5 Questions Every Contractor Asks That No Dashboard Will Ever Answer

74% of residential contractors see AI as key to efficiency, per ServiceTitan's 2026 State of AI in the Trades report. Only about 25% actually use it. Most are still staring at dashboards.

Dashboards are good at one thing. They tell you what already happened, in the shape someone predicted you would ask about it. The questions contractors actually ask at 6 AM with a cup of coffee are a different shape entirely.

This post is the pillar on the "dashboards vs chat" split. Five real questions. Why dashboards can't answer them. And what those same questions look like when you text an AI brain instead.

1. "Why did we miss revenue last week?"

This is the first question every $1M-$10M home service owner asks on Monday. A dashboard shows you that revenue was down 12%. It does not tell you why.

To answer why, you have to cross-reference jobs, tickets, techs, call volume, conversion rate, average ticket, and membership sales. All at the same time. All compared against the prior week, the prior month, and the same week last year.

A dashboard is a fixed shape. The "why" question is a moving target. By the time a BI developer builds the dashboard that answers it, the question has changed.

ServiceTitan users confirm this gap. One Capterra reviewer put it directly: "My least favorite thing about ServiceTitan is the reports." Another called the tool "too big to where my people are scared to dive in and learn." The data is there. The answer is not.

"My least favorite thing about ServiceTitan is the reports."

• ServiceTitan customer, Capterra review

Text Sully: "Why did revenue drop last week vs the week before?"

Sully pulls job-level revenue, compares against the prior 4 weeks, flags that two of your top techs were out Monday to Wednesday, that your call conversion dropped from 62% to 48% on Tuesday, and that three $8K quote jobs from the prior week closed this week instead. One answer, 30 seconds.

The infrastructure that makes this work is the same architecture behind how to build an AI agent for home services: a unified data layer that can reach across jobs, calls, quotes, and techs in one query.

2. "Which of my ghosted quotes are still warm?"

Every shop has 200-800 unbooked estimates sitting in the CRM. A dashboard can show you the count. It cannot tell you which ones are worth calling today.

To answer the real question, you need the quote amount, days since sent, whether the customer has read follow-up emails, whether they got competing bids based on reply patterns, whether they are a repeat customer, and whether the job type is seasonal.

A dashboard lists them. A good AI brain ranks them by close probability and writes the follow-up text.

Pete & Gabi's 2026 reactivation research shows existing customers close at 60-70% vs 5% for new prospects, a 12-14x gap. Ghosted quotes are the hottest segment of existing customers. Nobody gets to them because pulling the list is a project.

Text Sully: "Which quotes from the last 60 days are still live, ranked by close probability? Draft a follow-up text for the top 10."

Sully returns a ranked list with quote amount, days since sent, last customer activity, and a draft SMS tuned to each lead's context.

This motion sits next to AI customer reactivation for contractors, which covers the full reactivation playbook beyond just live quotes.

3. "What did the customer actually say on that call?"

A dashboard shows you call volume, average duration, and maybe conversion rate. It does not show you that the customer mentioned a leak in the master bath, or that your CSR said "we're booked until next Tuesday" when you actually had a Thursday opening.

Call data lives in call recordings, transcripts, and rep notes. None of that fits inside a dashboard tile. And if you are running Rilla Voice or CallRail, those transcripts are a separate system the dashboard cannot reach into.

This is the single biggest blind spot in every BI stack. The conversation is where the deal is won or lost, and the dashboard is blind to it.

ServiceTitan's 2026 report shows customer service and communication is a top-four AI use area, with 39% of AI-using contractors applying it there. The reason is that transcripts are text, and text is exactly what AI is good at searching.

Text Sully: "Show me every call in the last 7 days where the customer mentioned a second bathroom, and pull any that didn't get booked."

Sully reads call transcripts, extracts the mentions, cross-references against the schedule, and surfaces three missed-opportunity calls with the transcript excerpt and the CSR who took the call.

This is the same capability that powers an AI missed-call follow-up agent for contractors. Conversation data is structured data if the extractor is good enough.

4. "Is marketing actually paying off this month?"

Every owner wants a single number: revenue-per-ad-dollar by source, this month, actual dollars booked, not leads generated.

That number requires Google Ads spend, Meta spend, LSA spend, Angi invoices, CRM lead source data, job status, invoice data, and a way to attribute each closed job back to the ad that produced it. Seven sources minimum.

Power BI can build this dashboard. A specialist will build it. It will take six weeks and break the first time Google changes a report field.

The Jobber 2026 Home Service Economic Report notes that 75% of service businesses expect revenue to rise in 2026, and margin pressure is real. Owners need attribution to know where to double down. A dashboard that breaks when an API changes does not cut it.

Text Sully: "Which ad channel produced the most profitable jobs this month? Factor in labor and parts, not just revenue."

Sully pulls spend by source, joins against closed jobs, subtracts labor and materials cost, and returns a ranked list with profit per dollar of ad spend.

The tooling side of this is covered in AI lead qualification agents for home services, which handles the ingestion of leads before they hit the attribution model.

5. "Who on my team is actually at risk of quitting?"

No dashboard answers this. Ever.

To get close, you would need to cross tech schedules, job pay, customer complaint patterns, PTO requests, missed shifts, conversation tone in internal texts, and benchmarks against peers. It is a human question with data trails scattered across six systems.

Yet it is the most expensive question in a home service business. Losing a senior tech costs $25,000-$50,000 in recruiting, ramp, and lost revenue per multiple industry estimates. Owners know it. Dashboards ignore it.

Tommy Mello runs A1 Garage, a $200M+ home service business, on a dashboard-plus-gut hybrid. He uses Rilla Voice to listen to calls and scorecard techs. That is not a dashboard. That is conversation plus judgment, and it is what the dashboard-first BI stack is missing.

