An RFP for reconciliation software should force precise answers on matching method, exception language, data retention, and Excel reality — otherwise you are buying a demo reel. Controllers who accept vague security and matching claims discover the gaps on close week, when switching costs are highest.
Which matching and output questions cut through the pitch?
- Do exact matches get reserved globally before any fuzzy scoring?
- Are outputs Matched / Partial / Not Matched — or a black-box score?
- Can every Partial carry a human-readable reason?
- How do you prevent a weak partial from stealing a true exact match?
- What amount, text, and date tolerances are configurable per job — and logged?
- Can we export a review-ready workbook with variance summaries without another BI project?
Which security and retention questions belong in every RFP?
- Where do source ledgers sit after close — and for how long?
- Are files used to train models or build derivative datasets?
- Are objects private-only, with TLS in transit and no public URLs?
- What metadata survives purge (counts, timestamps) versus row-level data?
- Is there a scheduled purge plus a storage lifecycle backstop?
Which workflow questions expose multi-year fantasy?
- Does the workflow assume Excel uploads this month, or only a multi-year ERP integration?
- Who owns Partials and Not Matched in your recommended operating model — named roles, not “the platform”?
- What happens when we cancel — do residual files remain, and how do we verify deletion?
Those twelve questions are enough to separate standardization from theater. Bring them to the briefing. Ask for a live pair of extracts, not a slide about AI.
How does ReconCore answer without a customer fairy tale?
Exact-first matching. Three-bucket Excel export. Zero file retention within 24 hours. Subscribe to begin — no free trial theater, no refunds, no founding discounts. If that commercial honesty matches how you buy other close-critical tools, start a job or talk to sales with your RFP list open.