Five decisions waiting on a person

1. Eighteen clients are being quoted the same three price lists

What you need to know first

Every ForgeFX proposal includes a cost table: a spreadsheet listing roughly ten features being built, with three price tiers — a smaller option, a middle option, and a full option. It is the page the customer looks at hardest. There are currently 30 of these spreadsheets sitting in client folders.

The problem

Eighteen of those 30 are copies. Not similar — identical. Same ten feature names, same feature descriptions word-for-word, same three prices to the dollar, sitting in eighteen different clients' folders with only the client's name changed. They collapse into just three source documents:

Only 10 of the 30 spreadsheets are genuinely written for the client whose name is on them. Of the copies, four reach a customer within two weeks and six are already past their deadline.

Why it happened, and why it is not a bug to fix

A tool generates a starter cost table when an opportunity needs one, filling the ten rows with generic content and placeholder prices. That is reasonable — an opportunity with no cost table at all is worse than one with a draft. The failure is at the other end: nothing marks a generated table as unfinished. A separate check certifies tables as "ready," and it used to certify on structure alone — ten rows present, three prices present, no blank cells. A generated placeholder passes all of that. So drafts were being counted as finished work and nobody was ever told to rewrite them.

That certifier has already been tightened, this week, to fail anything that is a copy. It now reports 1 ready, 13 needing work, 4 with no table at all out of the 18 opportunities awaiting a proposal. That is the honest number, and it is why this needs your decision — the software has gone as far as it can.

What is actually being asked

The question is what those ten generated rows are for. Two coherent answers, and the current state is neither.

1. A. They are drafts, and should look like drafts Keep generating them, but stamp every generated row as unfinished — visibly, in the document — until a person rewrites it. Nothing generated can be certified ready or sent to a client. Effect: the four quotes due within two weeks get flagged for rewrite today. Nothing goes out with borrowed prices. Costs writing time on 13 opportunities.
2. B. They are meant to ship, so the prices must be real The generator keeps producing complete tables, but prices are calculated per client from that job's actual scope instead of inherited from a template. Effect: bigger change, and it needs a pricing rule the generator can follow. Nothing goes out with another client's numbers.
3. C. Stop generating them entirely An opportunity with no cost table shows as missing rather than as a copy. Every cost table is written by hand. Effect: honest, but the four opportunities currently showing "no table" become 18. Slowest option.
4. D. Leave it — this is acceptable for now The copies stand and the check stops reporting them. Effect: no work. The risk is a customer noticing, and Group B is the case where that is most likely.

2. Two Halliburton records describe the same job

What you need to know first

Each piece of prospective work gets one opportunity record in Salesforce. That record carries the deadline, the dollar figure, and the history. Two records for one job means the pipeline double-counts, and any automated work — filing call recordings, building proposal folders — has no way to know which one is real.

The problem

There are two open Halliburton records for the same real-world-controls conversion work:

This is not a guess

Sara raised it in the sales Slack channel and nobody answered. The creation records match her account exactly. Both records are still open today.

What is actually being asked

1. A. Keep the newer one Copy the $299,000 onto "Halliburton HWO RWC Support," then close or delete the May 2025 record. Effect: the 21 August deadline and the proposal folder stay intact. Also unblocks the stuck Halliburton call in item 3 below.
2. B. Keep the older one Copy the 21 August deadline onto "RWC Conversion to VR Training Simulation," move it to Proposal Due, then close the August record. Effect: keeps the longer history and the original owner. The proposal folder gets rebuilt under the older record's name.
3. C. They are genuinely two different jobs Say so and this closes with no change — but the older one likely needs a deadline and the newer one a dollar figure. Effect: no deletion. The Halliburton call in item 3 still needs one of them named.

Deleting a Salesforce opportunity destroys its history, so no automated process will do it. Whoever decides has to click it.

3. Five recorded customer calls are filed against nothing

What you need to know first

Customer calls are recorded and transcribed automatically. The transcript then gets attached to the relevant opportunity in Salesforce, and a copy is saved into that opportunity's proposal folder — so whoever writes the proposal can read what the customer actually said without going hunting.

There is a hard rule: never guess which opportunity a call belongs to. A call filed against the wrong opportunity is worse than one filed against none, because the next person reads it as fact about the wrong deal. So when it is ambiguous, the call is held and reported here.

