The Monday report says dials are up, contacts are up and sets are up. The closers say they sat in three empty driveways last week. Both are true, and the gap between them is the difference between a KPI that flatters and a KPI that manages.
This is not a page of targets. There is no credible published benchmark for a solar floor — every figure circulating is self-reported, undefined and quoted by people who did not measure it — and we are not going to add one. It is instead a list of the numbers that usually sit on the report, what each one hides, and the number to put beside it so it stops hiding anything.
Why there is no benchmark to give you
Two reasons, and they are structural rather than a matter of nobody having bothered.
First, every rate on a floor has more than one honest denominator. Contact rate alone is four different numbers depending on whether you divide by dials, records, reachable records or right-party connects — what is a good contact rate walks through all four — and two floors quoting “contact rate” to each other are almost never quoting the same thing.
Second, the largest terms are properties of the file and the telephony, not the floor. A rate is set by how the phones were matched, how old the segment is, what share are mobiles and whether the caller ID is flagged, before a rep has said a word. A benchmark across floors with different files is a benchmark of their vendors.
We build a canvassing app rather than a dialer, and the one funnel we publish with a methodology is the door: 37.0% of knocks answered, 5.8% marked appointment-set, across 237,329 logged outcomes. We cite it as an example of what a sourced number looks like, not as a phone benchmark — a knock has no wrong-party term and no spam label.
The numbers that flatter, and what to put beside them
| On the report | What it hides | Put beside it |
|---|---|---|
| Dials per rep-hour | An exhausted file makes every dial a fast no-answer. The number rises as the list dies | Connect rate by segment age |
| Contacts per hour | Wrong-party conversations count. A file matched to whoever is associated with the address inflates this while converting at nothing | Right-party share of contacts |
| Sets per rep | Unqualified sets are free to make. A setter paid on this number books the spouse who cannot sign for a time nobody agreed to | Held rate per set, by setter |
| Contact rate | The denominator. Per reachable record reads highest and excludes the data problem; per dial reads lowest and punishes attempt policy | The written definition, and wrong-party as its own code |
| Talk time | Long calls with the wrong household or with a renter. Measures effort, not outcome | Set rate off right-party conversations |
| Cost per lead or per record | The invoice. Says nothing about how many rang, how many were the owner, or how many held | Cost per held appointment, by source |
| Appointments this month | Blends sources. One file booking well and sitting badly hides inside a total for months | The same figure split by source and segment age |
The pattern in the third column: every flattering number is rescued by a denominator that cannot be gamed, or by splitting it along the line that would otherwise hide the problem. That is the whole method. A number on its own is a claim; a number beside its denominator is a measurement.
The chain a floor should actually watch
Five numbers, in order, each one a fraction of the one before it. This is the same chain outbound dialer setup ends on, with one addition at the end that turns it into money.
- Connect rate — dials a human answered ÷ all dial attempts. Belongs to telephony and line type. Falls on every segment at once when a caller ID is flagged; falls on old segments only when the file is exhausted.
- Right-party rate — connects with the intended owner ÷ connects. Belongs to the data. The most useful split a floor can make, because it is the only one that separates the vendor from the reps.
- Set rate — bookings ÷ right-party conversations. Belongs to the people. The one number on this list that coaching moves, and it is third.
- Held rate — appointments the closer ran ÷ sets. Belongs to qualification and confirmation. The stage where money leaves quietly, because nobody counts what did not occur — solar appointment setting covers the habits that move it.
- Cost per held appointment, by source — everything the source cost, fully loaded, ÷ held appointments it produced. The only number on the floor that connects to the P&L, and the one that decides which file gets reordered.
Multiply the first four together and take the reciprocal and you have dials per held appointment, which is the staffing number — how many dials per appointment has the calculator. Recompute the chain per source and per segment age, never blended, or the report will tell you the reps got worse the week the list ran out.
The numbers with consequences beyond efficiency
Three figures do not belong in the chain because a bad reading is not a performance problem — it is an exposure. They are watched daily and they have a ceiling, not a target.
- Abandonment rate. Calls the dialer connected and then dropped because no agent was free. The federal Telemarketing Sales Rule’s safe harbor is three percent of calls answered by a person, measured per campaign over a 30-day period; treat any pacing change as an experiment with that ceiling as the stop condition.
- Suppression and registry hits per load. How many records each incoming file lost to the scrub. A load with zero hits did not get scrubbed. A load with an unusual share of hits came from a vendor who did not scrub either.
- Complaints and do-not-call requests per ten thousand dials. The number regulators and carriers act on, and the earliest warning that a file is reaching people who never had a relationship with you. A rising figure on a new source is a match-quality finding, not a script finding — DNC compliance for outbound solar covers the routine underneath.
The abandonment figure is the federal safe-harbor threshold as written in the rule, not a recommendation. Rules change and state law varies — confirm your dialing setup with compliance counsel.
Build the benchmark you can trust
The only benchmark worth having is your own floor, six weeks ago, on the same definition. Getting there needs three habits, none of them technical:
- Write each definition down, dated. Numerator, denominator, attempt policy, which records are excluded. A definition that drifts destroys the history that was the whole point.
- Stamp source and batch at import. A record with no source cannot be attributed later, and attribution reconstructed at close flatters whichever source the reconstructor already believes in. How to measure lead source ROI is the ledger this feeds.
- Cohort by entry, not by close. The records that arrived in week one, followed through. Dividing this month’s held appointments by this month’s dials mixes two populations and calls the result a rate.
Six weeks of that and the report answers the closers’ question directly: which source, which setter and which stage produced the empty driveways. That is what a KPI is for. The field-sales version of the same argument, with the close-rate denominators laid out, is in what is a good close rate in solar; the wider operating map this page sits inside is how to run a solar call center.
Disclosure: Scout Data is our product, and we sell the file that sets the second link in the chain. That is why this page keeps insisting right-party rate be reported on its own line — we would like it to be visible. We do not publish a benchmark for floors using our data, for the same reason we do not publish one for anyone else: it would be their floor’s number.
Frequently asked questions
What are the most important KPIs for a solar call center?
Five, in a chain: connect rate per dial, right-party rate per connect, set rate per right-party conversation, held rate per set, and cost per held appointment per source. Everything else on a dashboard is either one of those with a flattering denominator, or a compliance number that has consequences beyond efficiency.
What is a good connect rate for a solar call center?
There is no credible published figure, and we do not offer one because we do not run the dialers. Connect rate depends on line type, list age, caller-ID reputation and the hour of day before any rep is involved, so a number quoted without those is a decoration. Measure yours with a fixed definition and compare weeks to weeks.
Why do call center KPIs look good while appointments fall?
Because the report is built on the denominators that flatter. Sets per rep rises while held rate falls; contacts per hour rises while wrong-party share rises inside it; dials per hour rises because the file is exhausted and every dial is a fast no-answer. Each headline number is true and each one is hiding the number underneath it.
How often should a floor manager review KPIs?
Telephony and compliance numbers daily, because they fail fast and expensively. The four-rate chain weekly, per source and per segment age. Cost per held appointment by source monthly, because sales cycles make it meaningless on a shorter window. Anything reviewed more often than it can change is noise.