ReadyCIO
Menu

Note

My website said "security" sixty times and "revenue" eight. Owners buy the other way round.

A services site written by a technical person sells protection. An owner buys gains first: hours back, costs down, work won. I counted the outcome words on my own owner-facing pages and found the imbalance. What I changed, and how to run the same count on any AI proposal put in front of you.

For companiesBy ai strategyDraft

I sell AI readiness to owners of 10 to 200 person firms, and until this month my own site was selling them protection they hadn’t asked for. I know because I counted. On the pages an owner lands on first, “security” and its relatives appeared more than sixty times. “Continuity” more than thirty-five. “Revenue” about eight, all in passing. “Cost” about three.

That’s what a technical person does. We sell what could go wrong: what to check, what to require, what to watch. An owner, or the fractional CFO advising them, buys gains first. Hours back. Costs down. Work won. Protection second, and only once they believe the first part.

Count before you rewrite

The count takes ten minutes and settles the argument. Take the owner-facing pages, the home page, the start guides, the about page, any quiz, and tally five families of words:

  • Cost: cost, saving, cheaper, budget.
  • Revenue: revenue, sales, win, customers, growth.
  • Hours: hours, per week, time saved.
  • Security: security, breach, leak, privacy, data.
  • Continuity: leaves, handover, notice, outage.

One trap. “Hours” showed up often on my pages, but as the effort to do the work (“an afternoon”), not as hours saved. Effort words read like a cost to the buyer. Count them separately, or they flatter the result.

What I changed

Not new pages. A frame. Five changes, cheapest first.

An outcomes band near the top of the home page: five tiles, one per outcome, each with one real example and a link to the depth page. The headline does the reframe: AI is not the point. Hours, costs, revenue and risk are. Under it, one line: every example here is from my own work, and figures appear only where they were measured. For a small-business reader who has seen inflated ROI claims, that sentence is the asset.

Revenue clauses where the copy already described a process change. Same-day quotes win more of them. Enquiries answered first beat the competitor’s reply. A one-page AI policy answers a larger customer’s supplier questionnaire, which turns a compliance chore into keeping a customer. Usually one clause added to a sentence that was already there.

An outcome label on each item of the readiness checklist. The wording stays. The label says why to care: protects hours and cost, protects security and customer trust. The item that’s a prerequisite rather than an outcome, someone is in charge by name, gets labelled honestly: gets the rest done. The list is here.

A tag on every service: saves, grows or protects. A fractional CFO matching services to a client’s question scans the tags, not the descriptions.

And on a self-assessment result, say what staying at that step costs, not only what to do next. Cost of inaction is the executive frame for urgency, and it belongs in the PDF too, because the PDF is what gets forwarded.

Figures only where they were measured

Inventing a plausible number is the fastest way to lose an owner, and the site’s own rule forbids it. Across everything I’ve built for clients, exactly one outcome had a measured figure: fifteen supervisors each spending half a day a week on time cards, taken to a few minutes each, worth hundreds of thousands a year to that business. That’s the one tile with a number.

The others say what changed without one. Billing that no longer needs a person. Issues found within a day. Still running on Monday if they leave on Friday. The revenue tile names the mechanism and no figure at all.

There’s a boundary here. For deciding what to build next, I do estimate hours saved, in bands, as an internal prioritizing tool. Those estimates never go on a public page as client outcomes. Only measured results do. If you want the measured version before you build, the method is here.

Run it on the next proposal you’re handed

The count works on anything an AI person puts in front of you, including a proposal from your own staff. Tally the five families. If it’s all security and continuity, it was written by someone technical who’s trying to protect you, which is fine, but ask them which of the five it moves and what was measured. If it’s all revenue with a confident number, ask where the number came from. Seven questions for anyone selling you AI covers the rest.

One wording note. Say staff hours, not man-hours. Lead each claim with the outcome, not the mechanism. The mechanism is my job. The outcome is yours.