June 12, 2025

Why 'AI Writes It in Seconds' Is the Wrong Metric

Vendors like to quote how fast a draft appears. That number tells you almost nothing about whether the draft was worth producing.

Generation speed versus useful speed

A draft that appears in ten seconds and needs forty minutes of correction is slower, end to end, than an adapted template that took two minutes to get right the first time. The metric that matters is total time to a sendable document, not time to first output.

The variable that actually predicts the outcome

The single biggest predictor of whether an AI-assisted draft saves time is how much specific, verified information it started with. A draft built from the prospect's actual site, actual names, and actual recent events needs light editing. A draft built from a generic prompt needs a rewrite, because the model fills gaps with plausible-sounding fiction rather than admitting it does not know.

What good input discipline looks like

Teams that see consistent gains treat the input stage as the real work: gather the source pages, confirm the names, note the specific event that makes this prospect timely. Drafting, done well, is almost mechanical once the input is solid. Done poorly, drafting becomes a second research project disguised as writing.

This is also where the risk of a confident but wrong answer concentrates. A model asked about a company it has thin information on will not usually say so; it will write confidently and incorrectly. The fix is not a better prompt. It is not asking the question until the facts are already in hand.

A short checklist for the input stage

Before asking for any draft, confirm four things are already in hand: the correct, current name of the company and the specific contact, the source document the draft should be grounded in, the one recent event that makes the outreach timely, and the specific ask or purpose of the document being drafted. Missing any one of these is the most common reason a draft comes back generic or wrong.

This is a two-minute checklist, not a formal process, but treating it as a required step rather than an optional habit is what separates teams that get consistently good drafts from teams that get an inconsistent mix.

Why this discipline pays off beyond the current engagement

A well-documented input trail on one proposal becomes a reusable reference the next time a similar prospect appears, a comparable company in the same sector, facing a similar situation. Teams that keep this trail organized find that later proposals in the same practice area move faster, not because of any change in tools, but because the research discipline compounds into an internal reference library over time.

A short way to catch thin input before it becomes a problem

Before a draft goes out for review, a fast gut check works well: does the note mention anything that could only be true of this one company, a name, a number, a specific recent event, or could every sentence in it apply to a dozen other companies in the same industry. A note that fails this check needs more input, not a better prompt or a cleverer rewrite.

Setting the right expectation before rolling any of this out

Teams that get the most durable value from AI-assisted prospecting are usually the ones that set expectations narrowly and specifically before rollout, rather than promising a broad, undefined improvement in productivity. Naming the exact task being automated, first-pass research, a first-draft outreach note, and the exact task that is not, judgment calls, relationship context, final sign-off, gives everyone a shared, checkable standard to measure the rollout against six months later, instead of a vague sense of whether things feel faster.

It also helps to name who owns keeping this working once the novelty wears off, someone who checks periodically that the input discipline is still being followed rather than assuming it holds on its own, and who a colleague can ask when a new situation does not fit the pattern described here. That kind of named ownership tends to be the difference between a good habit that lasts a month and a standard that survives a full year of real client work.

A final, practical way to judge whether a rollout is working

Six months after adopting any of the practices described here, ask the team a direct question: name one specific week in the last quarter where this made a real difference, and one specific place it still falls short. A team that can answer both halves of that question specifically has genuinely internalized the process. A team that can only answer in generalities, everything is faster now, has probably not looked closely enough at where the actual gains and gaps sit, and is due for a more honest audit before the next round of investment.

VIPMarketing Editorial