A 30-day bid-software pilot: how to decide whether to adopt it.
Adopt the tool only if a labelled evaluation set, run against a frozen baseline, meets acceptance criteria you wrote before seeing results: error severity, reviewer time, and evidence traceability. This page is a method. It is not a test of selrob. Part of our guide on bid and tender software.
Published: 16 September 2026 · By Bill · Reviewed by Andre · selrob
Takeaways
Write pass, review and fail before anyone looks at the tool’s output.
Start with public, synthetic or authorised historical packs. Live confidential use follows the buyer’s permissions.
Record the baseline workflow: who reads the pack, how long that takes, where the call is written down, and what source the call cites. Build an evaluation set you can name. Freeze the expected answer for each pack, including the bid / no-bid / review call and the clauses that drove it. Freeze the acceptance criteria on the same day. Then run the tool. Score misses by severity. Count reviewer minutes. Check that every cited location exists in the pack.
selrob sells qualification software and publishes this site. The example plan is a method for any vendor. It is not a result from testing selrob, and it does not report selrob results.
Limits of this method: a five-pack set is not a population. Features you did not put in the test are untested. English Australian packs do not stand in for other jurisdictions. Qualification is not response drafting. Live confidential packs stay out until the buyer’s permissions and the vendor’s data terms allow them. Ask the vendor questions on the buyer’s guide before the clock starts. Use a compliance matrix if the tool is supposed to list requirements you will still have to evidence.
Example plan
Thirty days, 1 to 30 October 2026
Hypothetical. Dates and owners are invented to show the sequence; they are not a client result. Acceptance criteria are written on day 4, before any tool output is reviewed.
Hypothetical 30-day bid-software pilot plan
Days
Activity
Output
Owner
1–2
Record baseline workflow for qualification.
Steps, minutes per pack, where the call is stored.
Bid lead
3
Assemble evaluation set. Public, synthetic, or authorised historical only.
Five labelled packs. No live confidential files.
Bid lead
4
Freeze expected answers and acceptance criteria.
Worksheet dated. Criteria dated. No tool output yet.
Reviewer
5
Configure the tool in a non-production workspace.
Access log. Data terms checked.
Admin
6–20
Run each pack. Log output against frozen answers.
Results log.
Operator
21–23
Score error severity and reviewer minutes.
Severity and time columns complete.
Reviewer
24–26
Traceability: every cited clause exists in the pack.
Pass or fail per pack.
Reviewer
27–28
Compare to the day-4 criteria.
Pass / Review / Fail record.
Decision owner
29–30
Adopt, extend the set, or stop.
Dated decision.
Decision owner
Expected answers
Frozen on day 4, before the tool runs
Hypothetical evaluation set. SYN packs are synthetic. HIST-01 is labelled as authorised historical and anonymised. Public-style means a structure based on a public RFT, not a live buyer pack. Not a client result. Not a selrob test.
Hypothetical expected-answer worksheet frozen before tool output
Pack
Source
Baseline call
Why
Baseline minutes
SYN-01
Synthetic public-style services RFT
No-bid
Mandatory ISO 27001 not held. Conditions §3.2.
18
SYN-02
Synthetic panel RFQ
Bid
Known scope. Payment 30 days. Capacity clear.
12
SYN-03
Synthetic FM schedule
Review
Abatement unread by legal. Schedule 4.
22
HIST-01
Authorised historical, anonymised
No-bid
Indemnity uncapped. Draft contract §24.
25
SYN-04
Synthetic RFP
Bid
Roster fits. Liability capped inside insurance.
16
Results log
Filled after day 4, labelled hypothetical
Hypothetical log for a generic tool. Severity: High = tool says Bid when baseline is No-bid on a mandatory or unread liability fail. Medium = Bid versus Review disagreement. Low = same call, missing citation. Not a selrob test.
Hypothetical bid-software pilot results log
Pack
Baseline
Tool
Severity
Reviewer min
Citation exists
SYN-01
No-bid
Bid
High
14
No. Mandatory missed.
SYN-02
Bid
Bid
–
9
Yes
SYN-03
Review
Bid
Medium
11
Yes. Schedule 4 named, unread legal ignored.
HIST-01
No-bid
No-bid
–
16
Yes. §24 named.
SYN-04
Bid
Review
Medium
13
Yes
Decision record
Criteria written 4 October 2026, scored 28 October 2026
Hypothetical. Overall Fail in this example because a high-severity miss is present. The record exists to show the method. It is not a product claim.
Hypothetical pass, review and fail record for a bid-software pilot
Criterion
Threshold (day 4)
Result
Outcome
High-severity errors
0 on the evaluation set
1 (SYN-01)
Fail
Median reviewer minutes
≤ baseline median (18)
13 (9, 11, 13, 14, 16)
Pass
Traceability
Every tool call cites a location that exists
4 of 5
Fail
Export of working
Possible without the vendor
Yes in this example
Pass
Overall
All must pass
Two fails
Fail
Baseline minutes 12, 16, 18, 22, 25; median 18. Reviewer minutes 9, 11, 13, 14, 16; median 13. Faster review does not rescue a high-severity miss. Fail means stop or extend the set. It does not mean bargain the criterion after seeing SYN-01.
What does this mean for the bid/no-bid call?
A tool that says Bid on a mandatory fail is a new way to start a zombie. Do not adopt it on reviewer-time savings. Do not put live confidential packs through it because the calendar said day 30. Write the criteria. Run the set. Record the call. The buyer’s guide is the question list; this page is the evidence that those answers held still when you looked.