Yelo Bank overtakes Unibank and tops Azerbaijan's Central Bank complaint index again

Yelo Bank topped the Central Bank's August table with an index of 2.51. That is two and a half times the neutral level and the bank's highest reading since November 2025.

Turanbank, Yapı Kredi Bank Azərbaycan and Unibank completed the top four. These are exactly the same four banks that led the table in January 2025, where our observation series begins.

The market as a whole did not deteriorate. The average index stood at 1.08 against 1.06 in July, and the median was even lower than in July. The number of banks drawing a disproportionate share of complaints for their size did not change either: eight in July, eight in August. What changed was not the breadth of the problem but its height.

How we read the table from this issue on

Starting with the August issue, we no longer rely on the Central Bank's colour groups. Instead, we classify banks ourselves, by their position relative to the neutral level of 1.00.

The reason is that the regulator's colour scale cannot be derived from the methodology it publishes alongside it. The data themselves remain the Central Bank's data: the formula behind the index is disclosed, we have no questions about the figures, and the entire analysis below is built on them. Our only question concerns the layer of interpretation placed on top of the figures, namely the split of rows into “high”, “medium” and “low” index. We set out the full case at the end of the article so as not to keep the reader from the main story.

We used these groups in previous issues. With twenty months of observations behind us, we think it right to say so openly and move to a stricter criterion: the very one the Central Bank names in its own legend but, judging by its tables, does not apply.

What the value of 1.00 means

The complaint index is calculated as a bank's share of all complaints received by the Central Bank during the month, divided by that bank's share of the sector's customer base. A customer is counted if they hold an account with a balance of at least 100 manats and have made at least one transaction in the last three months. Repeat complaints, requests for loan relief and matters outside the regulator's remit are excluded from the calculation.

This leads to a simple property that makes 1.00 the natural reference point. The banks' shares of complaints add up to one hundred percent, and so do their shares of the customer base. It follows that the sector average, weighted by number of customers, always equals 1.00. This is not an estimate or an observation. It is an arithmetic consequence of the formula itself.

A reading above 1.00 therefore means that a bank generates more dissatisfaction than its size would suggest, and a reading below 1.00 means it generates less. It is worth remembering that the index does not measure the number of complaints. A large retail bank with hundreds of complaints can sit close to 1.00, while a small bank can climb sharply on a handful of cases. The index measures service quality adjusted for scale, and it only becomes meaningful over long series.

The market has not heated up. It has polarised

Sixteen banks made the August table, against seventeen in July. Premium Bank dropped out of the list, which according to the regulator's note means it received no complaints during the month. Overall tension stayed practically unchanged: an average index of 1.08 against 1.06 a month earlier, and a median of 0.94 against 0.95. By the standards of this series, the backdrop remains cold. The average was 1.66 in February and 1.65 in January.

The top of the table, however, has pulled away from the market. The maximum value rose from 1.95 to 2.51, and the spread between the first and last rows widened from 1.66 to 2.27 points. The gap between the leader and second place reached 0.58 points, compared with 0.18 a month earlier. The table has stretched upwards while remaining calm for the most part.

A simple calculation makes this clear: without Yelo Bank, the average for the remaining fifteen banks would have been 0.98, which is below the neutral level. In a certain sense, the whole of August's “heating up” rests on a single row.

On our own scale, the picture is steadier than the regulator's colours suggest. Eight of the sixteen banks are above 1.00 in August, exactly half the table. In July there were also eight, out of seventeen. The problem has not widened, but its peak has become noticeably sharper.

Yelo Bank: a trajectory, not a spike

Yelo Bank led the August anti-ranking with an index of 2.51. The bank is two and a half times above the neutral level and nearly twice as high as the second row of the table. In the regulator's report it is the only bank assigned to the high-index category, the first such case since May.

A single value says little on its own, so we look at the series. Since the start of the year, Yelo Bank has posted 0.93 in January, then 1.68, 1.80, 1.49, 1.96, 2.06, 1.77 and 2.51 in August. The January figure was the best in our entire observation history, and since then the bank has been rising almost without interruption. It has held one of the top two places for the third month running, and its current reading is the highest since November 2025, when the index stood at 2.88.

The year-on-year comparison is also telling. In August 2025, Yelo Bank stood at 2.39 against a market average of 1.42, roughly 1.7 times higher. Now it stands at 2.51 against an average of 1.08, which is 2.3 times higher. Over the year the market has cooled by almost a quarter, while Yelo Bank has returned to last year's levels.

