Refrakt

Three pillars. Four programs.

Every client moves through all four, in the same order. What varies is depth, not sequence.

Clarity: know who buys, and why.

The program every client starts with, whether it takes a week or a quarter.

Momentum: reach them and convert them.

Turning a defined buyer into a real target list and a repeatable way to reach them, then turning selling into a process, not a personality.

Control: see what's working, without you in the room.

One trusted system, leading indicators instead of lagging ones, a forecast you'd actually stand behind.

Clarity: every engagement starts here

The Foundation Program

Who it's for

Every client. This program is never skipped.

What you walk away with

A written blueprint you own, and could defend to a board.

The work

  • Rebuild the value proposition against evidence, not aspiration
  • Define and prioritise the ideal customer profile, or ICP (the type of business you're genuinely best suited to work with), and the segments underneath it
  • Map how buyers actually decide, and who's in the room when they do
  • Test the positioning against market reality, not internal assumption
  • Find where the story diverges between founder, sales and marketing

Where AI fits

Agentic research and synthesis works through internal conversations, with the founder and the sales team, on why past customers actually bought and what was happening in the market at the time. That turns into a clearer picture of what to look for in finding the next one.

Momentum

Demand and Targeting

Who it's for

Clients with a defined buyer but no repeatable way of reaching them.

What you walk away with

A working demand engine, and the logic behind every choice in it. You're taught the filters and tools to build your own target lists, not handed a purchased one.

The work

  • Build target account lists that follow from the ICP, not from convenience
  • Decide the data strategy: sources, enrichment, contact mapping
  • Design outreach across the channels where buyers actually are, chosen deliberately
  • Build messaging and campaign themes that can run more than once
  • Build the feedback loop that shows which channels and messages actually work

Where AI fits

Agentic research builds and enriches account lists and supports outreach at a scale manual prospecting can't match.

Momentum

Sales Process

Who it's for

Clients converting inconsistently, or only when the founder's in the room.

What you walk away with

A sales process a new hire could run in month one, plus the CRM, playbooks and proposal tool that keep it working after we're gone.

The work

  • Define the sales process stage by stage, and build in genuine adherence
  • Configure the CRM to match: the stages, the fields, what counts as a deal worth chasing
  • Build playbooks tuned to how your buyers actually push back
  • Hand over Proposal Assembly, a tool for producing strong proposals quickly
  • Set the standard for when AI gets used inside the process, and what's kept on record
  • Run a quarterly look at wins and losses, so the whole thing keeps improving

Where AI fits

Structured support on account research, first-pass proposals, and spotting the pattern in a quarterly review. The judgement calls stay yours.

Control

Systems and Measurement

Who it's for

Clients with activity they can't yet see, measure or forecast.

What you walk away with

Visibility you could take to a board, and the discipline to keep it true.

The work

  • Configure the CRM as the single source of truth, not a filing cabinet
  • Rationalise the stack: fewer tools, deliberately chosen, actually used
  • Build reporting that shows pipeline health, visible to the whole team
  • Define leading indicators, not just lagging revenue after the fact
  • Establish a forecasting and review cadence that holds

Where AI fits

Agentic support turns raw pipeline data into reporting and early signal faster than manual review allows.

Every client touches all four, in the same order. How fast you move through each one depends on how mature your commercial engine already is there.

Find out where the guessing is happening