Six Iranian Companies Measured the Same Twelve Days

Twelve days in June 2025 are the single most asked-about period in Iran’s digital economy and one of the worst documented. Until this week our own record carried a measured figure for the whole event from two companies: Snapp, whose own report prints intra-city trips down 42 percent, and Behpardakht Mellat, whose own report prints no outage at all. Four more Iranian companies have published their own account of those days, in Persian, as slide decks and report sections that nobody appears to have read in English. Tehran Index has now read all four end to end, page image by page image, because the Persian text layer in these files reverses digit order and cannot be trusted for a figure.

The four are Digipay, the payments and instalment credit company inside the Digikala group; YektaNet, a display advertising network that places advertising across more than 5,300 Persian websites; Khanoumi, an online cosmetics and personal care marketplace; and Digikala itself, through a market research service it sells to brands. With Snapp and Behpardakht Mellat, already on our record, that is six companies measuring the same event in their own operating data, across payment infrastructure, consumer credit, advertising, beauty retail, general retail and mobility. Our existing piece, Snapp’s 2025-26: The Resilience Numbers, read one company’s shock in depth. This is what the same twelve days look like across an economy, and the contribution is that the six readings disagree with each other in ways that are themselves the finding.

First, the thing that makes most of this uncitable if you skip it

No two of these companies measured the same days. YektaNet runs the war from 13 to 24 June 2025 against the fourteen days from 30 May to 12 June. Digikala runs it from 14 to 26 June. Digipay prints no dates at all and says only “the equivalent span before the war”. Khanoumi is not measuring a span at all; it is marking single days inside a full financial year. Behpardakht Mellat is the only one of the six that compares a dated twelve-day window against two dated monthly averages, which is also why its figure is the easiest to check. Anyone who stacks these percentages into one chart is producing a number that exists in none of the sources.

The rails stayed up. The commerce on top of them did not.

Start with the company that reports nothing happened. Behpardakht Mellat is one of the two largest payment processors in the country and a listed issuer, and its own report states that its systems ran continuously across the twelve days from 13 to 24 June 2025 with 99.99 percent availability and no recorded service outage. Average daily transactions over those twelve days were 27,171,268, which it puts 6.18 percent above its average for the month to 20 April 2025 and 5.46 percent below its average for the month to 21 May 2025. On the national card rails, in other words, the twelve days register as an ordinary fortnight.

Hold that against what the rest of this piece says and the shape of the event changes. The switch kept clearing payments while Khanoumi’s ordering in Tehran province fell about 95 percent in a day and Digipay’s super app transactions fell 66 percent. Nothing broke underneath. What broke was above it: the connection to the customer, the bank systems at the far end of a specific transaction, and the willingness to spend on anything that was not food, power or a way back online.

Snapp’s own report, already on our record, is the other side of that. It prints intra-city trips down 42 percent, active driver users down 35 percent and completed inter-city trips down 28 percent, against growth of up to 116 percent in inter-city trip requests. Demand to leave the city rose while the trips actually completed fell, which is the same looking-against-buying gap that Digikala’s modem numbers show further down. It names Tehran, Kish, Karaj, Qeshm and Kermanshah as the cities with the largest falls in intra-city trips, and records 10 billion toman of non-repayable assistance paid to 2,500 high-trip drivers along with a commission exemption.

Digipay: a credit default that was a banking outage

Digipay’s report carries two numbers and the second is the one worth the attention. Credit purchases at contracted merchants and other super app transactions fell 66 percent against the equivalent window before the war, which Hamidreza Saadati, the executive director of the super app, attributes to internet restrictions, a change in what people were prioritising, and people leaving some large cities.

Repayment of instalments through Digipay fell 13 percent, and the company is explicit that this was not a change in willingness to pay. It states the fall related solely to users who intended to pay their instalment and could not, because of the outage in the systems of the two banks it names, Bank Sepah and Bank Pasargad. That is a lender publishing a measured dip in its own collections and attributing it to a bank infrastructure failure rather than to its borrowers. We have found no other Iranian lender that has put a figure on that distinction in public, in any language. Digipay also waived late-payment penalties on instalments from the months through 21 June 2025, and gave users until 22 July 2025 to repay the following month’s instalments without a late-payment penalty.

YektaNet: the money left before the audience did

YektaNet’s is the most rigorous of the four. It defines four dated windows, publishes the daily series for each index, and separates supply from demand. On the demand side, total advertiser budgets on the platform bottomed on 22 June 2025, the tenth day, at 79 percent below the pre-war fortnight. Active campaigns averaged 5,585 a day before the war and 3,517 during it, down 37 percent, with a trough of 2,812 on 20 June. Cost per click fell by up to 46 percent, which the report frames as a straightforward consequence: with that much demand withdrawn, the auction cleared cheaper.

On the supply side the picture inverts. Total daily visits across the more than 5,300 websites carrying its advertising fell only about 14 percent, troughing on 18 June, the war’s sixth day. Advertising impressions inside mobile applications rose 28 percent and inside mobile games rose 41 percent. Attention did not disappear. It moved off the open web and into apps, and it moved between subjects.

The music category losing 71 percent of its page views while video gained 37 and political news gained 35 is the clearest description of a population’s attention under stress that exists in Iranian commercial data. The supply and demand split also produced a measurable hole in the market: with video traffic up 37 percent and video advertisers absent, YektaNet’s video fill rate ran at 35 percent during the war, against an average of 80 percent through the recovery and 93 percent afterwards. For twelve days roughly two thirds of the video advertising inventory on its platform went unsold for want of a buyer.