Text Sully: "Which techs have seen revenue drop, more callbacks, or missed more shifts in the last 60 days?"

Sully cross-references tech-level revenue, callback rates, attendance, and surface-level complaint signals in call transcripts. Returns a ranked list with the supporting data for each flagged tech.

The sentiment and behavioral signals feeding this are the same ones that drive AI lead qualification on the customer side. Same extractor, different subject.

The common thread across all five questions

Look at the five questions above. They have three things in common.

They are cross-source. Every one of them needs at least 2-3 data sources to answer well. Revenue plus tech schedules plus call conversion. Quotes plus customer history plus reply patterns. Call transcripts plus schedule plus CSR notes. Ad spend plus CRM leads plus closed jobs plus costs. Tech data plus conversation sentiment plus attendance plus peer benchmarks.

They are ad-hoc. Nobody asks these exact questions the same way every week. The "why is revenue down" question shifts to "what about last quarter" to "is it the new pricing" to "which service line is the drag." Dashboards are built for questions you ask every week. Chat is built for questions you ask once.

They have a "why" at the core, not just a "what." Dashboards answer "what happened." A well-connected AI brain answers "why did it happen and what do I do about it." The first is a lookup. The second is reasoning across the data.

LocaliQ's 2025 benchmarks put the average home services CPL at $90.92, with HVAC non-branded leads at $149 and plumbing non-branded leads at $167. Those CAC numbers put a floor on how much every "why" question is worth. A shop that misses "why is conversion dropping" for two months is burning five-figure ad spend on leads that were not going to close.

Why AI chat answers these differently

The technical reason chat answers what dashboards cannot comes down to three design choices.

Unified data layer. A good chat AI connects to every relevant source (CRM, accounting, email, calls, ads) and normalizes into a canonical model. The chat query runs across the whole graph in one pass, not one dashboard at a time.

Conversation data is first-class. Call transcripts, customer SMS threads, and internal team messages are text. Large language models read text natively. A dashboard can display a transcript, but it cannot reason across 500 transcripts.

The query shape is the question. A user types "why did revenue drop" and the AI picks the right data sources, joins them, and reasons about the answer. A dashboard user has to know which dashboard to open and which filter to apply before they see anything useful.

Tommy Mello runs A1 Garage on a combination of ServiceTitan reporting plus Rilla Voice transcripts plus direct coaching. His public playbook emphasizes using data to inform decisions, not using dashboards as a replacement for judgment. The dashboard is the floor. The conversation is the signal.

The pattern is not "dashboards are bad"

Power BI is genuinely good at what Power BI does. Metabase ships fast for SQL-literate teams. Sigma is excellent for spreadsheet-native analysts. Looker Studio at $0 is unbeatable for a quick revenue-by-month chart.

These tools all solve "show me the known shape of a known question." Contractors live in the unknown shape. The questions are ad hoc, cross-source, and usually asked out loud, not in a filter panel.

The shift is not "dashboard bad, chat good." The shift is that the home service shops winning in 2026 will run both. A dashboard for the three KPIs you check every morning. An AI brain for every other question that comes up during the day.

Where Sully fits

Sully is ad-hoc chat on your connected data. Connect your CRM, accounting, email, calls, and ads. Text it the question. It answers in 30 seconds.

Not a dashboard replacement. A different shape of answer, for the 80% of questions dashboards were never designed for. The morning brief, the ghosted-quote triage, the call-transcript search, the attribution question, the tech-risk question. Five questions, one interface, no filter panels.

Every $1M-$10M shop already has the data. It is just trapped in dashboards that answer the wrong question.

Frequently Asked Questions

Why can't dashboards answer "why did revenue drop"?

A dashboard is a fixed shape built against anticipated questions. The "why" question is a moving target that requires cross-referencing jobs, tickets, techs, call volume, conversion rate, average ticket, and membership sales at once. By the time a BI developer builds the dashboard that answers it, the question has changed.

How many data sources do contractor "why" questions need?

At least 2 to 3 per question. Revenue-miss diagnosis needs jobs plus tech schedules plus call conversion. Attribution needs ad spend plus CRM leads plus closed jobs plus costs. Most $1M to $10M shops run 4 to 7 disconnected systems, which breaks any single dashboard trying to answer cross-source questions.

What percentage of contractors use AI?

Only about 25 percent currently use AI per ServiceTitan's 2026 State of AI in the Trades report, even though 74 percent of residential contractors see AI as key to efficiency. The gap is integration complexity and unclear ROI.

Can a dashboard tell me which quotes to follow up on?

No. A dashboard lists unbooked quotes. It cannot rank them by close probability using quote amount, days since sent, reply patterns, competing-bid signals, and seasonal context. Existing customers close at 60 to 70 percent versus 5 percent for new prospects, so the ranking is where the money is.

What is the cheapest way to replace a contractor dashboard?

A conversational AI layer on your existing data. Looker Studio at $0 is unbeatable for a simple revenue chart, but the ad-hoc and cross-source questions contractors actually ask live outside any dashboard. Chat answers those without requiring a BI build.

How much does a missed "why" question cost?

LocaliQ's 2025 benchmarks put average home services CPL at $90.92, HVAC non-branded leads at $149, and plumbing non-branded leads at $167. A shop that misses "why is conversion dropping" for two months burns five figures on leads that were never going to close.

Sources

• ServiceTitan 2026 State of AI in the Trades
• ServiceTitan press release on 74% AI sentiment
• Jobber 2026 Home Service Economic Report
• ServiceTitan Capterra reviews
• Pete & Gabi reactivation research
• Owned and Operated on Tommy Mello's $200M build