Category: Tech Myths

  • The Hidden Cost of Running Your Business on a Spreadsheet

    The Hidden Cost of Running Your Business on a Spreadsheet

    A roofer in Mesquite is sitting in his truck at 6 p.m., scrolling his pipeline sheet, when he hits a row from three weeks ago. Homeowner in Sunnyvale. Hail damage. Wanted a quote “this week.” He never called back. He meant to. The row just scrolled off the screen and out of his head, and now the job is somebody else’s. That homeowner already signed with the next guy who actually picked up the phone.

    That row was a real job. A spreadsheet just doesn’t care.

    A spreadsheet is a great place to store a list. It is a terrible place to run a service business, because it can’t remind you, can’t trigger a follow-up, and can’t survive the day the one person who built it walks out the door. It feels free. The cost shows up later as deals you forgot to chase. Let me show you exactly where sheets break, when one is genuinely the right call, and what to move to when you’ve outgrown it.

    I’m not anti-spreadsheet. I keep a dozen of them. But I’ve opened too many service businesses running their entire pipeline on one to pretend it’s a system.

    Why “just a sheet” feels right at first

    It’s free, it’s open, and you already know how to use it. For your first twenty customers, it genuinely works. The problem isn’t the sheet. The problem is that the sheet does exactly what it did on day one while your business gets more complicated around it every month. The gap between what you need and what it does widens quietly. You usually notice it as a “how did I forget to call that lady back” moment, in the truck, at 6 p.m.

    Where the spreadsheet actually breaks

    No follow-up triggers. This is the big one. A CRM can say “it’s been 14 days since you touched this lead, call them.” A sheet sits there silent while the lead goes cold and signs with your competitor. Every follow-up depends on you remembering to scroll, and on your worst week, you won’t. The jobs you lose are the rows you didn’t look at.

    Version chaos. The minute a second person touches the file, you have a problem. Your office manager is in “Pipeline_FINAL_v3,” you emailed a copy to your lead tech last Tuesday, and now two people have updated two different versions of the truth. Which one is real? Nobody knows. A real system has one record everyone sees at the same time.

    Reporting you can’t trust. You want to know where your best jobs come from, your close rate, how many quotes are stuck waiting on a callback. In a sheet that’s a manual afternoon with a pivot table, and it’s already stale by the time you finish. A CRM answers that on a dashboard, live, while you’re standing in someone’s driveway.

    And then there’s the one that should actually scare you. Key-person risk. The color-coding, the hidden columns, the formula buried in cell H47 that quietly runs the whole thing: all of it lives in one person’s head. The day they quit, or just take a week in Galveston, your business can’t read its own pipeline. A system only one human understands isn’t an asset. It’s a liability wearing a green checkmark. I’ve watched a $2M company go blind for nine days because the woman who built the sheet had her gallbladder out.

    That whole list has a name. I call it the spreadsheet tax: the slow, invisible bleed of deals, hours, and sanity you pay every single month for the privilege of “free.”

    A spreadsheet stores what already happened. A CRM tells you what to do next. That one difference is the entire reason CRMs exist.

    Put a real number on the spreadsheet tax

    Generalities don’t move anybody, so let’s put a real number on it.

    Meet Andre. Andre runs a 9-tech electrical company in Garland. He’s an illustrative composite, not a real client, but every piece of him is stitched from real DFW service businesses I’ve sat across from.

    Here’s the math on Andre’s spreadsheet tax. Stated assumptions, so you can argue with them:

    • He logs about 80 leads a month in his sheet (call-ins, website forms, referrals).
    • Be generous and say only 15% slip through with no real follow-up. That’s the row that scrolled away. 12 leads a month.
    • His team closes about 1 in 4 of the leads they actually work. So 3 of those 12 lost leads would have become jobs.
    • His average job ticket is $1,800 (panel upgrades, rewires, the bigger stuff, not service calls).

    So: 3 lost jobs a month times $1,800 is $5,400 a month. Times 12 is about $65,000 a year, gone. Not to a competitor who’s better than Andre. Gone to a competitor who simply called back.

    Your numbers will be different. Maybe you log 40 leads, maybe your ticket is $600, maybe your slip rate is 8% because you’re sharp. Run it with your own figures. Even if you halve every number I used, that’s still 32 grand a year falling through a sheet, and 32 grand a year is a real hire or a second truck.

    Skip this if

    Let me be fair to the sheet, because it isn’t always the enemy.