YektaNet then does something none of the others attempt. It scores about 100 businesses, the largest advertiser in each industry, on two axes: how far each cut its budget during the war, and whether its budget afterwards came back above the pre-war level. Thirty percent of the businesses studied cut less than half and came back higher, and the report names banks, digital currency, car sales and services, and tourism and accommodation rental among them. Thirty-seven percent cut more than half and never recovered to the old level, among them home appliances, online retail, brokerages and portfolio managers, internet taxis, digital services and apparel. That second group is the larger one by count and the smaller one by money: on the report’s own chart its share of total advertising spend fell from 50 percent before the war to 20 percent during it, while the resilient group’s share doubled from 17 to 34.

Digikala: people looked at modems and bought a hardware wallet

Digikala’s report is the only one of the four that separates what customers searched for, what they looked at, and what they actually bought, and the three do not agree. Searches ran to the contents of an emergency bag: solar charger up 1,245 percent, canned food 969, dried bread 855, solar power bank 754, first-aid kit 653. Visits by category show food up 20 percent and fashion and apparel down 11.

Inside digital goods, the gap between looking and buying is the finding.

Visits to ADSL modems and routers doubled. Sales of them fell 27 percent. The same inversion runs down the whole category, and it is what a connectivity shock looks like from inside a retailer: the demand was real, the fulfilment was not. Average delivery time during the war was 5 days.

One line in that table goes the other way. The best-selling item in Digikala’s entire digital goods category during the war was a hardware cryptocurrency wallet, up 218 percent, and the report volunteers its own reading: that the hacking of digital exchanges during the war frightened people into looking for a way to protect their holdings. Our note on what Iran’s crypto exchanges disclose found eleven of them claiming about 45 million accounts between them and one holding a current state trust seal. This is the demand side of those same twelve days, measured at the till by the country’s largest online retailer. The two best sellers in the category after the war were a game controller, up 102 percent, and console games, up 93.

Elsewhere the report reads as a public health record kept by a shop. Blood glucose test strips were up 296 percent in sales, which the report puts, in its own hedged words, down to pharmacies being closed and to diabetic customers fearing a long event. Canned aubergine, foreign rice and canned lentils, up 87, 84 and 64 percent, were the best-selling supermarket lines. Through all of it the platform averaged 3.4 million daily visits and 2.5 million daily unique searches, and its heaviest ordering day of the whole period was 23 June 2025.

Khanoumi: the only one that kept the clock running afterwards

Khanoumi’s section is not a war report. It is the connectivity chapter of its annual report for the year to 20 March 2026, and it is the only document of the four that places the twelve days beside what came later in the same year. That is why it ends this piece.

Its granularity is the finest of the four. It names the three lowest-selling days of its year and the measured network availability on each: 17 June 2025, below 3 percent; 9 January 2026, below 1 percent; 1 March 2026, zero. It identifies the exact hours the business stopped, 18:00 on 17 June 2025 and midday on 28 February 2026, both of which are normally its two busiest ordering hours of the day. In Tehran province, which is the single province it singles out, ordering fell about 95 percent below its usual daily volume on 18 June 2025, and again on 1 March 2026. When its main social channel went quiet for 88 days, visits to that account fell nearly 90 percent against the 88 days before, and its video channel followers went from 1,700 to 2,500 over the same window.

The finding is in the comparison. Across the three interruptions, Khanoumi tracks five markers: the event, the network going down, the network coming back, sales falling, and sales returning. In June 2025 sales returned after the network did: the connection is marked restored on 25 June and sales on 30 June, and the company puts the gap at four to five days. In January 2026 the order reversed: sales were marked back on 19 January, eight days before the connection was. In the third episode, which the company dates from 28 February 2026, sales are marked falling on 1 March and returning on 28 March, and the reconnection marker sits later still on the same row.

Khanoumi draws the conclusion itself, and it is a sentence about planning rather than about any event: a business that schedules its behaviour around when the network comes back no longer has anything to schedule against. For the twelve days of June 2025, reconnection was still the thing commerce waited for. It was the last time it was.

What we do not know

None of these are audited. All four are company communications, three of them designed as slide decks. Not one prints an absolute number of transactions, orders or rial alongside its percentages, so every figure here is a ratio whose base is undisclosed. A 66 percent fall and a 13 percent fall tell you nothing about their relative size in money.

The windows do not line up and cannot be made to. Digipay prints none. Digikala prints a war window that starts a day after YektaNet’s and ends two days later. Any cross-company arithmetic is ours, not theirs, and we have not done any.

Two of the reports carry an internal inconsistency. YektaNet’s page 19 states that the worst-hit group of advertisers held only 10 percent of total budget during the war; its page 20 chart labels that same bar 20 percent, and only the chart figure reconciles to 100 across the four groups, so the chart is what we print. In Khanoumi’s timeline the third episode’s reconnection marker carries the wrong year against its own row heading, which is why we give that episode’s sales dates and the order of the markers and not a converted reconnection date.

Four reports is not a sector. These are four companies that chose to publish. The ones whose twelve days went worst are unlikely to be among them, and the survivor bias runs the same direction in every item above. Of the 376 public company records in our registry, 372 still carry nothing at all about this period.

And the one audited figure does not exist. No Iranian company subject to a statutory audit has quantified the event in a filed financial statement that this desk has found. Snapp’s 7 percent is a company announcement. Until an audited set of accounts carries a number for those twelve days, everything above is management reporting, and it is labelled as such on every row.

Cite as: Tehran Index, “Six Iranian Companies Measured the Same Twelve Days”, 6 October 2026. Company figures read from the first-party reports of Digipay, YektaNet, Khanoumi and Digikala on 6 October 2026, each page read as a rendered image; Snapp and Behpardakht Mellat figures from those companies’ own reports as recorded in the Tehran Index facts layer. Registry counts computed from the Tehran Index company registry and facts layer, 6 October 2026, n=376 public companies and 1,345 public facts. tehranindex.com