites, retire evidence older than 24 months from the proposal, target one fresh verified review per quarter, and mirror the full review wall on a Devya-owned page that survives a profile change.

  A high rating with a dead date stamp reads as a firm that stopped winning, and the profile URL itself is not durable — three separate proof channels in today's pool broke, disappeared, or redirected inside one rotation.

  DOIT Software holds 4.9/21 on Clutch with the review base running 2017-2022 and no submissions after late 2022; 10Clouds' Clutch profile now returns HTTP 410; clutch.co/profile/itechart-group redirects to Vention following the rebrand; Robusta Studio and EJADA render with no review wall at all.