    Stay on the spreadsheet if you’re tracking under ~40 contacts and you personally touch every one of them. Stay on it if it’s a one-time list and not a living pipeline: a material order, a job-site punch list, a crew schedule for one build. Stay on it if one person owns it, nobody else edits it live, and there’s no follow-up timing that hurts you when it slips.

    If that’s you, ignore the rest of this post and don’t let anyone sell you a $300-a-month platform to manage 30 names. The myth was never that spreadsheets are bad. The myth is that they scale. They don’t.

    What to graduate to

    The move isn’t “buy the biggest CRM in the catalog.” It’s matching the tool to where you actually are.

    Light and cheap first. A simple CRM like HubSpot’s free tier, or a structured tool like Airtable or Monday. You keep the spreadsheet feel and you finally get reminders, one shared record, and basic automation. For a lot of DFW service shops, that alone kills most of the spreadsheet tax overnight.

    Then, when your follow-up has real rules (speed-to-lead on hail season, multi-step nurtures, referral tracking back to the agent who sent the job), a generic template starts fighting you. That’s when you want a system wired to your process, not somebody else’s. That’s the part we build.

    The signal you’ve outgrown the sheet is dead simple. You forgot a follow-up that mattered, or two people edited two versions in one week. When that happens, it’s time to move. We handle that graduation inside the HTS Operating System tier, and you can see how one looked in a recent HTS engagement.

    Ready to find out if you’ve outgrown the sheet?

    If a forgotten row has already cost you a real job, or your whole pipeline lives in one person’s head, that’s your signal. Book a discovery call. We’ll look at your actual sheet, run the spreadsheet tax on your real numbers, and tell you straight whether you need to move yet. You walk away with the math either way.

    Book a Discovery Call →

  • AI Will Replace Loan Officers: What Actually Changes in 2026

    AI Will Replace Loan Officers: What Actually Changes in 2026

    It’s 9:47 on a Tuesday night. Janelle is on the couch, half-watching TV, when the post slides up her feed. A guy with a headset and a ring light, 40,000 likes: “AI just killed the loan officer. Adapt or die.” She double-taps without thinking, then sits there for a second with that low hum in her stomach. Twelve years closing loans in Fort Worth, and a stranger with a microphone just told her the job is over.

    He’s wrong. But the half-truth buried in his clip is the part Janelle needs to take seriously. (Janelle is a composite, stitched from a dozen real DFW loan officers. She isn’t one client.)

    Here’s the verdict up front: AI is not coming for the loan officer. It’s coming for the loan officer who runs the job out of an inbox and a notepad. Those are two different people, and the distance between them is what this whole post is about.

    The frame that’s wrong

    “AI versus loan officers” pictures a borrower applying by chatbot, getting underwritten in 90 seconds, and closing without ever hearing a human voice. That isn’t the trend. Nobody is signing the largest debt of their life with a bot.

    The real trend is quieter and more dangerous: loan officers using AI to do more of the work themselves, faster, so a smaller number of LOs produce the volume that used to take a team. Same borrowers. Same files. Fewer people closing them.

    What the data actually says

    A few things are demonstrably true in the current mortgage market:

    • Headcount in lending has compressed. MBA forecasts and lender employment data show the workforce shrinking since the 2021 peak. The rebound in volume is not bringing the same headcount back.
    • Top producers are pulling away. MBA and STRATMOR surveys show top-quartile LOs taking share from the middle of the pack, and the tool stack is part of why.
    • AI adoption in mortgage ops is lopsided. Most lenders deployed AI for document reading, fraud detection, and underwriting support years ago. Borrower-facing AI in retail mortgage is still rare. That gap is what 2026 starts to close.

    The bet was never that AI replaces the LO. The bet is that one top LO with an AI-run operation outproduces three LOs without one.

    What actually changes in 2026

    The first response stops being a human job

    A borrower applies at 9:47 at night, same hour Janelle saw that post. By 9:48 they have an email confirming it landed, a text with the next step, and a link to grab time on her calendar. That used to need an after-hours assistant. Now it’s a workflow.

    Borrowers don’t feel “automated at” when the reply is fast, polite, and hands them to a real person inside one business day. It’s the most-requested build in the HTS Operating System tier, and it’s the cheapest piece of the whole stack.

    The pre-call brief is the lever almost nobody is pulling

    This is the one that matters most, and barely anyone uses it. Before Janelle returns the call, a workflow hands her a one-page brief: stated purpose, property type, ballpark income, prior credit bumps, who referred them, a suggested opening line. She reads it in 90 seconds. The call goes faster because she isn’t asking for things the borrower already typed in.

