Actionable insights — QuinStreet (QNST)

1. The "AI-victim" mispricing screen — buy what the narrative orphaned

1. Build a list of stocks down hard and stuck in a downtrend with no breakout — the names the market reflexively dumps under a popular fear ("an AI victim," here).
2. Separate price action from operating results. Pull the last 4–8 quarters of revenue, EBITDA, and free cash flow; flag any name whose fundamentals are improving while the stock falls.
3. Ask whether the feared disruption is real or just a narrative — and what it would actually take for it to happen (a competitor building the moat from scratch).
4. Demand a thrifty valuation floor (here ~0.6–0.8× sales) so you're paid to wait while the story is "unproven."

• Stocks selling off on a thematic fear while reporting accelerating revenue/EBITDA; valuations that "price in failure."

2. Source-credentialing — weight an idea by the picker's track record

1. When an idea is sourced externally, evaluate the source's documented hit-rate before the thesis itself.
2. Look for one undeniable, dated, public win (entry price, date, current price) as the credential — not a vague reputation.
3. Treat a draft review by the source as added conviction, but still stress-test the bear case yourself.

• A specialist's concentrated, verifiable wins; ideas they've personally vetted vs second-hand tips.

3. The single-quarter shock as a moat-discovery signal

1. Hunt for a violent, single-quarter revenue spike in an out-of-favor vendor (here auto-insurance revenue +615% YoY) — it usually means a customer base suddenly needed what the vendor sells.
2. Trace the macro cause (Covid scrambled auto-insurance economics → carriers desperate for leads) to confirm the demand is structural, not a one-off.
3. Check whether the spike broadened (more carriers spending $1M+/month than ever) rather than concentrating in one client.

• A demand shock that converts a commodity vendor into critical infrastructure; broad-based vs single-customer growth.

4. Stress-test the disintermediation bear case before buying the cheap stock

1. For any "middleman" trading cheap, name the existential threat explicitly (Google/Meta building the tool in-house).
2. Apply the "they've had years and haven't" test — has the obvious disruptor already had time and incentive, and not done it? Ask why not (regulation, integration complexity, conflicting ad revenue).
3. Identify the part of the moat that isn't replicable cheaply (here QRP's 100+ carrier integrations + issuance — "the juice might not be worth the squeeze").

• A feared disruptor with the means but not the motive; switching costs once a platform is embedded in workflows.

5. The margin-milestone re-rating math — value the inflection, not the trailing print

1. Take management's explicit margin target (here 10% EBITDA, from ~7.3%) and apply it to guided revenue to get forward EBITDA.
2. Apply a defensible multiple range (10–12× for a high-growth digital platform) to get an implied market cap and per-share value.
3. Sanity-check that gross cash flow ≈ free cash flow (no interest, modest depreciation, non-cash amortization) so the EBITDA is "real."

• A credible margin-expansion path applied to guided revenue; whether EBITDA converts to actual cash.

6. The accretive bolt-on as a "purely additive" call option

1. When a turnaround buys a business, check whether it's funded conservatively (here $45M cash + a credit line, leaving ~zero net debt) and immediately cash-flow positive.
2. Treat year-one accretion as the floor, with un-modeled synergies as upside — but don't pay for synergies until they show.
3. Prefer deals that add a new capability (HomeBuddy's social/native reach + auction-exclusive leads), not just more of the same.

• Self-funding, immediately-accretive bolt-ons that diversify the revenue engine rather than concentrate it.

7. Size to the conviction — small for a "demonstrated but unproven moat"

1. Match position size to where the thesis sits on the proven/unproven spectrum — a moat that's "demonstrated but unproven" warrants a smaller stake.
2. Average into the position rather than buying all at once, especially on a name with no broken downtrend.
3. Define the confirmation that would let you add: "if the next few quarters confirm the turnaround is real, the stock likely re-rates."

• The next 1–2 quarterly confirmations as the trigger to scale up; sideways-to-down if the story resets.