Before you sign

Three proposals for the same project: $80k, $240k and $600k.

Each one prices a different amount of your risk, and nobody who wrote them sits on your side of the table. I read them for owners who carry the decision without a CTO, show you which assumptions moved each number, and tell you which one I would sign with my own money. I sell no development and no software, and I take no vendor commissions or referral fees.

Other decisions owners bring me

“Everyone keeps telling me we need an AI strategy.” Which part of the initiative actually needs AI, what it changes in the business, and what a $20 a month tool would have done.

“My team says we need an app. Do we?” Usually not. When you do, what the smallest version that proves it looks like, and who should build it.

“The project is six months late and I cannot tell what is actually wrong.” Fix it, cut scope, replace the team, or stop. Decided against what runs on your data, not against a status report.

“Our vendor says we need the custom version.” Of course they do. What an off-the-shelf tool would fail at for your business, specifically, and whether the process should change instead.

“We are about to sign for three years.” Data ownership, portability, the API, termination assistance, who holds the credentials. The exit clause, read before the entry.

Everyone you can ask gets paid if the answer is yes

Ask the agency whether you need the build. Yes. Ask the vendor whether you need the platform. Yes. Ask the AI consultancy whether you need an AI initiative. Definitely. Your CTO, if you have one, is often the person who proposed it.

None of them is lying. Each of them was hired to answer “can we”, and you are asking “should we”.

The vendor

Wants you to buy. Knows the product better than anyone in the room and is the least neutral person in it.

The agency

Wants you to build. Its estimate is also its revenue, and nobody prices their own upside conservatively.

The consultancy

Wants a larger engagement. The proposal covers all four parts of the problem because it sells all four.

Me

Paid only by you to find out whether any of them are right. No implementation behind the recommendation, no referral fees, no preferred vendor.

I am on your side of the table. That is the whole product.

Start here: the Second Opinion One decision. A recommendation within two weeks, a call where I defend it, and access to me while it plays out.

See the Second Opinion

You bring one consequential decision: a proposal, a contract, a build, an AI initiative, a platform. I review what the decision needs, speak with the people I need to, challenge the assumptions that carry it, price the alternatives including doing nothing, and tell you what I would do if it were my money. The reasoning comes in writing, so you can check it and defend it: decision, evidence, options, recommendation, risks, next actions. Then I stay available while you negotiate, sign and build, for whatever the decision turns up next.

What a verdict looks like

Illustrative example. The business and the numbers are invented; the structure is the one you receive.

Decision: A 60-person distributor holds three proposals to replace its order system: $80k, $240k and $600k.
Evidence: The $80k quote assumes customer pricing fits the platform's standard price lists. Eleven of the fourteen customer contracts use tiered rebates it cannot express, so that quote turns into a custom build after signing. The $600k quote includes a warehouse module the business already replaced last year.
Options: Sign the $240k proposal as written. Sign it without the warehouse and reporting phase. Or keep the current system another year and fix the two reports behind most of the complaints.
Recommendation: Sign the $240k proposal without phase three, for $165k, and make data migration a fixed-price milestone. As written it is billed hourly with no ceiling.
Risks: The migration estimate rests on one sample export. If the full export is messier, go-live slips by about eight weeks.
Next actions: Pull the full export this week. Ask for migration as a fixed milestone before the signing date.

From real work

Four cases with the names taken out. Each one changed what I check before an owner signs anything.

Two quotes for the same app

A franchise owner asked me to find studios for a new marketplace app. Two of them quoted the same first version, and one came in almost twice as high as the other. Most of the gap came from two features: video, and payments, "especially refund flows". The owner wanted to know which studio to pick. The question that mattered came first: did version one need video and refunds at all?

What I check now: what drives each number, line by line, before comparing the totals.

A year of turning a customer away

An online coffee retailer learned that a customer in Korea had been blocked from its store for about a year. A firewall rule was rejecting her. I found the rule quickly. Changing it took two days, because one developer held the only login to the account.

What I check now: the domain, the hosting, the firewall and every app store account belong to the company, with the owner as admin, before the first invoice is paid.

Three to one on the dashboard

The same retailer's ad dashboards reported about three dollars back for every dollar spent. Then the card statement arrived, and the marketing lead wrote: "This is not being reflected in the profit margins." Measured channel by channel over the next holiday season, Google returned about 2.3 times its cost. Facebook lost money.

What I check now: any return a platform reports, against the bank statement and gross profit.

A domain sold, and the content with it

A startup where I had been CTO sold its domain. Days later the domain was back online, showing about 21,000 of our pages. The buyer had copied the whole site before the sale closed. Everything after that came down to one question for the contract: did the sale cover the domain name only, or the domain and its content? Three weeks on, I offered to export every article from the old server before it was shut down. The hosting account was cancelled about an hour later, before the copy existed, and nobody knew whether any of it could be recovered.

What I check now: a sale contract names each asset separately (the domain, the content, the backups), and nothing is cancelled until a backup has been opened and checked.

CPAs, fractional CFOs and attorneys often see these decisions before anyone else does. Here is how referrals work.