[HN Gopher] Self-Serve Dashboards
       ___________________________________________________________________
        
       Self-Serve Dashboards
        
       Author : vieiralucas
       Score  : 172 points
       Date   : 2024-06-11 14:04 UTC (8 hours ago)
        
 (HTM) web link (briefer.cloud)
 (TXT) w3m dump (briefer.cloud)
        
       | engine_y wrote:
       | I wonder what people here think. Would self serve BI work for
       | something which is more context specific? So for example self
       | serve BI for your own sales. What would it take? Why not?
        
         | wswope wrote:
         | Hot take as a datamonkey who's done a lot of work in this vein
         | professionally:
         | 
         | General-purpose self-serve is hostile to non-technical end
         | users. Most do not have the mental model of SQL to guide their
         | usage of the tool, so giving them an "open-ended" option to run
         | their own vizes ends up really being a fence-toss of some
         | technobabble garbage that isn't useful to them at all.
         | 
         | "Slice n' Dice"-style filtering added to an existing set of
         | reports, however, is a completely viable middle ground.
         | Practically speaking, this means writing the basic dashboard
         | query, then hooking a bunch of query parameters up to some
         | front-end UI widgets to let end users pick the date range,
         | level-of-detail, filter which trendlines or categories of data
         | are shown, etc. Just use good taste to avoid going overboard -
         | don't try to make every last detail configurable.
        
         | rbetts wrote:
         | > If we assume that the problem with self-serve BI is not SQL,
         | but the context and semantics of the data, then it follows that
         | the solution is to teach people about the data they're
         | querying, regardless of interface.
         | 
         | This has been the basic truth of any self-serve BI system I've
         | used.
         | 
         | Even in smallish orgs there are often three steps - the
         | engineer who instruments code/implements a metric, the engineer
         | who builds the ETL pipeline into the underlying BI warehouse,
         | and the person querying that data. So there are minimally three
         | people in potentially three different roles who need a shared
         | specification and understanding.
         | 
         | Also, self-serve BI tools can be surprisingly opaque and their
         | output can be hard to validate/test. So even if you know
         | accurately what data you are querying, testing that your query
         | is what you intend is hard.
        
         | mead5432 wrote:
         | This is something I've been wrestling with for a little bit.
         | The challenge is that if everything is locked behind a data
         | analyst then things only move as fast as the data analyst. If a
         | business person needs some information now, not next quarter,
         | the challenge is managing the queue for the data analyst which
         | devolves into a world of "rush jobs". It's all the things
         | highlighted by the theory of constraints (The Goal by Goldratt
         | is sooo good). You can alleviate the DA bottleneck by getting
         | more DAs or you can find ways to help the business users do
         | some of the basic stuff themselves.
         | 
         | I wholeheartedly agree that just giving a business user access
         | to the straight database is not ideal for all the issues
         | mentioned - they don't know the context and the gotchas in the
         | data combined with probably not understanding how to write SQL.
         | I think an effective data warehouse strategy with
         | straightforward data marts of materialized views can simplify
         | the interaction and maybe even make it really simple for
         | someone to generate basic visualizations. A lot of business
         | people can make basic dashboards in Excel which, worst case,
         | could be connected to a data mart. It's not going to handle BIG
         | data but may cover a large number of use cases for most
         | businesses.
         | 
         | I'm in favor of creating some basic dashboards We've also been
         | experimenting with embedding dashboards in internal tools that
         | provide some slice and dice capabilities but with a high level
         | filter. A user can manipulate and tweak a dashboard for a
         | specific customer but to look at a different customer they need
         | to navigate to it in the internal tools, not via the dashboard.
        
       | pcloadletter_ wrote:
       | This reminds me of a time I was at a company using a BI tool for
       | dashboarding. The numbers weren't making much sense to me, so I
       | looked at the query building tool. I couldn't tell for the life
       | of me if a part of the query was doing an inner join or left
       | join. The business analyst who built the dashboard had no idea
       | either.
       | 
       | It turned out that it was doing a left join when the intent was
       | an inner join, and the data being shown was an order of magnitude
       | higher than it should have been. This is when I lost all faith in
       | these kinds of abstraction layers on top of SQL targeting people
       | who don't actually know SQL.
        