    Now put a number on it. Say the brief saves 5 to 10 minutes a call. Janelle takes 30 calls in a normal week. Call it 7 minutes saved, times 30, times roughly 48 working weeks. That’s about 168 hours a year handed back to her.

    What is an hour of a producing LO worth? If Janelle’s pipeline lets her originate even $200 an hour in real production value, 168 hours is roughly $33,000 of capacity she’s leaving on the table every year by reading cold into every call. Your numbers will be different. Your call volume, your close rate, your value per hour all move the total. But the shape holds: this is real money, and it’s hiding in a step most LOs treat as too small to fix.

    That gap between Janelle-with-a-brief and Janelle-reading-cold is the augmentation gap, and it widens every quarter she waits.

    The follow-up stops being the thing that falls through

    Here’s the honest reason most LOs don’t follow up at scale: follow-up is grinding work that doesn’t feel like producing. Eight texts a day. Thirty emails a week. A birthday card she means to send and never does.

    In 2026 the calendar is a workflow, the texts run off approved templates, the birthday card prints itself. Janelle approves and personalizes. The system handles delivery. The hours that frees up go to the conversations that actually close: the partner lunch, the closing-gift drop-off, the call that turns one realtor into a steady three loans a month.

    Compliance review gets faster, not lighter

    A common bad take: AI will “auto-approve” compliance. It won’t, and you shouldn’t want it to. What AI is genuinely good at is flagging items for a human: a missing disclosure, an ambiguous sales line, language a compliance officer would rather rewrite. The reviewer still reviews every word. They just stop reading 100 clean emails to find the three that need them.

    And the part the LinkedIn guy gets exactly backwards: the job gets more human, not less

    This is where the clean story breaks. Strip out the first response, the brief, the follow-up busywork, and the compliance triage, and you don’t get a smaller loan officer. You get a bigger one.

    Borrowers still want a human voice on the largest debt of their life. Realtors still refer to people they trust, not chatbots. The hard files, the jumbo files, the self-employed borrower with four income streams, those still need an LO who can think. What changes is that Janelle becomes the operator of an automated business instead of the manual labor inside it. Less salesperson with a Rolodex. More small-business owner running a stack. That’s not the job dying. That’s the job finally growing up.

    Skip this if

    This isn’t for you if you’re brand new and still closing one loan a quarter. You don’t have enough volume yet for any of this to pay off, and you’d be automating a process you haven’t even run by hand long enough to understand. Go close ten loans the hard way first. Come back when the follow-up starts slipping. It will, and that’s the right time.

    What this is really about

    This post is not telling you to fire your assistant, run a one-person shop, or trust a bot to talk to your borrowers. It’s telling you the gap between an LO running an integrated stack and an LO running on email and a notepad is widening fast.

    Three years ago, the augmentation gap was a productivity edge, a nice-to-have. In 2026 it’s a survival question for everyone stuck in the middle of the producer distribution. Janelle isn’t going to lose her job to a chatbot. She’s going to lose it, one referral at a time, to the LO across town who closed the gap first.

    What to actually do this quarter

    Three moves, in order:

    1. Audit your CRM. If your data is junk, no AI built on top of it will help you. (Related: the 5 fields most LOs ignore.)
    2. Install the starter AI stack. Claude, Perplexity, an AI notetaker. Nothing exotic. (Related: 3 AI tools you can use this week.)
    3. Build one real automation. Pick the bottleneck that costs you the most hours and build the workflow around that one. Just one. Finish it before you start a second.

    How HTS does this

    Same playbook in a 50-LO shop or a 1-LO shop, scaled differently. The first 60 days: CRM audit, starter stack installed, one automation the LO will actually use. After that we layer in pre-call briefs, referral-partner workflows, and past-client outreach. A recent HTS build followed the same arc, and the piece everyone used most was the simplest one we shipped.

    Ready to find out where you actually stand?

    If your honest answer is “I don’t really know what my stack does, and I don’t know what’s missing,” that is the entire point of a discovery call. It’s 30 minutes. You leave with a clear next step, hired or not.

    Book a Discovery Call →


    Compliance note: This post discusses general industry trends in mortgage lending. It doesn’t offer legal, tax, regulatory, or financial advice. AI workflows that touch borrower data, marketing communications, or compliance review require sign-off from your shop’s compliance officer. Trends referenced are based on publicly reported industry data (MBA, STRATMOR, lender employment reports); verify current figures with primary sources before citing externally.

    HTS does not provide loan advice, legal advice, or financial advice. Mortgage and finance professionals are responsible for their own regulatory compliance.