Customer Health and the next step on each deal
Use Customer Health and a clear next step on each deal.
Read the lesson
Goal: Read win rate, lead sources and health scores. Use Ask Emma to find the next best action, and check it.
You need: Deals in the pipeline for at least a month (lessons 4 and 5). Contacts → Reports and Analytics access.
Why this matters
A pipeline full of deals is only half the value. The other half is what it teaches. Which lead sources turn into jobs? Which stage loses the most deals? Which customers are drifting away?
Owners usually guess these answers. They remember the big win from a referral and forget the ten referrals that went nowhere. Numbers correct those memories. Then the owner spends money and time where it pays.
Where the numbers live
| Question | Where | Name on screen |
|---|---|---|
| What share of deals do we win? | Contacts → Reports | Win Rate |
| Where do leads come from, and which convert? | Contacts → Reports | Lead Sources |
| How many leads become customers? | Analytics → Overview | Conversion Rate |
| What is open right now? | Dashboard | Active Deals |
| Is this customer healthy? | The contact's Customer Health | Health Score |
Contacts → Reports has a date range, a "This Quarter" choice and Export. Use the same range every time you compare, such as one full quarter. Export the numbers each quarter into one sheet, so you can see the trend over a year.
Win rate, the honest way
Win rate is the share of decided deals that you won.
Win rate = Won / (Won + Lost).
Open deals are not in the formula. They are not decided yet. Check how the app's Win Rate counts in the client's account. Then use the same method every quarter, so you compare like with like.
Look at win rate by source, not just overall. The overall number is an average, and averages hide things. A 40% overall rate can hide a 60% source and a 15% source.
Health score is not a prediction
Each contact has Customer Health → Health Score. It is based on calls, tickets, surveys, invoices and recent activity. The figure shows the four bands, from Critical at 0 to 39 up to Healthy at 80 to 100.
Health Score bands
Treat it as a health score, not as churn prediction. Low scores often come from overdue invoices, negative sentiment or no recent activity. Use it to pick who to call. Do not use it to promise the owner who will leave.
Where a contact's score lives
Customer Health
Ask Emma: the AI suggests, you decide
Ask Emma is the in-app assistant. You can ask questions about the data in plain words. "Which lead source had the most won deals this quarter?" "Which open deals have had no activity for two weeks?"
Treat each answer as a lead, not a fact. Emma cannot see what happens outside the data. A vacation, a storm or a broken calendar sync can all move the numbers. Before you act or tell the owner, check the raw counts yourself.
The next best action is a decision. Emma can suggest it. You check it, and the owner approves anything that touches prices, offers or customers.
Using an Ask Emma answer safely
- Ask Emma a question about the data
- Do the raw counts match the answer?YesGo onNoUse your counts; note the gap
- Could a vacation, storm or sync problem explain it?YesWeigh it before any actionNoGo on
- Does the action touch prices, offers or customers?YesThe owner approves firstNoAct, and log it
- Bring one finding to the monthly review
Build it in ServiceAgent
- Go to
Contacts→Reports. Set the range to last quarter. - Open
Win Rate. Write the number down. - Open
Lead Sources. Write down deals and wins for each source. - Work out win rate by source by hand. Check it against the screen.
- Go to
Analytics→Overview. Write downConversion Rate. - Open three contacts with low
Health Score. Read why: invoices, tickets, activity. - Ask
Ask Emmaone question about the pipeline. Check its answer against the counts. - Write a one-paragraph summary for the owner: best source, worst source, one action.
- Check it works. Repeat the same steps next quarter, with the same range.
Where deals get stuck
Win rate tells you how many deals close. It does not tell you where the rest stall. For that, count deals by stage once a month.
Filter Deals by each stage, and write down the count and the oldest deal. A healthy pipeline has deals moving. A stuck one has a pile in one stage. A pile in Qualified means visits are booked but estimates are slow to go out. A pile in Proposal means quotes go out but nobody follows up.
Then read the lost_reason counts for the quarter. Price, timing and no response each point at a different fix. Price may need a better-option offer. Timing may need a longer follow-up with a dated task. No response usually means the follow-up stopped too soon.
Bring one stuck stage and one loss reason to the monthly review. Not ten. One fix a month is a pace the owner can keep.
Size changes the reading
An owner-operator has small numbers. Ten deals a quarter can swing on one deal. Look at whole years, and do not chase small changes.
A fleet has enough deals for quarterly patterns. Compare by source and by estimator.
A multi-location business should compare locations. Use location tags or separate exports. One location's weak win rate is often one fixable habit.
Worked example: Cedar HVAC
Cedar HVAC has 9 techs on Growth. Last quarter it decided 45 replacement deals: 18 Won and 27 Lost.
Overall win rate: 18 / (18 + 27) = 18 / 45 = 40%.
Cedar HVAC's decided deals last quarter
- Won18
- Lost27
Decided deals45
By source:
| Source | Won | Lost | Win rate |
|---|---|---|---|
| Past customers | 9 | 6 | 9 / 15 = 60% |
| 7 | 13 | 7 / 20 = 35% | |
| Directory | 2 | 8 | 2 / 10 = 20% |
| Total | 18 | 27 | 18 / 45 = 40% |
The lost_reason field shows 6 of the 8 directory losses were "Price". Directory leads shop on price.
Cedar HVAC win rate by source
Say Cedar's average replacement is $10,000. Past customers brought 9 x $10,000 = $90,000. The directory brought 2 x $10,000 = $20,000.
Your one action for the owner: move effort toward past customers, such as the maintenance plan and a replacement reminder. Keep the directory, but stop discounting for it.
Common mistakes
- Counting open deals in win rate. Fix: Won / (Won + Lost).
- Comparing different date ranges. Fix: the same range every time.
- Reading health score as churn prediction. Fix: use it to pick calls, not to forecast losses.
- Acting on an AI answer without checking. Fix: check the counts first.
- Chasing tiny samples. Fix: for small businesses, look at longer periods.
Do this now
Take one quarter of real or practice deals. Work out overall and per-source win rates by hand. Ask Ask Emma the same question. Write down where its answer matched and where it did not.
Practice the conversation
- Operator Last quarter you won 40% of your replacement quotes.
- Owner That sounds low.
- Operator It's the average. Past customers closed at 60%. The directory closed at 20%.
- Owner Huh. I pay that directory every month.
- Operator And most of those losses were price. They're shoppers.
- Owner So I drop it?
- Operator Not yet. Stop discounting for it, and put more into past customers.
- Owner Can the AI tell me this every month?
- Operator Emma can help me find it. I check the numbers before you see them.
Quick check
- A client won 12 deals, lost 18, and has 10 still open. What is the win rate?
- 30%
- 40%
- 60%
- 25%
- A contact's `Health Score` is 45. What does that mean?
- The customer will cancel within 45 days
- The customer is Healthy
- The band is At Risk, so it is worth a call
- The score is broken
- `Ask Emma` says Wednesday jobs fell 50%. You know one of two crews was on vacation. What do you do?
- Send a Wednesday discount today
- Report a 50% drop to the owner
- Stop using `Ask Emma`
- Check the counts and weigh the vacation before any action
ServiceAgent UniversityLearn to run a service business on autopilot. Free courses, Blueprints and certification for Service Operators.