AI can make a go-to-market team faster at executing a flawed strategy.
Its more valuable role is to improve the decisions behind execution: which accounts deserve attention, what each opportunity warrants and how resources should be allocated across a market.
Speaking at the EUVC Summit & Awards Show 2026, Harrison Rose, Co-Founder of Goodfit and Paddle, argued that decision-making is AI’s real go-to-market advantage. His argument suggests a four-step test for GTM leaders.
1. Can AI see the market accurately?
While running go-to-market at Paddle in 2017, Harrison struggled to identify the software companies his team wanted to target. Standard industry filters included development agencies while missing businesses such as Hootsuite, which categorised itself under marketing and advertising.
Paddle employed seven or eight lead development representatives to review thousands of websites and enrich company records manually. Harrison and Alexander Berry, now his Co-Founder at Goodfit, replaced much of that work with a classification model trained on labelled examples.
The model gave Paddle a continuously updated view of its addressable market. This raises three questions:
Can the system distinguish a genuine target from a superficial match?
Does it capture companies that conventional filters miss?
Can it reassess the market as companies change?
AI cannot allocate GTM resources well if its view of the market is incomplete or inaccurate.
2. Does it have enough context to judge each account?
A representative can research one account deeply. AI can compare every account against a much larger body of company knowledge, including:
Previous emails and replies
Won and lost opportunities
Historical contract values
Past messaging and creative
Results from similar companies
The useful test is whether AI can connect that history to a specific recommendation:
Which evidence informed the recommendation?
What comparable accounts or interactions support it?
Is the system learning from failures as well as successes?
Is the underlying CRM data reliable enough to guide resource allocation?
3. Can it estimate what an account is worth?
At Goodfit, Harrison has been using expected value to prioritise accounts:
Expected value = probability of winning × predicted contract value
One model estimates the likelihood of winning an account. Another predicts its potential contract value. Together, they assign an expected monetary value before a sales conversation begins.
That estimate can help determine whether an account warrants a representative’s time, automated outreach, paid advertising or a more complex channel sequence.
Expected value does not remove uncertainty. It creates a consistent basis for comparing opportunities before internal enthusiasm, familiarity or sales intuition influences the ranking.
4. Can it turn the estimate into a better treatment?
A score only becomes useful when it changes what the team does.
For each account, AI could help decide:
Whether to pursue it
When to engage
Which people to contact
Which message and channel sequence to use
How much human time and budget to commit
The final test is whether those recommendations improve with every attempted interaction. Harrison’s vision is a system that updates its decisions as accounts change and new performance data arrives.
Programmatic advertising already works this way, with machines deciding who sees an advert, where it appears and what price is paid. Harrison expects machine-led allocation to influence a much wider share of go-to-market strategy.
Start by applying this four-step test to one market segment.
Compare the AI-ranked accounts with the team’s current priorities, then investigate every major disagreement. Those gaps will reveal whether the constraint lies in market coverage, context, prediction quality or the team’s own assumptions.
Listen to Harrison’s full talk for his perspective on expected value and the shift towards machine-led GTM decisions.