         | appplication wrote:
         | I have seen it play out over and over with low code tools. The
         | difference between a sensible default and a footgun is just a
         | matter of perspective.
        
         | garciasn wrote:
         | As someone who leads a Data Engineering and Data Science team
         | and has for 15+ years, this is exactly the problem. Too many
         | folks with access to data who do not understand the data, its
         | relationships, and what the outputs their individual efforts
         | create mean.
         | 
         | Decentralized/embedded Engineers/Scientists; self-service
         | dashboards; low-code BI/data tooling; and, now, LLM-driven text
         | to SQL/viz lipstick on a pig have been floated as some of the
         | solves to the problems seen in the analytics space over the 25
         | years I've worked in the space. Unfortunately, to date, nothing
         | has actually solved the root issue: lack of data understanding
         | and, its end result, trust in the deliverables.
         | 
         | But, to your specific point, SQL isn't the solve here, either.
         | Too many folks know enough SQL to pull data and use it as they
         | see fit, but too few folks understand the data, its
         | structure/schemas, and valid use of those data. THAT requires
         | time, energy, knowledge, and experience in the space. NO
         | TOOLING, other than experience, solves for this--today (note:
         | will LLMs get to a place where they can? Maybe; but, let's be
         | honest--probably not).
         | 
         | Dashboards are great at giving quick hit information of KPIs
         | and the ability to drill down into them; but, the most
         | important thing to solve are always:
         | 
         | 1. Data Management practices
         | 
         | 2. Understanding of data, its relationships, and proper use of
         | those data/metrics in deriving insight to drive the business
         | forward.
         | 
         | I am excited to see what the future holds, but my grey beard
         | doesn't allow me to ever, Ever, EVER trust any next-gen tooling
         | being it hasn't held true to date.
        
           | cjk2 wrote:
           | Ah it's much higher than that.
           | 
           | I always ask: tell me the question you are answering with
           | this.
           | 
           | 99% of people can't answer that question.
        
       | pelagicAustral wrote:
       | Haha Yes, this is it. I maintain a number of databases at work
       | and I've implemented Metabase about a year ago. Started the first
       | dashboards, translated some of the old reports, and hoped that my
       | dear users would engage with the platform for their own needs...
       | it never happened, and it's probably never going to happen.
       | 
       | I agree, a user with no grasp of the schema, or structure of the
       | database, or with zero knowledge on how to wire a query (with or
       | without a graphic interface) will not make the effort because
       | even when they do, they feel like a caveman in front of an
       | iPhone, and everybody hates that feeling.
       | 
       | Most of the time, even people that have worked with PowerBI, or
       | other tools wont make the effort because they also know, push
       | comes to shove, I will end up querying for them, so why would
       | they?
       | 
       | This is a problem with no easy solution I'm afraid... me, I love
       | writing SQL, so it's a bit of a break each time I need to produce
       | stats of some sort, or get a report out.
        
         | sumoboy wrote:
         | Huge gap for building self service tools/dashboards for execs,
         | just no sense of the schema and data to make them sql writers
         | which you don't want. When metabase showed a version last year
         | with AI, you could see right away without some knowledge of the
         | data model your just grasping at writing the correct prompt.
        
       | drinkzima wrote:
       | The straw man is always that self-serve fails because every user
       | cannot use data well at work. The reality is some users will be
       | inclined to solve their own problems and others will not, but
       | self service is available to many users with deployed BI, and SQL
       | is nearly always not the way they are doing it.
       | 
       | Most times I see this type of article, it's with folks that have
       | never worked in a modeled BI tool. Salesforce data, for example,
       | is very complex. But an ability to make a table of live
       | opportunities with metadata and order them freely, next to usage
       | data in an app is self service BI. It's not hard; it takes some
       | setup; but it's self service.
       | 
       | The idea that folks can jump from business understanding to fully
       | mapping the data as it lives in the data warehouse, on the other
       | hand, is not trivial and won't be. The nuance of the real world
       | is hard.
       | 
       | Different types of users need different interfaces - SQL all the
       | way down to point and click. And there's no free lunch on
       | modeling raw data to bring it to a consumable place for the
       | company.
        
         | snowfield wrote:
         | Yes, provide different ways for different users to sove their
         | own issues. Then technical people can solve the hard ones.
        
