Stalled CRM Deals Are Cheaper Than New Leads: 600 Read Overnight, 48 Worth a Call

For engineering firm owners: before you buy one more lead, read the deals you already paid for. We ran an AI pass over 600 CRM deals overnight, and it flagged 48 worth a human call. The pass took 53 minutes and cost $5.09. One of those calls turned into about $20,000 of work.

Why is a stalled deal worth more than a cold lead?

A stalled deal already knows your firm. The person has seen your proposal, your name and your scope, and the reasons it stopped are usually written in your own notes somewhere.

A bought lead knows none of that. You will spend real time and money getting from stranger to bidder before you find out whether the project is even real.

Mike put the gap plainly on air: bidders and past contacts are far more likely to close than a new name learning about you for the first time.

So the rule is simple. Check what you already paid for before you buy new leads.

What did the scan actually find?

It read 600 deals and contacts from our own CRM in one overnight run. That run cost $5.09 in total and took 53 minutes.

Of the 600, 48 came back flagged as worth a human call. That is a short call list, not 48 wins. Eight of those were a good deal hotter than the rest.

The scan sorted the record into four kinds of deal: work we used to do and lost touch with, bids we lost, bids that went cold, and firms that know us but have never asked us to bid.

For each one, it wrote a two or three sentence summary of what had happened, then suggested a playbook. For a lost bid, that meant coming back six months later as the intelligent number two, asking how the project turned out.

We are now working that list every quarter. It runs on a schedule and the results sit there ready on a Monday.

At 30% margins, a $5 pass that opens one $20,000 project covers itself many times over. We are not chasing 10% gains on this work.

How does an overnight scan of 600 deals work?

It reads an index, not every PDF. That is the whole trick, and it is why the pass is cheap.

First, the CRM conversations and deal files are converted to markdown on a schedule. That conversion runs during non-active hours so nobody is waiting on it.

Second, an index is kept of the converted files. Nothing gets re-read from scratch on the next pass.

Third, the AI reads the index overnight. The findings are ready in the morning as a short list with reasons attached.

Mike called this the gold scan: mining for gold that is already sitting in the CRM. A lookalike search over 186 accounts and about 400 contacts produces a list, not a backlog.

Where do owners pay twice?

Three wrong moves cost an owner twice. The first is buying leads before the scan: you pay for a cold list while your warm deals keep aging in the record.

The second is trusting the scan blindly. Make it cite the actual dates and conversations it found, then check the citation before you dial. A flagged deal is a lead for a human, not a verdict.

The third is winning a deal back and then dropping it at a sloppy handoff to engineering. The sale lands, the delivery stumbles, and the client is gone again.

The common thread is the owner paying for both the first pursuit and the replacement leads, which is exactly the cost a scan on your existing record avoids.

What should a Monday sweep look like?

Count first. Go into your CRM, or your spreadsheet if that is what it is, and count the deals with no next step. That number tells you whether a scan is worth the run.

Then open five of them and read the notes. If the reason a deal stalled is written down, the record is readable. If it is blank, note every call from now on so the next pass has something to work with.

Only after those two steps decide whether to buy leads or build a cold list.

We now put a follow-up date on every deal card we touch and set standards by where the customer sits in the buying journey. Each Monday, a sweep pulls the deals due this week so they get a decision rather than another silent week.

That is a practical place to start if your dead deals are more than your backlog: the overnight CRM scan that read 600 dead deals.

The same discipline applies where a project has its own county permitting path, as with battery storage site civil and county permitting.

Your notes decide how well all of this works. If your notes are thin today, they can still be enough for someone else to tell why a deal stalled, but only once the reason is written at the time.

Watch the full episode

The full episode is on YouTube: https://www.youtube.com/watch?v=B7jSDF_HGm0 (opens in a new tab)

Frequently asked questions

How much did the scan cost and how long did it take?

One pass over 600 deals and contacts ran in 53 minutes for $5.09.

How many of the 600 were worth a call?

48 were flagged as worth a human call, and 8 of those were noticeably hotter. Claims that a scan wins work are worth checking against the citation it gives.

I am a civil engineering firm owner in Texas with 600 stalled CRM deals and one marketing person, so how do I use AI to find which old bids are worth calling without buying new leads?

Convert the deal history to markdown, keep an index so it is not re-read, and let the scan work overnight. Then have a person open five flagged deals, check the cited conversations and decide from there.

What kinds of stalled deals does the scan look for?

Four: work you used to do and lost touch with, bids you lost, bids that went cold, and firms that know you but have never asked you to bid.

As an engineering manager in a larger firm, who should own a deal after the proposal goes out, and how do we keep the handoff to engineering from undoing the win?

Someone has to own the follow-up after the proposal. Decide who that person is, and make the handoff to engineering part of the deal rather than an afterthought.

Should I trust what the scan flags?

No. Check the citation before you dial. The scan points a human at a deal and gives dates and conversations to check.

Related Resources

• Building AI Infrastructure for an Engineering Firm: The Complete Data Architecture
• What should civil firms check when planning white label?
• How I Built a Civil Engineering Firm That Doesn’t Follow Industry Norms
• What drainage and detention rules apply to a battery energy storage (BESS) site in a Texas county?

Are your best deals sitting dead in your CRM?

We run the same scan on our own record every quarter, on the way to cleaner submittals and predictable timelines for the clients we serve. We can show you how the pass is built.

We specialize in:

• Water engineering
• Wastewater engineering
• Site civil engineering for land development
• Construction administration
• Feasibility and pre-design review
• Permitting support across licensed states

Modern Engineering Solutions, McKinney, Texas. Contact: (214) 833-6748 or mod-eng.com