RFP Win/Loss Picture — Submitted Issues (Last 12 Months)

Scope

This document reviews GitHub issues in datopian/sales where:

• the issue has label rfp
• the submitted label was added within the last 12 months

Exclusions applied for this first pass:

• RFIs are separated into their own table and excluded from the main RFP picture
• #533 and #493 were removed because they are older historical wins and not useful for the current-year picture
• #753 was removed because it is a G-Cloud application rather than a real bid outcome

Summary

30 submitted RFPs reviewed (last 12 months, as of the 2026-06-05 analysis snapshot). Win rate: 10% (3/30 full set), 17% on known outcomes — but a large share of the tracker still has no recorded outcome, so both figures remain uncertain and should not be treated as current live stats without refresh. The biggest finding is not a refined theory of why we lose; it's that the learning signal is fragmented and outcome recording is broken.

Where feedback exists, losses cluster into four buckets: proposal too generic/not deep enough on scored criteria; missed mandatory compliance requirements; pricing unjustified (in both directions — too low on 3PODS, uncompetitive on UAE); avoidable process failures after submission (e.g. LG Inform shortlisted then disqualified because remote presentation not accepted). Source analysis was corrected on 2026-07-22 to classify Statens Vegvesen (#702) and NHSBSA (#613) as invited / relationship-led rather than proactive. With that correction, all 3 confirmed wins in this dated snapshot are invited / relationship-led hosting/support contracts. Next step: refresh outcomes since 2026-06-05, then close out remaining pending/unknown issues and select 3–5 bids for deeper retrospective.

Summary Stats

• Main RFP set reviewed: 30 issues
• accepted: 3
• rejected: 15 (updated: #782 moved from unknown to rejected based on current GitHub labels)
• pending: 7
• unknown/to be updated: 5
• Known outcomes recorded: 18/30
• Still not closed out: 12/30

Two useful ways to read the current picture:

• Win rate across the full reviewed set: 3/30 = 10%
• Win rate across known outcomes only: 3/18 = 17%

The second figure is more informative for now, but also potentially biased because the unknown set is large.

Source split (see Source Analysis below, corrected 2026-07-22):

• invited (including as subcontractor): 11 — 3 accepted, 5 rejected, 0 pending, 3 unknown
• proactive: 19 — 0 accepted, 10 rejected, 7 pending, 2 unknown

Main RFP Table

Source Analysis: Invited vs Proactively Sourced

Counts

Win Rates (Known Outcomes Only)

Invite Channels

What This Tells Us

After correcting #702 Statens Vegvesen and #613 NHSBSA to invited, all confirmed wins in this snapshot are invited / relationship-led: 3/8 known invited outcomes were wins (38%) vs 0/10 known proactive outcomes (0%). This is still provisional because the numbers are small, 12/30 issues were still pending or unknown as of the snapshot, and several outcomes have changed since 2026-06-05.

The more important pattern may be the combination of source and contract type:

1. All 3 wins are invited / relationship-led hosting/maintenance/support contracts — #741 Malmö, #613 NHSBSA, and #702 Statens Vegvesen. Zero wins on new-build or custom development in this snapshot.

2. Invitation / relationship-led entry appears helpful, but not sufficient. Statens Vegvesen (#702), NHSBSA (#613), and Malmö (#741) were invited / relationship-led and won; the two highest-stakes invited bids — UAE (~$300k) and Scotland (~$1M) — were still rejected.

3. Subcontractor bids via partners have not produced wins. Both resolved sub bids (#775, #776) are rejections. This route may not fit Datopian's profile, or the partner selection criteria need sharpening.

4. Implication for follow-on priorities:

   • #52 Improve RFP response quality: still a primary lever, especially because invited bids can still fail on mandatory criteria, pricing, or scored-response quality
   • #51 Improve RFP sourcing via Meiran fit reports: still valid, but the gain lies in fit quality (surfacing opportunities that match the hosting/CKAN/CKAN-adjacent profile where we win) rather than in increasing invitation volume

---

RFI Table

These are in the GitHub rfp set but should not be counted as part of the core RFP win/loss picture.

What Seems True So Far

• Outcome recording is still weak. 12/30 submitted RFPs are still not properly closed out (improved slightly: #782 resolved to rejected via current GitHub labels).
• Rejections are much more visible than wins in the current-year RFP set.
• The tracker often contains useful outcome evidence in comments before the issue is fully updated with outcome labels. GitHub is usable as the source of truth, but only if closeout discipline improves.
• Some losses already point to concrete operational causes rather than generic bad luck.
• Corrected 2026-07-22: all confirmed wins in the 2026-06-05 snapshot are invited / relationship-led after reclassifying #702 Statens Vegvesen and #613 NHSBSA. The winning profile is concentrated in relationship-led hosting/support contracts, and the source-channel analysis needs a live refresh before drawing stronger conclusions.

Early Learnings from Recorded Feedback

• #653 DfT Local Transport Data: feedback suggests the response was too generic and did not go deep enough on one concrete example against the scoring criterion.
• #646 Executive Council EEIA: public feedback suggests our responses may not have felt sufficiently relevant to the solicitation.
• #678 3PODS: comment suggests pricing was considered too low / not well justified, which is a useful counterpoint to the more common "too expensive" theory.
• #780 UAE Ministry of Investment: prior notes indicate both mandatory-criteria failure and commercial uncompetitiveness.
• #661 LG Inform 2026: we were shortlisted but then effectively lost because the next stage required an in-person London presentation and remote presentation was not accepted.

Interpretation

At this stage, the biggest conclusion is not yet a refined theory of why Datopian loses. The biggest conclusion is that the learning signal is still fragmented.

However, there are already enough concrete signals to justify deeper review of at least these buckets:

• proposal quality / specificity against scored questions
• mandatory compliance / qualification requirements
• pricing competitiveness
• process constraints that cause avoidable losses after submission, such as presentation logistics or follow-up gaps

Recommended Next Step

Use this table as the first pass, then focus only on the unresolved cases:

• review the pending issues
• review the unknown/to be updated issues
• mark confirmed losses vs still-pending bids
• add one-line reasons where known
• then select 3-5 representative bids for deeper retrospective review