       | camjw wrote:
       | This feels like one of those spaces where everyone says LLMs and
       | AI are just going to make it a solved problem, but no-one in the
       | comments here is even mentioning it - maybe not a good problem
       | for LLMs after all?
        
         | mritchie712 wrote:
         | Most people that went after this tried for text-to-sql (e.g.
         | ask a question and generate a ton of SQL to answer it). That
         | approach has pretty much failed. The LLM could never have
         | enough context to generate accurate SQL at a high enough rate
         | to trust.
         | 
         | What we've found to actually work at Definite (I'm the founder)
         | is text-to-semantic-query. This is an older video, but here's
         | an example: https://www.youtube.com/watch?v=44mhLgUYOp8
        
           | camjw wrote:
           | How is this any different (text-to-sql vs text-to-semantic-
           | query). Isn't this just comparing text-to-sql to text-to-
           | slightly-simpler-sql?
        
             | mritchie712 wrote:
             | Yes, it's simpler, but there's a few key differences:
             | 
             | 1. You also have complete control over what the LLM can do
             | / access thru the semantic layer (e.g. you can remove
             | tables that the LLM shouldn't consider for analytical
             | questions).
             | 
             | 2. One of the biggest choke points for text-to-sql is
             | constructing joins. All the joins are already built into
             | the semantic layer.
             | 
             | 3. Calculating metrics / measures is handled in the
             | semantic layer instead of on the fly with SQL (e.g. if you
             | ask something like "how much revenue did we generate from
             | product X", you wouldn't want the LLM to come up with a
             | calculation for revenue on the fly. Instead, revenue is
             | clearly defined in the semantic layer).
             | 
             | 4. The query format for our semantic layer (we use
             | cube.dev) is JSON, which is much easier to control then
             | free form SQL.
             | 
             | The semantic layer gives the LLM a well defined and
             | constrained space to operate within whereas there are
             | hundreds of ways for it to fail writing raw SQL.
        
         | TheGRS wrote:
         | The post talked about that too: LLMs could come up with a query
         | for any question. A technical person would know how to spot the
         | problems with that query or even the question, a non-technical
         | person wouldn't know that they just asked for a bunch of hot
         | nonsense.
        
         | pcloadletter_ wrote:
         | With all the hallucinating (or "bullshitting") going on, it's
         | hard to imagine LLMs working well for query generation. But
         | hey, we're _very_ early days for all of this.
        
         | tlarkworthy wrote:
         | I have watched this indie hacker do quite well over the last
         | year or so, with ai2sql.io
         | 
         | https://x.com/mustafaergisi/status/1793657418435432649
        
         | PheonixPharts wrote:
         | > but no-one in the comments here is even mentioning it
         | 
         | The post has an _entire section_ discussing this.
         | 
         | The problem with text-to-sql is that, as the post elaborates,
         | _writing SQL_ is not the problem. It 's understanding the
         | context and the data:
         | 
         | > On the other hand, a technical person would notice that the
         | question doesn't make sense, and they would ask for more
         | context. They would ask for details about the business person's
         | hypothesis and the problem at hand. Then, they would explain
         | what type of data is available, and work with the business
         | person to formulate a precise and useful question.
         | 
         | Text-to-sql in practice is a solution that nobody was asking
         | for, despite the insane number of SV startups shipping GPT-
         | text-to-sql wrappers as products.
         | 
         | There certainly is like places where LLMs _can_ help (post
         | touches on this briefly), and that is in semantically exploring
         | databases /tables/etc and contexts around data, but this is a
         | very different project and would require a lot of curation from
         | data teams to make it happen.
        
       | mritchie712 wrote:
       | You're right that text-to-sql doesn't work, but text-to-semantic-
       | query works very well. This is an older video, but here's an
       | example from Definite (I'm the founder)
       | 
       | https://www.youtube.com/watch?v=44mhLgUYOp8
        
       | ssahoo wrote:
       | Modern devs are not familiar with OLAP and cubes, therefore the
       | argument is valid until time swings again.
        
         | qwertyuiop_ wrote:
         | Just a nitpick. The term "modern" is thrown around as
         | equivalent to evolutionary progression to best practices. The
         | term that could used to refer to the developers in you are
         | describing could be "contemporary".
        