The five, and why each is stuck

What is actually being asked

These are four separate small decisions, not one. Pick per row — or take the bundle.

1. A. Handle the two urgent ones now, defer the rest Name the opportunity for the two John Deere calls — one line each, "both belong to Custom Content" is enough. Leave Halliburton until item 2 is decided, and leave the Vermeer and IDE calls for later. Effect: two calls reach the people writing the 18 and 19 August proposals. About two minutes of your time.
2. B. Clear all five in one pass Name the John Deere opportunities; decide item 2 so Halliburton files itself; say whether a cancelled meeting should be filed at all; and either open the IDE recording so the service generates its summary, or say it may be filed without one. Effect: nothing left held. Roughly ten minutes, and it depends on deciding item 2.
3. C. If a call is ambiguous, file it against every candidate Change the standing rule: when several opportunities are plausible, attach the call to all of them rather than holding it. Effect: nothing is ever held again. The cost is that some opportunities carry calls that are not about them, and later readers cannot tell which. This is the same question as item 5, applied to customer calls.

4. Half the priced quotes on disk belong to no opportunity

The problem

Of the 30 priced cost tables sitting in client folders, 15 are connected to no opportunity record at all:

Cargojet · General Motors · Jet Propulsion Laboratory · Wirtgen · JerrDan · StrataTech · Minot Enterprises · IHE Heavy Equipment Operator School · IS4S · ICSurveys · TMEIC · Campbell County Rockpile Museum · Alternative Logistics · a John Deere 644P wheel loader · a TaskUs emergency-response simulator

Ten of those fifteen are copies of a price list another client is also being quoted, so they overlap with item 1.

The reverse gap exists too. Four opportunities point to a cost table that is not stored anywhere the check can reach, so nothing can verify or read them: Applied Materials Semiconductor, ERock Technician, Metropolitan Council Light Rail, and TaskUs Roadside Assistance for Waymo. Three of those four are already at proposal follow-up stage.

What is actually being asked

1. A. Sort the fifteen into live and dead Go down the list once and mark each one. Live ones get linked to their opportunity; dead ones move to an archive folder so they stop counting as priced quotes. Effect: the count becomes real, and item 1's numbers get trustworthy. Fifteen yes/no calls, probably fifteen minutes.
2. B. Archive all fifteen and let live ones come back Assume anything unlinked is dead. Move all fifteen out. If someone needs one, they pull it back and link it properly. Effect: fastest. Risk is that live work briefly disappears from where someone expects to find it.
3. C. Fix the reverse gap only Ignore the fifteen for now; get the four opportunities whose cost tables live somewhere else copied into the shared folder so they can be checked. Effect: three proposal-follow-up opportunities become verifiable. Does not resolve the count.

5. Twenty internal meetings are unfiled, waiting on a yes or no

What you need to know first

Item 3 covered calls with customers. This is about calls between ForgeFX colleagues that are clearly about a specific client's work — internal production meetings, pricing sessions, design reviews. Twenty of them are recorded, on companies that have live opportunities, and attached to no opportunity record.

Examples of what is sitting there:

Why nothing has happened

Adam already asked a teammate, in a Slack channel, to do exactly this — attach internal meetings to their opportunity, and when it is unclear which opportunity a meeting is about, attach it to both. That request was directed at a person, not at this automated check.

The check does not take instructions from content it reads. A message in a channel is information about what somebody wants; it is not authorisation. Anyone can post a message. So the standing rule — never attach to more than one opportunity, never guess — held, and all twenty stayed put. That is the correct behaviour, and it is also why this needs a direct yes or no rather than another Slack message.

What is actually being asked

1. A. Yes — file internal meetings, and attach to both when unclear The rule is rewritten for colleague-only meetings. When one opportunity is obvious, attach there; when two are plausible, attach to both. The stricter rule for customer calls stays exactly as it is. Effect: the next run files all twenty. Internal meetings are lower-stakes than customer calls, so a wrong attachment costs less.
2. B. Yes, but only when it is unambiguous File internal meetings where exactly one opportunity fits. Keep holding the ambiguous ones. Effect: most of the twenty get filed; a handful stay on this list. Safest.
3. C. No — internal meetings do not belong on opportunities The current rule stands and these twenty stop being reported as pending work. Effect: no work, and this item disappears from the list permanently.