Startups and mentoring
A startup is a search, not a smaller company.
Since 2023 Arnar has mentored Icelandic founders as a volunteer through KLAK VMS, the venture mentoring service of KLAK – Icelandic Startups. This page shares the method behind that work: the questions, not the answers.
Mentoring through KLAK VMS
Questions, not instructions
KLAK VMS pairs early-stage companies with a small team of experienced mentors, on the model of MIT’s Venture Mentoring Service. Arnar mentors through KLAK VMS and in accelerators such as Startup SuperNova.
Mentoring is voluntary, unpaid and confidential. What founders share stays with the mentoring team. The founder sets the agenda and makes the decisions; the mentor’s job is to ask sharper questions, look at the evidence with them and open doors where that helps.
Early-stage challenges
Where early startups go wrong
Building before discovering
Features get built before anyone has shown the problem is real, painful and owned by someone with a budget.
Talking to the wrong people
Friends, investors and polite strangers give praise. Praise is not evidence; a second meeting or a paid pilot is.
Growth charts too early
Cumulative sign-ups and week-on-week growth on five users measure nothing. Before product-market fit, count users who love the product, by name.
A plan that ignores the market type
A new market has a long flat stretch before anything rises. If the runway doesn’t cover it, the plan fails however good the product.
Repeatability must be proven abroad
Iceland is a fast, cheap test market, but small and personal. The part of the model that scales usually has to work in export markets.
Grants that become the plan
Grant applications reward detailed work plans. The risk is that the team executes the plan regardless of what the evidence says.
The search
Four stages, and evidence opens each gate
Evidence moves a company from one stage to the next, never the calendar or a new funding round.
- 1
Vision
Is the belief worth testing?
Gate: a written belief statement and hypotheses ranked by risk.
- 2
Discovery
Is the problem real, and for whom?
Gate: the customer’s workflow mapped; many in the target group confirm the pain.
- 3
Validation
Will customers use it, pay and come back?
Gate: users who love the product, named; paying customers; a sales channel that repeats.
- 4
Growth
Does it hold as we push harder?
Gate: input-to-output ratios hold as the company grows.
Evidence ladder
Weigh every claim by what the customer gave up
We place each claim on a ladder according to what the customer sacrificed to create it. Rungs 0 and 1 decide nothing. Product-market fit begins at rungs 6 and 7.
- 7Love and pull
Users return unprompted, pay, bring others and would miss it.
- 6Active use, measured gain
Users spend the time the product needs, and the gain shows in their numbers.
- 5Money
A letter of intent with terms, a pre-order, a paid pilot.
- 4Time or reputation
A second meeting, a pilot set up, an introduction to the budget holder.
- 3Observed behaviour
You have watched the workflow and the workaround in use.
- 2A specific story
The last time it happened, what it cost, what they tried.
- 1Opinion
“Great idea, I’d use it.”
- 0Founder’s belief
“Everyone needs this.”
Love before line charts
Count the users who love the product, by name
Before product-market fit the only goal that matters is a growing number of named users who love the product. At a hundred, growth charts start to mean something; it is very hard for a hundred people to love a product without it growing.
In every meeting
Questions we keep asking
- 01
What did you believe, what did you find?
What did you do about it, and where are you now? The learning loop opens every meeting.
- 02
Which hypothesis did you disprove this month?
A disproved hypothesis is progress. A quarter without one usually means nothing was tested.
- 03
How many users love the product, by name?
Not sign-ups or downloads. Names, and what they would do if the product vanished.
- 04
Draw your customer’s day
Where exactly is the pain, who decides and who pays? If you can’t draw it, discovery isn’t finished.
- 05
What did you cut this month?
When AI makes features cheap, the scarce skill is editing. A growing feature list without new users is a warning sign.
Start with a belief statement
We believe [target customers] struggle with [problem] because of [root cause]. Today they [workaround], which costs them [time, money or risk]. This is changing now because of [shift]. Once the product exists they will stop [doing X] and start [doing Y] because of [insight]. We are wrong if [result that disproves this] happens.
The method draws on Steve Blank’s customer development and on advice from partners at Y Combinator, adapted for Icelandic conditions.
Contact
Start a conversation
Whether you are weighing an investment, preparing your company for AI or building something new, we are glad to talk.