       | NightMKoder wrote:
       | I've come to a similar conclusion about "self-service BI" myself
       | but my solution is somewhat different. The solution I have is
       | move the layer of abstraction higher: make extremely customizable
       | dashboards but do not expose SQL to business users.
       | 
       | An example of this might be a dashboard with 20 filters and 20
       | parameters controlling breakdown dimensions and "assumptions
       | used." So asking "how did Google ads perform in the last month
       | broken down by age group" is about changing 3-4 preset dropdowns.
       | Parameters are also key here - this way you only expose the knobs
       | that you've vetted - not arbitrary SQL.
       | 
       | Obviously this is a hard dashboard to build and requires quite a
       | bit of viz expertise (eg experience with looker or tableau or
       | excel) but the result is 70% of questions do become self service.
       | The other 30% - abandon hope. You will need someone to translate
       | business questions into data questions and that's a human
       | problem.
        
         | snowfield wrote:
         | Yes, know which questions gets asked often and make those
         | dashboards. Now the cfo or whoever can just open those
         | dashboards when they want answers to those questions for any
         | given timeframe.
         | 
         | And tinker with basic parameters
        
       | jasmcole wrote:
       | If you're interested in a mix of SQL, Python, DuckDB and low-code
       | interfaces to SQL databases, you may want to check out
       | https://count.co
        
         | Dyac wrote:
         | MASSIVE count fan here. Been using it over a year and nothing
         | else comes close for exploratory analysis.
         | 
         | I've not felt as impressed by a tool or had it so quickly
         | change how I work ever before.
        
       | datadrivenangel wrote:
       | The conventional BI interface shown in the post is Metabase,
       | which has one of the better interfaces for doing BI out there
       | today. Metabase also has the ability to see the GUI-generated SQL
       | for a question, and convert that question into pure SQL, which is
       | a great way to move from self serve to governance as it's easy to
       | modify and validate the logic and give less technically skilled
       | people an easy route towards improving their skills.
       | 
       | The post is fundamentally correct though, and I say this as a
       | professional data person, a BI tool rarely gives more people
       | correct understanding of the data or the technical skills needed
       | to use it correctly. If your data is well managed, the tools are
       | easy and people can figure things out, but the world is
       | complicated, so your data will become complicated, and the cost
       | of data management is very visible while the benefits are
       | invisible.
        
       | NoboruWataya wrote:
       | I think BI dashboards can work but only for very simple queries.
       | If you are even at the point where you are asking non-technical
       | people to execute joins on data I think you are in too deep and
       | should just be writing SQL at that stage. Joins might seem very
       | basic to some but honestly I find it difficult to wrap my head
       | around them sometimes and with a UI like a dashboard which is
       | less expressive than SQL I think it would just be a recipe for
       | confusion.
       | 
       | It's ultimately a trade-off: you can make your tool more
       | accessible to non-technical users than SQL, but it will
       | necessarily be less powerful than SQL. And I still think there
       | are plenty of use cases within that space. IME so much "BI" is
       | just "I have two columns of data and I want to plot one against
       | the other".
       | 
       | The author describes SQL as the only "self-serve" BI tool, but
       | honestly, I think that is Excel. So many of these BI tools are
       | just reinventing Excel with new (and therefore less familiar)
       | interfaces. It is a meme to hate on Excel and I think that is
       | because people have in the past tried to use it for complex stuff
       | that really should be done in SQL. If we used SQL for complex
       | data manipulation and Excel for "give me that as a pie chart",
       | there would truly be no need for BI tools.
        
       | smithcoin wrote:
       | We use Metabase (featured in the image) and for the most part
       | non-tech people do it fact use it. What helped us implement it
       | was having 'office hours' where I walked people through examples
       | of how to use it "This is how to get sales at a specific location
       | or in a state". While it hasn't solved every
       | problem/query/export, a large portion of requests that went in
       | the direction of engineering before no longer make it that far.
       | Other reasons to love metabase, you can self host it and use your
       | GSuite SSO.
        
