Network path mapping
Use this skill when there is a list of firms worth working with and no warm way in, or when outbound is the only channel running and the network that could replace it has never been mapped. Produces a path per target account: who in the network can reach it, how strong that connection is, what to ask them for, and the message that makes the ask easy to say yes to. Also use for "who do I know at [company]", "map my network", "find a warm intro", "who can introduce me".
Bailey DarlingCo-Founder, GiaApplies when a target list exists and cold outreach is the only plan. Produces a ranked set of warm paths with a drafted ask for each.
The play
- Build the target list first, 20 to 50 firms, before looking at the network. Starting from the network produces a list of whoever is easiest to reach, which is a different and much worse list.
- Export the connections list before anything else. Nobody can recall their own network from memory, and the export is the only way to see it whole. Requesting the archive and turning the CSV into the inventory is in references/network-inventory.md, along with the other sources worth pulling: former colleagues, former clients, vendors, peer firms that refer out, and the people who went quiet two years ago. Most agency owners have four to five hundred usable relationships and can name forty.
- Match the network against the target list. For each target, who in the inventory can reach someone there, and through how many hops. Stop at two hops; a third-hop introduction is a cold email with extra steps.
- Score each path on connection strength, not on seniority. A strong relationship with someone junior beats a weak one with a VP, because the ask only works if the person wants to make it. The strength tiers and what each can be asked for are in references/path-strength.md.
- Branch by what the path actually is:
- Former client at a new company: the strongest path available and the most neglected. They have already bought, and the firm they joined is a new account.
- A current client who knows the target: valuable and spendable once. Never ask until the engagement is visibly going well.
- A peer agency that does adjacent work: a referral relationship, not a one-off ask. Build it as a standing channel.
- A vendor or a tool rep selling into the same market: they meet the same buyers weekly and are rarely asked.
- A dormant contact, two years quiet: re-establish first, ask second, never in the same message.
- Write the ask so the connector does almost nothing. One forwardable paragraph, a named reason, and an explicit permission to say no. The patterns are in references/the-ask.md. An ask that requires the connector to compose anything will sit unanswered.
- Work the list in order of path strength, a handful a week, and record every outcome. An intro that never arrives is information about the relationship, not just about the target.
- Re-run quarterly. The network changes faster than the target list: people move, and every move creates a path that did not exist last quarter.
What good looks like
- First thing I check that most people skip: where former clients work now. People who have already bought from you, sitting inside new accounts, is the single highest-converting path in any network, and almost nobody keeps a list of it.
- The mediocre version: opening the contacts list, picking the names that feel senior, and asking each for "any introductions you think make sense". It transfers all the work to the connector, and it is why most intro requests go unanswered.
- I know the mapping is right when at least a quarter of the target list has a named path, and each path has a specific reason attached.
- Second test: the ask could be forwarded without editing a word.
- Red flag: every path runs through the same two people. That is two relationships, not a network, and the finding is that the network needs widening before it needs working.
Rules
- MUST build the target list before looking at the network.
- MUST work from an exported connections list, never from memory.
- MUST stop at two hops.
- MUST score on relationship strength, not the connector's seniority.
- MUST write the ask as a forwardable paragraph the connector can send unedited.
- MUST give the connector an explicit way to decline.
- MUST record every outcome, including silence.
- NEVER ask a current client for an introduction while the engagement is wobbling.
- NEVER ask a dormant contact for anything in the first message after a long silence.
references/network-inventory.md
Building the inventory
Most people can name forty contacts and actually have four hundred. The gap between those numbers is the whole point of this exercise, and it is why the first step is an export rather than a brainstorm.
Start with the connections export
Request a copy of your own data from your professional network's privacy settings and choose the connections archive. Most services email a download link within minutes, though a full archive can take up to a day. Request it before doing anything else in this skill; everything below is faster once the file is in hand.
What arrives is a CSV, one row per connection, with roughly these columns:
| Column | Use |
|---|---|
| First name, last name | The identifier |
| Company | The match against the target list |
| Position | Whether they can reach a buyer |
| Connected on | A proxy for how the relationship started |
| Email address | Present only for some connections |
| Profile URL | For checking where they are now |
Three things to do with it before matching anything:
- Sort by company and look for any name that appears on the target list. These are the one-hop paths and they are found in about five minutes.
- Sort by connected-on date, oldest first. The oldest connections are usually former colleagues and former clients, which are the two highest-value groups. The newest are usually event and inbound contacts, which are the weakest.
- Flag anyone whose current company differs from where you knew them. That is the former-client-at-a-new-company case, and it is the best path in the file.
The export is a snapshot and it says nothing about relationship strength, which is scored by hand in references/path-strength.md. It also goes stale: people change jobs constantly, so re-export quarterly rather than maintaining the CSV by hand.
Treat the file as personal data. It holds other people's employment details and sometimes their email addresses. Keep it where client data is kept, do not upload it anywhere you would not upload a client list, and do not use the email column for bulk sending. It is for mapping paths, not for building a send list, and using it that way burns the network this skill exists to build.