This is where the two-scenario model we return to in every issue comes in. A customer turns to the regulator either because their complaint went unanswered inside the bank for too long, or because the problem is systemic and simply cannot be solved at the level of a branch or call centre. The first scenario produces spikes, the second produces plateaus and slow climbs. Over the past seven months, the shape of Yelo Bank's curve is closer to the second.

A fair sense of scale also requires looking the other way. A reading of 2.51 is far from the historical peaks of the series: Bank BTB reached 7.65 in February 2025, Bank Avrasiya 6.89 in January 2026, and Yelo Bank itself hit 3.75 in June 2025. This is not a service crisis. It is the highest level of tension in a year that has been noticeably calmer.

Turanbank: a reversal that calls for further observation

Turanbank took second place with an index of 1.93, almost twice the neutral level. As recently as spring, the bank sat at the very bottom of the table: 0.32 in April, 0.36 in May, 0.40 in June. In two months its reading grew almost fivefold, and the bank climbed from fifteenth place to second.

This is the bank's highest value since January 2026, when the index stood at 2.03. Turanbank belongs to what we call the volatile group in this series: over twenty months its reading has ranged from 0.32 to 3.24, a tenfold spread. Banks like this rarely produce smooth series, and their sharp moves are usually linked not to a gradual build-up of dissatisfaction but to specific events, such as changes to fees, product terms or procedures.

We have no grounds to say what exactly happened: the Central Bank publishes only the ratio, not the breakdown of complaints. What we can record is the shape. After three months of exceptionally clean readings, the bank moved into the upper part of the table within two months.

The old four are back

The most interesting feature of the August table emerges when you look not at individual rows but at who sits at the top. The first four places went to Yelo Bank, Turanbank, Yapı Kredi Bank Azərbaycan and Unibank. The very same four banks made up the top four in January 2025, where our observation series begins, only in a different order.

In the year and a half between those two points, the table was led by entirely different players. Bank BTB and Bank Avrasiya, which posted the record values of the series, are absent from the ranking today for certain reasons (one of them has left the sector altogether). Expressbank, Accessbank and AFB Bank rose to the top in one-off spikes and dropped back just as quickly. Now the upper part of the table is held by banks with a long history of high readings.

Unibank slipped to fourth place in August with a reading of 1.64, and formally this is an improvement: a month earlier it stood at 1.95 and topped the table. Yet over twenty months of observations, the bank's index has never fallen below 1.53, and its average is around 1.94. It has not had a single truly clean month in that period. This is what chronic underperformance looks like: the bank almost never tops the table, and almost never leaves the upper part of the table.

Yapı Kredi Bank Azərbaycan took third place with 1.71 and has stayed above the neutral level for the third month in a row. In 2026 its readings have held within a range of 1.33 to 2.05, with no dips and no surges, showing the same evenness as Unibank.

Above 1.00, but under the radar

Four more banks in the August table exceed the neutral level while remaining in the regulator's “green zone”. This is where our scale and the Central Bank's diverge most visibly.

Expressbank, with a reading of 1.18, rose from thirteenth place to fifth. In its case the monthly move is best viewed over a longer horizon: a year ago, in August 2025, the bank topped the table with an index of 2.83, and in February 2026 it stood at 3.01. Against that background, the current value reflects a marked improvement, albeit with continued volatility: in July the bank recorded 0.71.

Bank of Baku, at 1.16, continues its slow decline from the June peak of 1.92 but remains above 1.00. This is characteristic of the bank: in 2026 it has not once fallen below the neutral level. Bank Respublika, at 1.07, has hovered around 1.00 for the fourth month running.

Kapital Bank requires a separate explanation. Its index came to 1.02, its first reading above 1.00 in 2026, and in eighth place the bank formally falls into the upper half of the table. However, the excess is two percent, and this is the country's largest retail customer base, for which a ratio this close to parity points to stability rather than to a problem. This is a case where the figure needs commentary, not a ranking.

Rabitabank: the best result in our entire observation history

The biggest improvement of the month came from Rabitabank: from 1.62 to 0.75, and from third place to twelfth. The value is unique for this bank. In twenty months of our observations it had never fallen below 1.00, and its previous best was 1.09 in July 2025. The current reading is almost a third below that previous low.

