What AI reference matching is
Picking a reference for a deal is a matching problem. The prospect has an industry, a size, a set of worries and a stage in their evaluation. Your library has customers with their own industries, sizes, stories and limits. The right pairing is the customer whose experience answers the question this prospect is asking.
In most teams that pairing happens from memory. A rep asks in a chat channel, someone suggests the customer they talked to most recently, and the same few names get used again. The library might hold a better fit that nobody thought of.
AI reference matching is a way to search the whole library by describing the need instead of building a filter. You write what the prospect cares about in plain language. You get back a ranked list of references, each with a score and a short reason.
How matching works in lyynx
Describe the prospect
You type a description the way you would brief a colleague. For example: a mid-sized logistics company, replacing an older system, worried about how long the rollout takes and whether their operations team will adopt it.
Review the ranked list
lyynx returns matching references from your own library in ranked order. Each result has a score and a short reason explaining why it was suggested. The reason is the useful part. It tells you whether the match is about industry, about the problem solved, or about something in the advocate's story, so you can judge it quickly.
| Rank | Reference | Score | Why it was suggested |
|---|---|---|---|
| 1 | Sample Logistics (Sam Sample, IT Director) | 92 | Same industry and similar size. Story covers a phased rollout led by IT. |
| 2 | Test Retail Group (Casey Test, Operations Lead) | 81 | Different industry, but the story is about adoption by a warehouse operations team. |
| 3 | Fictional Foods (Riley Demo, COO) | 64 | Replaced an older system. Larger company, executive viewpoint. |
A person on your team chooses. No request is sent until they do.
A person chooses
Matching produces a shortlist and stops there. Someone on your team reads it, picks a reference or none, and starts the request in the normal way. Nothing is sent to an advocate automatically.
History and feedback
Match history is kept, so you can go back to a past search and see what was suggested. Your team can leave feedback on the results. Over time that history is a useful record of which kinds of prospects you are being asked to support and where the library is thin.
Search is still there
Matching does not replace filters. The Find a Reference search still lets you narrow by industry, size, location, product and tags, with quick filters for references who are willing to take a call or who have a testimonial. Use whichever gets you to a shortlist faster. Both draw on the same reference library.
How matching respects the advocate
The risk with any automation in a reference program is that it makes asking too easy. If software can pick a customer and fire off a request, the customer starts receiving requests nobody thought about. lyynx does not do that.
- No automatic outreach. A match is a suggestion shown to your team. The advocate hears nothing until a person decides to ask.
- Their settings still apply. The privacy level, contact method and approval status on a reference are the same whether you found it by matching or by browsing.
- Their limit still applies. If an advocate has reached the engagement limit they set, new requests route to your team instead of to them. See advocate burnout prevention.
- A better fit is a kinder ask. An advocate who is asked to speak with a prospect facing the same problem they solved has an easier, shorter conversation than one who was picked because their name came to mind.
Who uses it
- Sales reps who need a reference for a live deal and do not know the library well.
- Customer marketing handling a queue of reference requests, who want a second opinion beyond the usual names.
- Customer success managers checking whether one of their accounts keeps coming up, which is a sign it may be overused.
Practical tips
- Describe the worry, not only the profile. "Healthcare, large" finds the same references a filter would. "Healthcare, large, nervous about moving data from the old system" finds the advocate who lived through that.
- Read the reason before the score. A score tells you the order. The reason tells you whether the match makes sense for this deal.
- Look past the first result. The top match is often the reference everyone already uses. The second or third may fit nearly as well and have had fewer requests lately.
- Feed the library. Matching can only work with what is written down. Thin profiles get thin reasons. Fill in the use case summary, challenges solved and Q&A answers.
- Leave feedback. A quick note on whether a suggestion was useful makes the history worth reading later.
- Prepare the advocate after you choose. Once you have picked, send them context. The reference call prep brief template gives you a structure.
For the reasoning behind good pairings, with or without software, read match prospects to the right reference advocate every time.