         | Galanwe wrote:
         | From my experience, the issue is not the means to query the
         | data, it's the actual idiosyncrasies of the data. The key
         | metric is thus not "users seem to be more independent as we are
         | receiving less inquiries for help", because there is a very
         | high chance said users are pulling and interpreting completely
         | bogus metrics from the data.
         | 
         | I have seen this over and over. Once low tech users have access
         | to the data, they start building pyramids of bogus analysis,
         | somehow convinced that after all it's not that hard.
         | 
         | The real blocker is all the context that is - let's be honest -
         | always required to perform a correct analysis:
         | 
         | - "Oh no, you cannot use the delivery date to compute monthly
         | sales, since the finance team refill it for recurrent sales"
         | 
         | - "Oh no the prices are stored in USD in the catalog but we
         | actually adjust the rate monthly based on the
         | `monthly_discount` table"
         | 
         | - "Yeah you have to remove items that have a null purchase date
         | from the sales report we have that convention to mark last
         | year's unsold stock"
         | 
         | - "No you cannot sum the sales without joining with the FX rate
         | table since prices are in local currency"
         | 
         | - Etc, etc, etc
        
       | serverlord wrote:
       | My product has a GPT-4 API to convert text to SQL and it works
       | just fine. Reason: Good data engineering and using ClickHouse to
       | have a single materialized view to create any dashboard. It will
       | be out soon. It works like magic.
        
       | mmcgaha wrote:
       | I have been providing data to business users for the last 24
       | years. It doesn't matter if we give them a query tool, MS Access,
       | Power BI or a data cube in Excel, there are only a small number
       | of users who will actually use the tools. My guess is these are
       | the same users who would have done analysis with data scraped
       | from terminals and printed reports 40 years ago. For what it is
       | worth, execs do like a dashboard with key metrics and the new BI
       | tools make it a lot easier to write and maintain KPI dashboards.
        
         | whartung wrote:
         | > My guess is these are the same users who would have done
         | analysis with data scraped from terminals and printed reports
         | 40 years ago.
         | 
         | Back in the early computing days, most of the work was done on
         | the large, shared, mainframe computers. The computers were so
         | costly, that the "computer division" was it's own, separate
         | division of the corporation that had premises with the other
         | divisions. For example the Western Division made products, but
         | if it wanted computer resources, it contracted with the
         | Computer Division, who handily had mainframes installed onsite.
         | 
         | Our group was a feisty, small internal analysis group using the
         | new, "cheap" mini computers. One of our points of service was
         | simply being much more reactive to the users needs, we could
         | simply respond more easily because of how the funding worked.
         | 
         | To you point about "data scraped", we were walking through the
         | plant and saw one of our users with one of our reports. They
         | were cutting the lines out of the green bar report, taping them
         | to another piece of paper, and photocopying it. They were
         | sorting the report by a different criteria. We told them "You
         | know, we can do that for you!" "Oh really!?"
         | 
         | People that need to Get Stuff Done, get it done. Our goals as
         | service providers (which is what we in the computer systems
         | groups are, service providers to our internal customers), is to
         | make that as efficient as possible.
         | 
         | Another person was using a PC, our tablet digitizer, and
         | Autocad to record the points on aircraft to generate radar
         | profiles. It was an inventive use of the digitizer, not for
         | fundamental CAD work, but simple data capture from the drawings
         | in Jane's Combat Aircraft.
        
       | pphysch wrote:
       | > That story usually starts with an engineer or data scientist
       | who's frustrated because they spend too much time writing queries
       | and preparing dashboards for business people. They think that if
       | they make BI easy enough, everyone will be able to "self-serve",
       | but that rarely ever happens.
       | 
       | The solution is usually to paste/implement that query in a low-
       | tech automated email report, web page, or Grafana/shared
       | spreadsheet.
       | 
       | It's a (simple) automation problem, which should not be conflated
       | with cross-training an MBA to be a data scientist.
        
       | nosefurhairdo wrote:
       | Our self-serve dashboards feature is the most used feature on our
       | platform according to our analytics. Our customers include many
       | Fortune 100 companies and our company was acquired for some
       | billions of dollars.
       | 
       | I largely agree with the sentiment but it's just not what we've
       | seen in practice.
        
       | DarkerIsBetter wrote:
       | Excellent post.
       | 
       | Unless the data model is either extremely clean or simple, users
       | lacking deep context will struggle.
       | 
       | There are always tables for abstract concepts, code paths for
       | legacy behavior, etc. Should we expect users to embed this
       | internal business logic into their SQL queries? How do these
       | users know when they need to change their embedded assumptions?
        