In the July issue, Rabitabank appeared in the headline as one of the top three in the anti-ranking, so its current result deserves a separate mention. One month does not make a trend, and the September table will show whether the bank can hold this level. Still, a move from the top three to the lower half of the table in a single month is rare in this series.

The second consistently positive example is AFB Bank. June's worst performer, which posted 2.27 that month, has held 0.47 for the second month in a row, half the neutral level. At the bottom of the table, PAŞA Bank (0.47), Xalq Bank (0.59) and Bank VTB Azərbaycan (0.24) look as steady as usual, while ABB (International Bank of Azerbaijan), at 0.83, remains one of the most stable participants in the series: over twenty months its readings have stayed within a narrow band around 0.85.

Why we no longer use the Central Bank's colours

The regulator's table divides banks into three categories: high, medium and low complaint index. The explanatory note to the table defines only two of them. A high index is a value above the average obtained as a result of the calculation, and a low index is one below that average. The note does not say anywhere what a medium index is.

The first objection is purely logical and requires no calculation at all. A single criterion divides any set into exactly two parts: what lies above it and what lies below it. Three categories require two boundaries, yet only one has been declared. If we assume the medium category is a band around the average, a second parameter is needed, namely the width of that band, and it has not been published. In other words, the third category cannot be derived from the stated methodology at all.

The second objection is arithmetical. Even if we accept that there are in effect two categories, it remains unclear which average is meant. Two readings are possible. The first is the structural sector average, which equals 1.00 and follows from the regulator's own formula; with that boundary, eight banks would have been above average in August. The second is the arithmetic mean of the published column, 1.08 in August; with that boundary, there would have been six. In practice, only four rows fall outside the green zone. Neither reading matches the picture.

The third objection is empirical, and it is the most awkward one. The boundaries of the colour zones do not move with the market. Across the twenty published tables, the lower boundary of the yellow zone consistently falls between 1.18 and 1.21: a value of 1.18 was coloured green in both July and August 2026, while a value of 1.21 was placed in the yellow zone as far back as July 2025. The boundary of the red zone lies just as consistently between 2.34 and 2.51. Over the same period, the column average ranged from 1.06 to 1.66. What we have is a fixed absolute scale, published nowhere and inconsistent with the wording of the legend.

In fairness, the bias created by this fixed scale does not run in one direction only. In February 2026, with an average of 1.66, the regulator placed ten of seventeen banks outside the green zone, whereas the “above average” criterion would have produced seven. In hot months, the scale is stricter than its own legend. But 2026 has been a year of steady cooling, and in that phase the effect works the other way. In August half the table exceeds the neutral level, yet only four rows out of sixteen look problematic at a glance.

None of this means the Central Bank's data are bad. It means only that the colour in its table is not a function of the average, even though it is presented as one, and so it does not tell the reader where a bank stands relative to the market. We take the only criterion the regulator itself names, apply it transparently and arrive at exactly two categories: above the neutral level and below it. Gradation within the upper group comes not from colour but from the size of the excess and the trend. Yelo Bank is two and a half times above the norm and has stayed above it for seven months running; Kapital Bank is two percent above it, for the first time this year. No single shade can convey the difference between these two cases.

What comes next

The September table will show whether Yelo Bank's August peak was a turning point or the continuation of a year-and-a-half trend, and whether Rabitabank can hold on to its unexpectedly clean result. We will keep measuring against the 1.00 mark, regardless of how the table is coloured.

Central Bank complaint index, August 2026

Source: Central Bank of the Republic of Azerbaijan, “Bank müştərilərinin bank xidmətlərindən istifadəsi üzrə şikayət indeksi” reports for January 2025 to August 2026. Averages, medians, time series and colour-zone boundaries calculated by First News Intelligence Unit. Premium Bank is not included in the August table: according to the regulator's note, no complaints were received during the month about banks that do not appear in the table.

About the author:

First News Intelligence Unit (FNIU) is the analytical unit of the 1news.az newsroom, specialising in research on Azerbaijan's financial sector and economy.

Read in other languages:

«Не-фантастическая четвёрка»: Yelo Bank, Turanbank, Yapı Kredi и Unibank снова во главе индекса жалоб Центробанка Азербайджана

Heç də fantastik olmayan dördlük: “Yelo Bank”, “Turanbank”, “Yapı Kredi” və “Unibank” yenidən Azərbaycan Mərkəzi Bankının şikayət indeksinin zirvəsində