Marketing Intelligence Brief — AB Fleet · 2026-W34

Cadence per [[marketing-intelligence]] (B2B lane). Scan date: 2026-08-17.

TL;DR

• Biggest finding: a direct competitor with published pricing finally exists — Điều hành Xe. Same buyer, same six reports (P&L, lãi/lỗ chuyến, per-vehicle/driver/customer, định mức xăng dầu), ₫499k / ₫990k / ₫2,490k per month.
• Their Growth tier is ₫990k for up to 30 trucks. Our Standard is ₫1,500,000 for 20. We're ~2× a published price in the same band, and the playbook's "nobody publishes prices" anchor is now false.
• Their gap is exactly our wedge: no OCR, no receipt photo, no GPS integration mentioned anywhere on the site. Costs are typed in.
• But they already use Zalo (expiry alerts). Zalo alone is no longer a differentiator; the receipt→truck join is.
• Regulatory: Decree 254/2026/NĐ-CP took effect 2026-07-01, replacing Decree 123. Fuel invoices must carry each station's business-location code. (meInvoice)
• This week's experiment (carried, 7th): onboard customer #1 + 1 testimonial line. Metric: customer #1 live + 1 quote.
• Last week's result: no result recorded → carried forward. content-log empty. Competitor-watch: dead, see below.

Last week's experiment → result

So what: seven weeks, same experiment, still zero attempts. Nothing in the vault since 8/10 touches onboarding.

Commits since W33 are trading-bot, freight-docs purchases and Gain research — no fleet onboarding work. The three mechanical steps are unchanged: pull the bot account out of the AB Logi group, attach the bot process, one line in org-groups.json. No new experiment.

W33 "Next scan" — worked through

So what: the item I was told to stop scanning (price) got answered by a competitor I didn't know existed. And the feed item resolved by discovering the feed doesn't exist.

The new competitor — the one thing that changes the pitch

So what: our price is no longer unanchored. Either justify 2× on the receipt→P&L join, or the number moves.

Confidence: pricing/tiers/reports verified (own pricing page). "No OCR/GPS" is absence of evidence, not proof — a trial signup settles it.

Two conclusions, both actionable without a customer:

1. Stop pitching "per-trip profit." They sell that at ₫990k. Pitch what typed-in software structurally cannot do: the litres on the receipt vs the kilometres the truck actually drove. [[HANDOFF]] 8/08 is the proof — the non-integer-VND rule caught ₫12.7M of understated fuel that OCR confidence rated 1.0.
2. A ≤5-truck tier now exists at ₫499k. We have no answer below 20 trucks.

Competitive map — sourced changes only

So what: one new direct competitor, correctly priced and shipping; everything else unchanged from W33.

Competitor-watch feed — killing it

So what: W33 set the ultimatum, W34 executes it. The script isn't in this repo and the file hasn't moved in five weeks.

marketing/competitor-watch.md still dates 2026-07-14: all 🆕 baselines, zero 🔴 CHANGED rows, Binh Anh still ⚠️ fetch-failed. No competitor-watch.mjs anywhere in llm-wiki. Stop reporting it as a live feed. Either it lives uncommitted on the mini (check once), or it was never built. The manual scan found this week's only real signal anyway.

Copy bank (exact phrases, sourced)

So what: owners describe the pain as revenue-without-profit, and blame depreciation they never counted. Lead with that sentence, not with "receipts".

Counter-copy to test, updated for the new competitor: "Phần mềm nào cũng tính lãi/lỗ chuyến — nhưng số lít đó ai gõ vào?" (Every tool computes per-trip profit — but who typed in those litres?)

Content loop — own outlier math

content-log empty — no posts recorded yet.

This week's ONE experiment (carried forward, 7th)

So what: unchanged. A competitor now sells our pitch at ₫990k while we still have zero customers — more scanning can't fix that.

Onboard customer #1 end-to-end (org + bot in their Zalo group + first OCR'd receipt approved) and ask for one quotable sentence. Metric: customer #1 live + 1 testimonial line. Lead with receipt→odometer reconciliation, not per-trip P&L (commoditised at ₫990k) and not GPS (only the demo adapter is ready, [[HANDOFF]] 8/01).

Next scan (W35)

• Sign up for Điều hành Xe (or Bách Khoa's 7-day trial) and confirm by hand whether either accepts a receipt photo or a fuel XML. Only remaining question that changes the pitch — a signup, not a search.
• Pricing decision, owner call: hold ₫1.5M and justify on the join, or add a ≤5-truck tier against their ₫499k.
• Delete the competitor-watch section unless the script surfaces on the mini.
• Watch for Điều hành Xe adding OCR — the single move that erases the wedge (replaces last week's XML-ingest watch item, which came back empty).
• Human-in-loop: verbatim pains from closed Zalo/FB groups (9th week). oto-hui/otosaigon are open and productive — use as fallback surface, not as an excuse to keep deferring the account.

Note: the Write to marketing/2026-W34-abfleet-brief.md was declined, so this brief is not yet saved to the vault. Say the word and I'll write it.