Contact graph

[ramanshah]

As sales continues, I'm struck by how much of my hustling resembles long chains of professional introductions. Coding the channels by degree of introduction is becoming more unhelpful as meetings are starting to materialize with fourth-degree connections that might convert into meetings with fifth-degree connections.

I often meet people who are tremendously valuable as sources of new leads but get coded in the CRM as "sales graveyard" along with duds. The data is missing something.

Much better would be to organize contacts through a social graph. Either I meet strangers through a few channels (existing family, existing friends, professional meetups, hobbies, etc), or I'm introduced to them via existing contacts.

Having a (directed) graph representation of contacts in the CRM, linked hygienically to the relational sales database, would give me insight regarding my most valuable relationships so I can make sure to keep them warm and show my appreciation. (Note that an org-contact pair is still the starting point for a sales process; the structure from that point is still well modeled as relational.)

Analytic insight I'd want from such a graph:

• Who in my world has directly generated the most sales prospects? The biggest subgraph of sales prospects? (Note that my existing channel coding screws this up completely. Who cares if a superstar in my referral graph is a second- or third-degree contact?)
• Who in my world gives me direct sales prospects with the best conversion rates? Subgraphs with the best conversion rates?

[ramanshah]

Brainstorming options:

• Have a graph database and a relational (e.g., SQLite) database in a single app with some coherence logic to keep them consistent.
  • Use a simple native-Python graph database like cog.
  • Use a more industrial-grade graph database like neo4j.
• Model the contact graph relationally, like this.

In all cases, there are questions about how to harness the bureaucracy of the database for good: How do I model the algebraic structure of how I know a person? (Family | OldFriend | ProfessionalMeeting | ColdEmail | ... | IntroducedBy Person)? How do I maintain integrity of a sales-pipeline database if the person-graph can be mutated? (I could make it so that one can only add people to the person-graph: introductions are, after all, irreversible. But what happens when there's a data-entry issue?)

[ramanshah]

I'd like to finish How To Be A Power Connector before getting too involved in such a graph implementation, as perhaps a different perspective will be the most useful. Having minimal duplication will help me maintain and use any database produced.

[ramanshah]

Almost finished with How To Be A Power Connector. It has changed my attitude about this. My sense from this book is that an internal social graph is indeed a vital asset but that it isn't well-coupled to sales. Instead, it's well-coupled to contact management and relationship (power circle) management. Namely, some people I know are power connectors and some aren't, and deepening relationships with power connectors ought to be a priority. Knowing who's a power connector depends on some graph-based insight. Here are some things I'd want from a social graph:

• To have both people and institutions on an equal footing. If I met someone, I was either introduced to them by another person or met them via an institution.
• To be able to update it periodically (say one or a few times a year) and ergonomically with new contacts. Munging LinkedIn, Facebook, and my phone contacts in some automated way would cover a lot of the cases.
• To be able to review the graph for editing my "power circles" and investing/divesting time in particular institutions.

An information system that has this ought to deduplicate information across some other use-cases:

• Syncing contacts across phones and serializing contacts in case of physical phone loss.
• Reviewing my power circles to keep the relationships appropriately warm and to search for gaps in my power circles to strategically remediate.
• Refreshing my power circles periodically as people come and go.

The idea, then, is that "customer relationship" management might not be the best use-case for an open tool. In fact, with a bit more business experience, out-of-the-box tools like nocrm.io seem to fit nicely with the truly actionable parts of sales: asynchronous prospect generation when doing high-volume outbound sales and then follow-up management for open sales.

[ramanshah]

Looking at the mission statement and the other issue in this repo, there's one thing (beyond the portfolio piece and open-source motivations, which other projects could reasonably satisfy) that I'm not sure that nocrm.io would satisfy:

• I want flexible analytic access to my own sales funnel to monitor sales and marketing efforts, understand what kinds of prospects are working for me, and estimate the demand curve for my services.

[ramanshah]

I'd want to check that a vendor product can categorize leads and show me conversion rates grouped by category. Estimating the demand curve is pretty pie-in-the-sky. That said, I would want a very easy way to review the quotes I've given people over time - especially, juggling different rates for different clients, to avoid misquoting my rate to a given client.

[ramanshah]

Signed up for a free trial of nocrm.io and am investing some time in learning its idioms/data model to make sure I understand how my business maps onto it. If I'm confident that it will suffice for organizing my sales activity and informing my sales and marketing strategy, then here's the plan for this repo:

• Rename it something evocative. One bit of Judy Robinett branding is "Titanium digital Rolodex." Maybe something along those lines? Something else on the subset or enhancement on a "CRM-y" feature set, speaking to the open-data or graph features?
• Edit the README to re-focus this repo's goals around the non-sales aspects of contact management and strategic social capital management.

[ramanshah]

Possible name: contector as a portmanteau of "contact" and "connector".

[ramanshah]

Sold! NoCRM is good enough for my needs. The analytics there are good enough for making strategic decisions about optimizing my sales funnel. The main sales activity that came up that's non-obvious to handle with NoCRM is to set up a policy for canceling sales when "ghosted" and to set up a tickler for recycling lost sales (on the theory that sometimes the lead likes you but there was no immediate need for the service, and a follow-up sale might be easier due to the social capital).

Respective ideas for these gaps:

• Standardize a ghosting policy when leads stop communicating. First crack:
  • Day 0: I communicate with the lead (e.g., have a sales call and e-mail a summary).
  • Day 7: I send a follow-up.
  • Day 21: I deem myself "ghosted" and cancel the sale.
• Remind myself in my physical tickler (say bimonthly) to review "lost" or "cancelled" leads from 6 months before for ones worth recycling. Duplicate the lead, set up a client folder, and reach out again.

Beyond that, contector, here I come!