       | esafak wrote:
       | I think the missing product the author alludes to is the data
       | _catalog_ , to document what the data means, and where it comes
       | from.
        
         | zurfer wrote:
         | As someone who has build a data catalog tool and is now
         | building one of these looked down on LLM agents(1): Data
         | catalogs are mostly used by technical people/the data team and
         | a handful of power users. Why? Because it's intellectually
         | really difficult to understand data deeply and it's similarly
         | hard to present data to make it easy, so people just ask the
         | people who know (no self service).
         | 
         | An LLM agent can shine here because a) you can give all the
         | relevant context that is needed to make sense of the data and
         | b) it can be personalized and proactive, so the user just gets
         | appropriate reports when they need it, instead of wading
         | through a "Customer360 Cockpit" with hundreds of visuals and
         | options.
         | 
         | (1) https://www.getdot.ai/
        
       | jppope wrote:
       | It always cracks me up that you can pay an upper level manager
       | >$XXX,XXX and they can't run BI sql queries... SQL was invented
       | for the purpose of making it easier for managers to query data.
       | lol
       | 
       | As a former sales guy/ manager I also have no sympathy for those
       | people...
        
       | timacles wrote:
       | As someone who has been a part of implementing BI in every
       | possible form across several companies in the last 10 years, its
       | always the same problem.
       | 
       | No one uses it for jack shit.
       | 
       | All the colorful graphs, charts and cool visualizations have very
       | little actionable information for 99% of people. The other 1% is
       | management and executives who need some random 3 KPI points
       | charted over 3 months to look at so they can feel like they're
       | doing their job or have something to complain about to their
       | underlings.
       | 
       | I have never seen any substantial discussion about any BI metric.
       | Its always a passing thought of curiosity, get the data, chart
       | it, go "Oh Wow, would you look at that." and then immediately
       | move on.
       | 
       | And to take a step back, the only time people actually use any
       | sort of metric consistently, is when they have an obsessive
       | curiosity about a particular area.
        
         | camjw wrote:
         | Why is it that $BILLIONS is spent on this problem then, if it
         | doesn't exist. I get that, yeah, management and execs only care
         | about 3 KPIs but presumably people lower than them in the chain
         | care about improving those KPIs and digging into the data helps
         | you do that?
         | 
         | Or is this not true?
        
         | gofreddygo wrote:
         | > No one uses it for jack shit.
         | 
         | Dashboards are lossy, compressed selective subset of hyper
         | processed data. They don't show the why, or how, or when.
         | 
         | yet they survive in the form of a report or a chart because
         | they shift the control of the narrative on the designer of the
         | dashboard. You pick and choose what you want to emphasize and
         | what to hide. Convenient, if thats what you want.
         | 
         | OTOH, i've never found or resolved or identified a single issue
         | from a dashboard. I've spent hours on why something wasn't
         | highlighted or something was and found the culprit to be the
         | dashboard itself.
         | 
         | Yep, jackshit
        
       | hamasho wrote:
       | I think sometimes encouraging business people to make SQL queries
       | themselves can be beneficial if the database schema is well-
       | defined.
       | 
       | When I worked for an ad-tech company, we built an ETL system to
       | convert our messy, complex, and technical debt-ridden application
       | database into a well-organized analytical database. The main
       | purpose was to make the jobs of data scientists and machine
       | learning engineers easier.
       | 
       | However, it also helped business people with some technical
       | knowledge create the dashboards they wanted. Although some
       | queries were a bit messy, it made it easier for them to organize
       | their requirements and communicate effectively with the tech
       | team. Unfortunately, it also resulted in a lot of half-baked and
       | unused dashboards, but overall, it brought positive change to the
       | company.
       | 
       | That said, I don't think it's worth developing an ETL system just
       | for business people. It requires multiple dedicated devs, whereas
       | writing SQL for dashboards occasionally only takes a few days per
       | month for a single dev. I agree that the most important part is
       | fostering a good relationship between tech and business people.
       | If business people have a mental barrier, it becomes challenging
       | to create new dashboards and update or fix existing ones.
        
