Chief Marketing Officer / China
Three decisions that cost us revenue deliberately
- Digital screens
- 41,000
- Office lobbies, residential entrance halls and transit interchanges in six cities
- Revenue in 2025
- RMB742m
- After RMB701m in 2024 and RMB746m in 2023
- Revenue from local advertisers
- 29%
- 2025, up from 17 per cent in 2023
Decisions & consequences
01Stopping a camera attention pilot and building district holdout tests instead
In November 2022 a technology vendor offered to fit small cameras to our office-lobby screens. Its software would detect faces, estimate how long each person looked at the screen and report verified attention without, it said, storing images. Two national advertisers, a beverage group and a car brand, wanted those figures for their 2023 plans. Our sales team had already agreed a pilot on 2,400 screens in Shanghai and Suzhou.
National advertisers then provided 46 per cent of our revenue, and a competitor was promoting similar measurement. Refusing could look like hiding weak audiences. Proceeding meant processing facial information in lobbies where nobody had agreed to it. Counsel advised that the Personal Information Protection Law treats facial information as sensitive and tightly limits image collection in public places.
I recommended ending the pilot. Haoyu Rong, our chief commercial officer, argued that we would lose campaigns and that the vendor carried the legal exposure. The chief executive accepted my recommendation in December. The car brand moved an RMB9.5 million spring campaign to another operator.
We then built a method that identified no one. Property managers in 310 office buildings agreed to share hourly entry counts from their access gates, aggregated and stripped of identity data. We designed matched-district holdout tests: a campaign ran in some clusters of compounds and offices but not in comparable ones, and the advertiser compared its own sales or redemptions. Recall surveys used consenting panel members.
By the end of 2024, twenty-three advertisers had run holdout tests. Seventy-one per cent renewed, against 52 per cent of similar advertisers without a test. The beverage group measured a 6.8 per cent sales difference in exposed districts and doubled its 2024 booking. Holdout clusters cost us roughly RMB4 million of unsold inventory a year.
Measurement became a product rather than a defence. Next time I would brief the agencies' research directors before announcing the refusal, so that they heard our reasoning from us rather than from a disappointed sales team.
Holding our national price through the 2024 slowdown
In the first quarter of 2024, sell-through on our office-lobby screens fell to 54 per cent from 71 per cent a year earlier. Vacancy in our Shanghai and Suzhou office buildings was rising, and national advertisers were cutting outdoor budgets or demanding deep discounts through their agencies. Revenue for 2023 had been RMB746 million; the first-quarter run rate pointed nearer RMB650 million.
Haoyu Rong and the chief executive proposed a 25 per cent cut to every rate card. Their reasoning was straightforward: an empty loop earns nothing. My concern was that a general cut would become the reference price for every agency negotiation in 2025, and that it would do nothing for small businesses unable to afford even a discounted city-wide campaign.
I proposed a different plan. We held list prices and set a floor for each building tier, below which a deal needed the chief executive's agreement and mine. We offered national buyers a free holdout test instead of a discount. And we launched district fortnight packages: two-week flights on residential screens around a single neighbourhood, priced from RMB6,800 to RMB18,000, with self-booking. The board approved the plan in April, against the commercial team's preference.
The cost was real. One agency moved a beverage client's RMB12 million second-half booking to a rival offering 35 per cent off. Sell-through stayed below 60 per cent until the fourth quarter, and 2024 revenue fell to RMB701 million.
Realised national pricing ended the year 7 per cent below list rather than 25. Local advertisers rose from 17 per cent of revenue in 2023 to 29 per cent in 2025, when revenue recovered to RMB742 million. Our sales team now negotiates against agreed floors rather than in isolation.
I launched the local packages before the content-standards desk was ready for them. Clinic and training advertisements arrived without approval documents, review queues reached nine days in May and some small advertisers left. I would staff review first and sell second.
Rewriting residential category rules and opening partner fees to owners
In the first half of 2025, our property-manager partners logged sixty-three owner complaints about our residential screens. Most concerned unsecured consumer-lending apps and paid-livestream platforms shown in entrance halls where children waited for school buses. Owners' committees in four Suzhou compounds also asked where the screen fees went, and one voted to have the screens removed.
Residential screens produced 38 per cent of our revenue, and we paid property managers RMB96 million a year. The two questions were connected. Owners who felt the screens worked against their families were also the owners most likely to examine the fees. The Civil Code treats income from the shared parts of a residential building, after reasonable costs, as belonging to the owners, so opaque payments created exposure for our partners as well as for us.
I proposed two changes. We would exclude unsecured consumer lending and paid-livestream apps from all residential screens, and we would give every compound partner an annual statement of screens, hours, fees and complaints to share with its owners' committee. Haoyu Rong argued that the categories were lawful and the advertisers had done nothing wrong. Some property managers, including colleagues at Hexiaolan, preferred the existing arrangements. The board approved both changes in July 2025.
The category rule removed about RMB21 million of annual bookings. A hundred and forty compounds left the network when their managers declined to issue statements, taking 1,900 screens with them. Hexiaolan's buildings received exactly the same terms as everyone else, which its residential team accepted only after two difficult meetings.
Complaints fell to fourteen in the first half of 2026. Compound renewals reached 94 per cent, and three national food and children's-products advertisers increased residential spending, citing the category rules in their briefs.
I wrote the category rule after the complaints arrived. I would consult a sample of owners' committees first, and test each category on evidence rather than excluding a broad group, because our first wording briefly caught a licensed bank's credit-card campaign.