Build, scrape, ask, route.
Recon sits between list-building and outreach. Everything downstream of it gets better: it goes to the companies worth engaging, and it stops opening with a guess. Here is where it plugs in.
The order is: build, scrape, ask, route
Four steps, and the middle two are the ones most stacks collapse into one.
Build the list however you already do, a data provider, a maps scrape, an export from a partner. Raw is fine. Recon doesn't need a clean list; it produces one.
Scrape everything answerable from public sources. Services pages, listings, socials. This is fast, cheap, and it does two jobs: it answers what it can, and it tells you what it can't. A fact that appears on a services page should never become a phone call. You would be paying to ask something the company already published, which wastes the dial and signals you didn't look.
Ask only what's left. That residue is the actual product of the scrape step. Sometimes it's the whole list, because the fact is never published anywhere. Sometimes it's a fraction, because two-thirds resolved from a website. Occasionally it's nothing, and you know the whole list before anyone picks up a phone.
Route on the answer. Not filter. Route. A no isn't always a discard, and this is where most of the value gets left behind.
Route, don't filter
The instinct is to treat recon as a filter: keep the yeses, bin the rest. That throws away most of what you paid for.
An answer usually implies an approach, not merely an inclusion. Some examples of the same list splitting three ways instead of two:
| Answer | Filtering | Routing |
|---|---|---|
| Need already served by a competitor | Discard | Different sequence: displacement, timed to renewal |
| Need unserved | Keep | Direct offer, lead with the gap |
| Uses a system already | Keep | Integration and migration angle |
| Runs on paper | Keep | Education angle, a different pitch entirely |
| Nobody could be reached | Discard | Nothing by phone; another channel, or nothing at all |
The list didn't get smaller. It got sorted into groups that want different things, and each group now gets a message written for it rather than an average written for nobody.
A no that tells you why is worth more than a yes that tells you nothing.
What each downstream channel should receive
Recon output isn't one artifact. Different channels need different slices of it.
Cold email
The verified fact and the words it came in, so the first line stops manufacturing relevance. There's a large difference between "I noticed you're in the auto services space" and opening with something specific that's true, which you established rather than inferred.
Cold calling
The routed segment plus the reason, and what to open with when the questions asked for it. A rep working a list where every row was confirmed a candidate is doing a different job from a rep working a list built on firmographics, and they will feel the difference within an hour.
Paid
Language and rate. If a recon pass across a few hundred companies shows a consistent way operators describe the problem, that's your ad copy, and it came from the market rather than a brainstorm. If it shows the rate at which the precondition actually holds, that's your addressable number.
The CRM
The fact, dated. Most recon findings decay slowly: whether a location has a waiting area, whether a business runs trucks, whether it operates multiple sites. Those are good for a year or more. Contract dates and vendor relationships move faster. Store the answer with the date it was established, and it stays useful across campaigns instead of being consumed by one.
What it costs, honestly
Recon isn't free and it isn't instant. It adds a step before outreach, which means a delay measured in days, and a cost above an enrichment lookup.
The arithmetic that matters isn't recon versus nothing. It's recon versus discovering the same facts through the outreach itself, where you pay in leads that are now cold twice, in a rep's calendar, and in a result you can't attribute because the commercial frame confounded every variable.
Where recon doesn't pay: a list small enough that a rep can qualify it manually in an afternoon, a question already answered by public data, or a motion that has never worked and needs a strategy answer rather than a better list.
Ask where your last handful of customers actually came from. If they all arrived through referral or a channel partner while the plan is to scale cold outreach, recon will make the cold motion sharper, but it won't make it the motion that produces your revenue. That's worth knowing before it's worth optimizing.
The habit worth keeping
Whatever you use to do it, the discipline is the same and it survives any tooling change: name the fact that makes someone a candidate, check whether it's published before you go asking, ask only what's genuinely unknown, and route on the answer instead of filtering.
Teams that do this stop running campaigns to find out whether a market exists. They already know, and the campaign is about execution, which is a much better problem to have.