       | mehulashah wrote:
       | If this is true for BI (which I tend to agree with), then this
       | even more true for ETL. End users have an even less of a chance
       | of doing data wrangling. Rather, it's the technical folks that'll
       | do it.
       | 
       | What we learned in building AWS Glue was that it wasn't just
       | about the context -- it was also about escape valves. Escape
       | valves are tools necessary to get out of a situation that wasn't
       | anticipated. When the answer doesn't make sense, the technical
       | users are the only ones that have the know how to debug it.
        
       | kthejoker2 wrote:
       | The assumption that business users can't be taught or are too
       | impatient (even a hint of too stupid?) to learn the relationship
       | between their questions, the data models, and the drop downs is
       | ridiculous.
       | 
       | I could flip this problem around ..
       | 
       | In my experience they are always willing to learn but often times
       | the data modelers don't understand their domain well enough to
       | capture all the nuances of the questions they ask.
       | 
       | Instead they just hide nuance (and increase the time it takes to
       | answer a question) or eliminate it (and therefore produce
       | inaccurate and misleading answers) all in the name of dumbing
       | down self-serve.
       | 
       | I hate the euphemism "non-technical" you can absolutely find a
       | middle ground between LLMs and BI query generation tools and SQL,
       | instead of just declaring by fiat some impenetrable wall of
       | competence.
        
       | kkfx wrote:
       | The main issue is that modern tools are not like classic
       | desktops, like Smalltalk workstations or Emacs, meaning a single,
       | fully-integrated environment, where anything is at user disposal,
       | with end-user programming concepts built-in.
       | 
       | In org-mode I can create "good looking slides" in a snap, I can
       | quickly craft some chuck of code, run it and get some results,
       | it's damn limited in "dashboard" terms, let's say I can quickly
       | plot some data but the plot is just a crude static image, make it
       | glow with PGF/TikZ it's _very_ time consuming so it 's not an
       | option either and it would be still static, because Emacs itself
       | it's the right tool but from an older era. Modern tools offer
       | more eye candy and quick manipulation but only for very limited
       | actions in a very inflexible UI obviously not integrated with
       | anything else. R probably is the quickest with R Studio/quarto to
       | produce contents quick, dirty and still nice to see, but it's
       | still far from the flexibility of Emacs. I think there is no
       | solution without re-writing the entire modern software stack with
       | the classic paradigm and modern stuff doable thanks of much more
       | horsepower under the wood.
        
       | Angostura wrote:
       | The author spends a lot of time convincing himself that SQL is
       | not the problem, but that it's the context and semantics of the
       | data.
       | 
       | I disagree. Just because understanding the data is a difficult
       | problem doesn't mean that SQL isn't _also_ a problem.
       | 
       | I understand the data and the set up of the database just fine.
       | Darned if I can remember SQL syntax
        
       | codexb wrote:
       | I've also found this to be the case.
       | 
       | That being said, I like being able to create dashboards for
       | myself, but the interface for every tool to make these queries
       | tries to be too clever and it ends up being painful to use --
       | _cough_ JIRA _cough_
        
       | dapearce wrote:
       | I've implemented self-serve analytics at three organizations
       | successfully. It's important your data is well organized, well
       | labeled/defined, and the BI tool needs to allow you to configure
       | guardrails around how data is queried. I have used Looker as the
       | BI tool at all three organizations. End users do not need to even
       | define joins. They just select the grouping and aggregate columns
       | they want to return and click run, then configure their
       | visualization from the data. It is true that some users are not
       | data literate and still will not run their own queries, but from
       | my experience a lot of non-technical business users love being
       | able to easily explore data themselves, and the business receives
       | a lot of value from it.
        
       | lovasoa wrote:
       | I do agree that data teams are probably not going to disappear in
       | the short term, and their understanding of the structure of the
       | data they work with is even more important than their SQL syntax
       | skills.
       | 
       | But that said, I think we can do better than just static
       | dashboards. What I'm trying to do with https://sql.ophir.dev is
       | to let the same data teams write apps instead of dashboards. Apps
       | are more flexible, and allow deep dives and navigation to a level
       | that is not possible with just a dashboard.
        
       ___________________________________________________________________
       (page generated 2024-06-11 23:00 UTC)