Where else to pull from
In order of value, not of convenience.
| Source | Why it is worth the time |
|---|---|
| Former clients, and where they work now | Already bought from you. Highest conversion of anything here. |
| Former colleagues from every job | They know your work directly. Scattered across the market by now. |
| People who referred you work before | A proven referrer, often never asked twice |
| Vendors and tool reps in your market | They meet your buyers weekly and are almost never asked |
| Peer firms doing adjacent work | Referral partners, not competitors, when the services do not overlap |
| Investors, advisors, accelerator contacts | Wide networks, and introductions are part of what they do |
| Event and conference contacts | Weak by default, but they map to specific firms |
| The first two rows produce more usable paths than everything below them combined, and neither is fully visible in the export, which is why the export is the start and not the whole inventory. |
Fields
| Field | Example | Why |
|---|---|---|
name | ||
where now | VP Marketing, Harbour Logistics | The target-list match |
where before | Was at Northwind when we worked together | The strength of the tie |
how we know each other | Ran a six-month engagement, 2024 | Decides what can be asked |
strength | 1 to 4, see path-strength.md | The ranking |
last contact | date | Whether an ask is premature |
their network | Freight and logistics, Toronto | Where they can reach |
has referred before | yes / no | The strongest predictor there is |
where before matters more than it looks. The tie formed at the old company is the real relationship; the new company is just where it now points.
Keeping it current
Job changes are the engine. Every time someone in the inventory moves, a new path appears, and it has a window: the first three months in a role, when they are building their own vendor list and have budget nobody has spent yet.
Check for moves monthly. It takes ten minutes and it is the highest-yield maintenance in the whole system.
The first pass
Do not try to be complete. The export plus two hours on former clients and former colleagues will produce 100 to 200 scored names. Run the matching on that. The long tail can wait until the obvious paths are worked.
Completeness is the enemy here: an inventory built over three weeks is usually built instead of doing the asks.
references/path-strength.md
Scoring a path
Strength is about the relationship, not the connector's title. A tier 4 connection to a manager is worth more than a tier 1 connection to a CEO, because the ask only works if the person wants to make it.
The tiers
| Tier | What it means | What can be asked |
|---|---|---|
| 4 | Worked together closely. They would take your call today. | A direct introduction, with their endorsement |
| 3 | Worked together, or known for years, contact is occasional | An introduction, if you make it easy |
| 2 | Met a few times, friendly, no shared work | Permission to mention their name, not an intro |
| 1 | Connected online, never really spoken | Nothing. This is a cold contact wearing a warm label. |
Tier 1 is where most intro requests go to die. Treat it as cold and write to the target directly instead.
Modifiers
| Condition | Effect |
|---|---|
| Has referred you work before | Up one tier, whatever else is true |
| You have referred them work | Up one tier |
| Last contact over 18 months | Down one tier until re-established |
| You have never given them anything | Cap at tier 3, and fix that before asking |
| They have declined an ask before | Down one tier, and change what you ask for |
The "never given them anything" cap is the one worth enforcing. A network worked only in one direction stops producing within a year.
Ranking the list
Sort by tier, then by how well the target fits the ICP. A tier 4 path to a marginal account outranks a tier 2 path to a perfect one, because the tier 4 ask will actually happen and the tier 2 ask probably will not.
Work five to eight a week. More than that and the asks get templated, which the connectors notice immediately.
Two hops
One hop is someone who knows the target. Two is someone who knows someone who does, and it only works when the first hop is tier 4 and knows exactly why the introduction matters.
Three hops is a cold email with extra steps and a worse hit rate, because each handoff loses the reason. Do not build them.
Recording outcomes
Four results per ask: intro made, intro promised and not made, declined, no response.
The middle two are the useful ones. An intro promised and never made usually means the connector could not find a way to phrase it, which is a problem with the ask, not with them. Rewrite and try the pattern on the next one.
No response twice from the same person means the relationship is a tier lower than scored. Adjust it in the inventory rather than asking a third time.
references/the-ask.md
The ask
The connector should have to do almost nothing. Two sentences to you, one forward, done.
Shape
[Why you're asking them specifically, one line.]
[What you want, named target and named reason.]
[A forwardable paragraph they can send as-is.]
[An explicit out.]
Example
Hey Dana, you were at Northwind when we rebuilt their quoting flow, so you've seen how we work.
I'm trying to reach Priya Shah at Harbour Logistics. They're going through the same mess Northwind was in 2024 and I think we could help.
If you're up for it, here's something you could forward:
"Priya, meet Bailey. His team rebuilt our quoting flow at Northwind and cut turnaround from five days to a few hours. Given what you said about your intake backlog, worth twenty minutes."
And if it's not a connection you want to make, no problem at all, just say so and I'll leave it.
Four parts: why them, what you want, the forwardable paragraph, the out.
The forwardable paragraph
The part people skip, and the part that decides whether the intro happens. A connector who has to compose the introduction themselves will put it off, then feel bad, then avoid you.
Write it in their voice, not yours. Third person, short, with one concrete result in it. Three sentences at most.
The out
Always included, always explicit. Two reasons: it makes saying no cheap, which protects the relationship, and it raises the yes rate, because an ask with no escape reads as pressure.
"No problem at all if you'd rather not" costs nothing and changes how the whole message lands.
What kills an ask
- "Any introductions you think make sense?" Transfers all the work. The most common intro request and the least effective.
- No named target. Same problem, dressed up.
- No reason. "I'd love an intro to Priya" gives the connector nothing to say in the forward.
- A pitch attached. The ask is for an introduction, not a chance to sell the connector.
- Asking a dormant contact cold. Re-establish first, in a message with no ask in it at all, then ask weeks later.
- Asking a current client mid-wobble. Wait until the work is visibly going well. An intro request lands very differently when they are unhappy.
After the intro
Reply within a day, cc the connector on the first message, then drop them from the thread so they are not copied on a negotiation.
Tell them what happened either way. The connectors who hear back are the ones who introduce you again, and that follow-up is the cheapest relationship maintenance available.
Giving first
The inventory caps anyone you have never helped at tier 3 for a reason. Before a quarter of asks, make a handful of introductions for other people, with no expectation attached. Networks that only ever receive stop producing, and it takes about a year to notice.